Figure 1. Location of possible debris from MH370 (red) detected during the surface search (green) on March 29, 2014. Possible flight paths aligned with waypoints also shown (black). (Click on image to enlarge.)
Introduction
Australia’s CSIRO recommends that the next subsea search for MH370 be carried out in a 25,000 sq km area that extends along the 7th arc as far north as 32.6S latitude. The recommendation is based on extensive drift analyses as well as satellite imagery which may have captured debris from the impact.
Despite the work of CSIRO, there are reasons to believe that the wreckage from MH370 is along the 7th arc, but at latitudes to the north of 32.6S latitude. Some of these reasons are:
- The absence of debris discovered along the shores of Western Australia is better explained by an impact to the north of 32.6S latitude.
- The timing and location of the fragment of the engine cowling with the letters “Roy” that was discovered in Mossel Bay, South Africa, in December 2015, is better explained for an impact north of 32.6S latitude.
- The particular impact points that CSIRO considers to be most likely are located at latitudes along the 7th arc that have already been searched to about 19 NM from the arc. With the increasing descent rates indicated by the final two BFO values, the impact likely occurred within that distance from the arc.
On the other hand, CSIRO has provided reasons why they believe the impact could NOT be north of 32.6S latitude. These are:
- For latitudes north of 32.6S, debris should have arrived in Madagascar, Mozambique, and other locations in Eastern Africa, before the discovery of debris was reported. But this could easily be explained by the delay between when debris arrived near a location, and when it is discovered and reported to authorities. In fact, the circumstances surrounding Blaine Gibson’s discovery of the horizontal stabilizer fragment with the letters “No Step” suggests the arrival of that part in Mozambique well before it was discovered.
- Reconstructed paths terminating along the 7th arc to the north of 32.6S latitude are less consistent with the satellite data. Yet, if we allow for pilot input between the last radar capture at 18:22 and the second handshake at 19:41, and if we assume automated flight after 19:41, there are reconstructed paths that match the satellite signaling data within allowable error limits. In fact, there is a range of possible paths that terminate to the north of 32.6S, and these were the focus of my last post. Within this range, there are some waypoint-derived paths that might deserve extra attention.
- If an impact had occurred north of 32.6S, the floating debris would have been detected by the surface search that was conducted by aerial surveillance in the weeks following the crash.
In this article, we take a closer look at the last item in the list.
Many believe that CSIRO’s estimate of search efficiency is over-stated, and it is possible that debris was missed during the surface search. Here, we consider the possibility that floating objects from MH370 were in fact detected by the aerial search and photographed, but those images were ignored after path reconstruction models placed the probable impact site further north, and then further south along the 7th arc. The images captured by the surface search demand a new level of attention in light of the prospect for re-starting the subsea search along the 7th arc in areas that will likely include latitudes to the north of 32.6S.
Obtaining the Images
Large areas of the ocean were surveilled by the Royal New Zealand Air Force (RNZAF) using P-3 Orion aircraft as part of the overall surface search effort. One image from March 28, 2014, which included an object described as a “blue panel”, received some attention because of the object’s resemblance to a flaperon. This object was discussed in an article by Bernard Lagan, who flew with the New Zealand Orion crew on March 28. (The video that accompanies the article is worth watching.)
Curiosity surrounding this debris prompted IG member Brian Anderson in October 2015 to request more information about this and other images collected by the RNZAF search that could be debris from MH370. His request was fulfilled, and he received a large set of images that were made publicly available on Duncan Steel’s website. For convenience, here is a compressed file (1 GB) of the all the images.
Brian continued to query the RNZAF, and was able to obtain more information about the coordinates of the images. IG member Don Thompson was also able to extract the position and timing from the meta data embedded in some of the images. IG member Richard Godfrey in turn helped assemble the data for the completed set of images and also provided some descriptions of the images. The time and position data that we have for all the images is compiled in this Excel file.
As far as we know, none of the objects identified in the surface search on March 29, 2014, were recovered by ship, so the relationship to MH370 remains unknown.
What the Images Show
The images taken on March 29, 2014, captured a variety of objects, some of which could be floating debris from the impact of MH370. The surface search also found objects that are without a doubt not from MH370, including fishing articles such as nets, floats, and marker buoys. Some of the more interesting images from March 29 are shown in Figures 2 – 5, including two debris fields of small fragments, an object that resembles a suitcase, one that resembles a portion of a panel with wires, and a rectangular box that resembled a cargo package.
Figure 2. Enlarged images of debris fields on March 29, 2014. Left: 28.3927S, 97.7750E from image 5832. Right: 29.19333S, 95.1250E from image 5853.
Figure 3. Object resembling a suitcase near 28.3176, 97.8234E on March 29, 2014. (Enlarged from image 5833.)
Figure 4. Object resembling a portion of a panel with wires near 28.8867S, 96.0844E on March 29, 2014. (Enlarged from image 5847.)
Figure 5. Object resembling a rectangular cargo package near 29.1936S, 95.1094E on March 29, 2014. (Enlarged from image 5854.)
Drift Analysis
As only 21 days elapsed between the impact of MH370 and the discovery of the debris, it should be possible to “backtrack” the objects to March 8 to determine a potential point of impact. To simplify this procedure, I looked at the forward drift results for a point of impact on the 7th arc at 29.7S latitude, which corresponds to the path aligned with Wilkins Runway (YWKS), Antarctica, and is also the closest path to the objects among the four waypoint-derived great circle paths that I examined in the last article.
Using the CSIRO results for drift of low windage debris from 30S latitude, I selected 31 particles that were positioned within a 15-NM radius of the assumed impact point at 29.7S latitude, among the hundreds of particles that CSIRO studied over a much larger area. The selection of multiple particles within a circular area allows for a diversity of drift paths which reflect the uncertainty in the impact point and the stochastic nature of ocean drift. Ideally, more than 31 particles would be modeled as we know the crash produced many more, but the geographic spread of the injected particles in the available CSIRO data sets limited the number of particles I could select and get meaningful results about a particular impact point. The selected particles (green) among the total particles available (translucent blue) are shown in Figure 6.
Figure 6. Selected particles (green) within a 15-NM radius (white) of the 7th arc crossing (intersection of black and blue lines) at 29.7S latitude. (Click on image to enlarge.)
The drift of the particles is shown in the following video for the time period between March 8 and March 29, 2014. The direction of the particles is from the impact at 29.7S latitude (white circle) and towards the detected debris (red dots). The video also shows how the paths of the particles diverge so that after 3 weeks, the particles are separated by more than the 150-NM distance between the two detected debris fields (images 5832 and 5853). The drift results indicate that the detected objects are consistent with an impact on March 8, 2014, along the 7th arc at 29.7S latitude. (The video is best seen in “full screen” mode by selecting the four arrows next to the Vimeo logo.)
Implications for New Search
Unless there is evidence that objects detected on March 29, 2014, are not from MH370, we have to consider the possibility that one or more objects are from MH370. This greatly increases the probability that the impact was further north than the 25,000 sq km area recommended by CSIRO for the new seabed search. Ocean Infinity seems committed to searching beyond the first 25,000 sq km, and based on the results presented here, a hot spot around 30S latitude deserves special attention.
I have asked the ATSB and CSIRO for more information about these “contacts”. The ATSB had record of the detection of some of the objects on March 29, 2014, but no record of the detection of the two debris fields. Meanwhile, David Griffin of CSIRO has agreed to do some “backtrack” drift calculations to see where along the 7th arc the debris would have originated if the debris was from MH370. As these results are very relevant to future search efforts, I will make the results available when I can.
Update on April 5, 2018
On November 29, 2017, I received an email from David Griffin in which agreed that the objects spotted by the RNZAF on March 29, 2014, could have been from an impact on the 7th arc around 29.7S latitude. At that time, I had no permission to share his email. I now have that permission, and what follows is the email in its entirety:
Hi Victor,
First, a few responses to your post (which I thought made some good points):
- The 25,000km^2 search area was not recommended by CSIRO. It came out of the First Principles Review. CSIRO was just one voice in the room.
- Roy: “better explained” – well, sort of, see below
- “could NOT be north of 32.6S” – if you check the reports I think you will find words like “less likely” are what we said.
- I won’t comment on what the images may or may not be. I leave that to others. CSIRO’s role is to comment on how floating things move.
- Your use of the kmz files: well done! I think you’ve done much of the job. See below for additional info.
- I think you have made a good point that the items seen by the RNZAF were NOT all conclusively proven NOT to be from MH370.
Next, two asides:
- The surface search targeted regions of ocean where items from the 7th arc may have drifted to. So backtracking any items seen should, by and large, conclude that they were near the 7th arc on 8 March. So anyone (one of your commenters, I think) who thinks that backtracking to the arc supports the idea that the items were MH370-related has missed the point. You are correct that the only point of doing the backtracking is to identify what area on the arc was the potential crash site.
- Forward tracking and backward tracking are mirror images when we do not include any random numbers (which is normally the case). The difference is sensitivity to starting point. Assume for the moment that the NZ photos are all bits of plane. If I backtrack them with our (guaranteed-imperfect) model, I fully expect that many would go wide of the 7th arc. Long story short: its simpler to use forward tracks, as you have done.
Finally, some fresh results:
I have now used both our models (BRAN2015 and BRAN2016, as used in our reports) to investigate the 29.7S crash site scenario. This is a “poor man’s ensemble” (just 2 members) but I think it really helps people not to over-interpret results when confronted with two models. Confucious said: “man with 2 watches does not know the time”. I say “man with just one watch thinks he knows the time exactly when he might not”.
The flow field
Trajectories from a line of points on the 7th arc (~30.5S to 29.5S on 8 March) up til 29 March:
http://www.marine.csiro.au/~griffin/MH370/br15_MH370_93103_tp3l1p2d_bh_arc7_305295_0/20140329.html
http://www.marine.csiro.au/~griffin/MH370/br16_MH370_93103_tp3l1p2d_bh_arc7_305295_0/20140329.html
It is around 19 March (click PREV to go back to then) that the line of points starts to stretch apart. You can see that this is because of the fanning-out of the sea surface height contours. 28.5S 96.5S is a saddle point in the sea level. Images 5847 and 5846 were near this saddle point. Neither model is keen to make trajectories go right there (corollary: backtracks will not come back). Wind and Stokes drift are what make it possible (in both reality and the modelling). Images west or east are more easily reached, from points on the arc that are not far apart, suggesting that it is indeed plausible (taking model imperfections into account) that all the photographed debris items were from a single origin (near 29.7 or 29.8S) on 8 March.
SST images
I have also looked at the satellite sea surface temperature imagery, and overlain this on modelled trajectories with three levels of windage. The debris will be somewhere between the 1.2% and 3% windage dots, while the deep ocean features will move more like the zero windage dots.
The images for 28 Mar
http://www.marine.csiro.au/~griffin/MH370/br15_L3S-1d_MH370_93103_tpb_arc7_305295_0/2014032812.html
and 30 Mar are clearer than 29th, and show (like others before) that there was a strong temperature front at about 29S, very close to the image locations. Currents tend to flow mostly along temperature fronts, not across them. This happens in the model as well as in the real world to a pretty good degree. But fronts are also often slightly convergent at the surface (water sinking) which is why buoyant material accumulates at fronts. So you can see that this provides a plausible explanation of why there were many debris items here – MH370-related or not.
Roy
I would not claim that BRAN2015 trans-Indian trajectories starting at 29.7S are very consistent with the finding of Roy in Dec 2015. It is only a very small fraction of the trajectories that go down there. Enough to say to it is possible – certainly – but not really a very conclusive result. See:
http://www.marine.csiro.au/~griffin/MH370/br15_MH370_IOCC_tp3l1p2dp_rw2_297986/index.html
Other items
March-April 2015 is when debris from a crash at 29.7S would (according to BRAN2015) have started to wash up on Madagascar, Tanzania and Mozambique. http://www.marine.csiro.au/~griffin/MH370/br15_MH370_IOCC_tp3l1p2dp_rw2_297986/20150312.html
Flaperon
Nov-Dec 2014 is the earliest that the flaperon might have arrived at Reunion if the crash was at 29.7S according to BRAN2015. 29 July 2015 is well after this earliest time but still perfectly plausible:
http://www.marine.csiro.au/~griffin/MH370/br15_MH370_IOCC_tp3l1p2dpf10_20_99_297986/20150729.html
So not particularly helpful.
I talked to Craig Longmuir at AMSA today about this, so have copied him in. He may have something to add about the photos and their interpretation.
So, as with the Pleiades images, these RNZAF sightings, if MH370-related, do potentially lead us to a crash site. But Pleiades and RNZAF are mutually-exclusive. Same with what was sometimes referred to as “the Esky lid”.
David
An excellent post as always! Keep up the good work.
@Victor
Thanks for reviewing once again the pertinent information that was derived from the RNZAF P3 Orion searches.
It should also be noted that the original bathymetric surveys undertaken were centred on 29.7°S, and were carried out before the ATSB decided that new interpretation of the SAT data provided terminals much further south.
@Richard: Thank you.
@Barry Carlson: Yes. The interpretation of the 18:40 BFO as an indication that the plane had already turned towards the south moved the search region further south along the 7th arc. Ironically, it was the IG that first alerted others of the significance of the 18:40 BFO. Some of us are now proposing that there was an ongoing descent at 18:40 rather than an already completed turn to the south, moving the search again towards the north. Of course, that recommendation recognizes the null result of the previous search.
Nice work, Victor. Worth pursuing further and I look forward to seeing what CSIRO models say.
@Victor,
Thank you for a very coherent article. A few additional comments:
The RNZAF Orion was only one of many aircraft tasked on 28th and 29th March. RAAF AP-3Cs, Japan CG, G-V, JSDF P-3C, RoKAF P-3C & C-130, PLAAF IL-76 and USN P-8A. The aircraft were all tasked within the bounds of the green shaded area shown in Figure 1 of the post (the area for 29th was a similar size but pivoted by about 20deg a/cw about the NE corner). AMSA has committed to provide me with an ArcGIS shapefile depicting the air search areas. While Bernard Lagan did state the crew were monitoring radar sensors, AMSA had considered that radar would be of limited use considering that floating debris was likely to be of composite construction and floating flat on the ocean surface.
A previous FoI request submitted to the ADF did not produce any images from 28-29th March.
AMSA’s daily updates recorded that HMAS Success and Haixun-01 were active in the search area on 29th & did recover objects from the ocean but nothing recovered was deemed of interest.
My estimate is that only 152,00km² of the planned 256,000km² was searched on 28th, and 232,000km² of the planned 252,000km² was searched on the 29th.
@Don Thompson: If you or anybody else has knowledge that any of the objects photographed by the RNZAF crew on March 29 were recovered, please let us know.
@Victor
I do agree with the previous comments that address your efforts (thanks), but I retain a healthy dose of skepticism relative to the analytics used over the last three plus years. Applying the “Prudent Man Rule” (you can Google that) would an underwater search even have been started? The answer is no. My personal experience with the Prudent Man Rule had to do with establishing mining claims on property that had high potential for future residential development in the Phoenix area. I lost my claims based on the correct conclusion that a prudent man would never attempt to extract anything of value at those locations. Oh well.
My first major heartburn has to do with Figure 1.1 of the most recent CSIRO drift publication linked below.
https://www.atsb.gov.au/media/5773567/mh370_ocean_driftiv_oct2017.pdf
This figure claims to show the cumulative probability of the detection of debris by the aerial search. My assertion is that there is no rational way to create such a graphic. Additionally detection is ambiguous without recovery. How could CSIRO possibly know that the debris shown in the figures of your recent post was not from MH370? None of that debris was ever collected and examined.
My second major heartburn has to do with figure 11.3 of the DSTG book. This figure shows the prior and posterior probability distributions after the flaperon was found. Frankly, I cannot discern any difference. The flaperon finding was given almost zero weight. Herein lies the basic weakness of the Bayesian methodology as applied to an MH370 terminus. Someone has to pull some weighting values out of their ass, and pretend that they are meaningful. Stone shows the same graphic along with the flaperon probability distribution is his NPS presentation (page 43) linked below. I had the same issue relative to the Metron AF447 search papers which showed the terminal probability with and without the assumption that the black box pingers had failed. Someone simply made a guess.
http://www.nps.edu/documents/103424533/106018074/Bayes+Search+for+Missing+Aircraft+NPS+20+Apr+2017.pdf/051a76bc-18cc-47a7-b8b8-52d92d618dfe
The reality is that we are no closer to defining a terminus with confidence than we were a couple of years ago. I was serious in earlier post relative to the failure of the human brain and conventional analytics as applied to this problem. We are simply not as smart as we think we are, and we are stuck in an analytical paradigm that is obsolete (valid math, but questionably applied).
@Victor,
I don’t have any other information about suspected debris recoveries than what is already in the public domain. I have collated all the AMSA, and later JACC, updates posted during the period of the surface search, into a single ‘journal’ PDF: that’s the source of my comment.
@DennisW: In the DSTG analysis, the posterior distribution didn’t change with the flaperon discovery because the assumptions for reconstructing the paths didn’t change. With the null result for the previous seabed search, we (independent investigators) are wiser. As for the probabilities for the surface search, we’ll find out more soon whether or not the debris from March 29 was ever recovered and/or considered. To me, the debris fields in particular look like crash debris, and the other objects look man-made. That makes all of them important. You might feel otherwise, but I see a slow progression of thought.
Thank you for the new article Victor. You have asked some very pertinent questions.
To me, it is notable that in the early days, there were significant voices calling for the search to be made further and further South along the 7th arc. I believe these opinions did succeed in moving the search area away from the real impact point.
The rational arguments are converging on an area North of the searched area and I really believe that significant progress has been made.
@Ulric
Progress is being made, but not in an elegant or deterministic manner. Cut-paste below from the Wikipedia summary of what constitutes a “conjecture”.
https://en.wikipedia.org/wiki/Conjecture
One method of proof, usable when there are only a finite number of cases that could lead to counterexamples, is known as “brute force”: in this approach, all possible cases are considered and shown not to give counterexamples. Sometimes the number of cases is quite large, in which situation a brute-force proof may require as a practical matter the use of a computer algorithm to check all the cases: the validity of the 1976 and 1997 brute-force proofs of the four color theorem by computer was initially doubted, but was eventually confirmed in 2005 by theorem-proving software.
Brute force is a good description. Time is, indeed, on our side. Eventually the solution will have no place to hide. No doubt about that. BTW, the four color theorem proof has been simplified relative to the original computer exhaustive solutions, but it remains a computer solution. I regard this as “cheating”, but as a practical matter, who cares?
It is an excellent article Victor. Timely too. The use of Griffin’s data to make the case for S30 is noted. Richard’s independent drift analysis concluding S29-S31 adds additional credibility. Perhaps the most compelling argument for S30 vs. S35 is that the RNZAF photographs are far higher resolution than the French satellite photos. You can’t look at the satellite photos and get a clear conclusion that the items are manmade (though “experts” did), much less aviation related. With the RNZAF photographs, one can easily see that some of the debris is manmade, and at least one or two could be debris from MH370. All very compelling.
@DennisW
I think “brute force” is not a good description of the reasoning applied up to this point. There are good reasons to have made exhaustive inquiries into the more likely aspects of the problem because one must eliminate the obvious before one looks for the non-obvious.
It has been clear for a long time that the geometry of the final turn south is the critical factor in determining an end point and there is still energetic debate on that subject. With the renewed attention on debris fields from the initial air searches we have additional considerations to make about an end point and of course part of that is to consider how a likely end point constrains the geometry of the FTS.
It is the sheer tenacity of the proponents here which will eventually tie the loose ends down but I think the method is far from “brute force”.
@Ulric
It is not important, but brute force is a perfectly acceptable problem solving methodology. I think the description of the method is almost an exact fit to what we are doing. I won’t comment any further, and you are free to have whatever opinion you like.
https://en.wikipedia.org/wiki/Brute-force_search
Bruce Lamon said…
Mr. Dolan, thanks for taking these questions.
Early on the Chinese (March 22) and the French (March 26) published satellite images of floating objects of significant size found in the vicinity of the initial Australian search area. Were any of these objects ever relocated or otherwise analyzed to see if they came from MH370?
JULY 14, 2014 09:14
Martin Dolan, Chief Commissioner (author) said…
No recovered debris has been identified as originating from MH370, and the search effort has not found any debris associated with the satellite images or other visual detections.
JULY 17, 2014 11:27”
@Bruce Lamon: Welcome, Bruce.
That quote seems to imply none of the debris seen on March 29 was recovered. This seems contradictory to statements that the surface search detected no debris. No recovery does not equate to no detection. I hope this becomes more clear in the coming days.
G’day Victor, thank you for another interesting and insightful article. The amount of time and effort that you are putting in – research, generating high quality content (including graphics and AV), moderating comments, contributing to discussions, etc – is not lost on me. Thank you for all of your hard and very diligent work.
To the topic du jour, I’d like to think that a search block centred on 30°S either side of the arc would have to be OI’s Plan B if the CSIRO’s 25k km² yields nothing. As I recall Richard had produced a 180 CMH at best holding speed solution that terminated at around 30.2°S 98.2°E on the 7th Arc. With regards to the surface search, the CSIRO’s “The search for MH370 and ocean surface drift – Part II“, Fig. 4.2. Efficacy analysis of the surface search (p.16), illustrates that the surface search coverage was not completely efficacious between 30°S – 33°S; interestingly, coincidentally or otherwise, there is a sizeable lacuna (0% probability of detection) of about 3,500 km² at about 30° 10’S 98°E.
@Mick Gilbert said: As I recall Richard had produced a 180 CMH at best holding speed solution that terminated at around 30.2°S 98.2°E on the 7th Arc.
I don’t recall that result. At this point, Richard and I both favor great circle (LNAV) or CTT paths with a descent and loiter between 18:28 and 19:41. We both predict nearly the same termination point for the path towards YWKS, which is what I used for the drift model results.
What do the ATSB and Boeing mean by “Maximum Range Cruise”?
In the publication “MH370 Flight Path Analysis Update” from 8 Oct, 2014, Figures 2 and 3 purport to show “Aircraft Performance Boundaries” based upon MRC (maximum range cruise) performance. Specifically, the figures shows several hypothetical flight paths based on an early (18:28) and a late (18:40) turn. The figures show the flight plaths for a range of altitudes, and they show the crossings of ping rings 4 through 7, plus the location of fuel exhaustion. These flight paths were computed using several assumptions, but not all are explained in detail. What was assumed regarding speed? Did the models use cost index = 0, which defines MRC? If so, it means that the Mach and airspeed changed along the path as the aircraft burned off fuel. What about temperature – was the GDAS temperature (or equivalent) used, or did the calculations use ISA temperatures? Was a fixed point for the FMT used – if so, where? For cost index = 0 routes, total weight is needed – what was used at at FMT? What route was used previous to the FMT?
As you can tell, there are lots of options, and the ATSB report provides no useful information for most of them. I have explored a variety for the case of an early turn and altitude (or more specficially, flight level) of 35,000 feet. Here are the results.
First, did the models use the GDAS (or equivalent) temperatures plus a cost index = 0 model? Definitely not. In order to make such a model fit the data, it is necessary to have the FMT occur at latitude of 8.1 degrees. Reaching such a latitude from the last (or even penultimate) radar point requires ungodly speeds before the FMT and makes a mockery of the Lido radar image. A similar (albeit not as compelling) result follows if one assumes that the models were based on an ISA (i.e., a standard atmosphere).
Instead, it seems that Boeing assumed either a constant true airspeed or a constant Mach (which are equivalent for constant altitude and ISA), and tuned it to give the maximum range. I can get good fits using either ISA or GDAS temperature models. For an ISA model, I find a constant Mach of 0.773 works well (FMT at 6.9,95.7). For a GDAS model, I find a constant Mach of 0.757 (similar FMT). In both cases I impose a requirement that the pre-FMT path pass close to the last radar position. For a GDAS model, I find that fuel exhaustion occurs at latitude -37.0 degrees, in reasonable agreement with Fig. 2 latitude of -36.9. Now, Mach 0.757 at the FMT is lower than the holding speed for that altitude and mass (I use mass = 208.9 tons), but perhaps that is OK.
Anyone seeking to invest > $10 million in a resumed search for MH370 may wish to first obtain clarity regarding how the performance limits of the aircraft were derived. Publicly available information regarding previous calculations of such limits is not sufficient to allow verification of those results with any confidence.
@Andrew,
Thank you for responding to my question on the SLOP. I found that an initial turn of about 55 degrees fits the data better than 45 degrees for the 12 NM SLOP, but this small difference is immaterial to fitting the subsequent path.
@VictorI
@DennisW: In the DSTG analysis, the posterior distribution didn’t change with the flaperon discovery because the assumptions for reconstructing the paths didn’t change. With the null result for the previous seabed search, we (independent investigators) are wiser.
That is a strange statement to make when the graphics clearly implied an a posteriori and post posteriori result. No matter, I am pretty much done with you people.
@sk999,
You said: “Anyone seeking to invest > $10 million in a resumed search for MH370 may wish to first obtain clarity regarding how the performance limits of the aircraft were derived.”
I have requested this info several times from ATSB, but it has not been provided. My conclusion is that ATSB literally does not know what assumptions were used by Boeing and do not now appear interested in learning what they were.
At this point, I also think the question is mostly moot since the previous unsuccessful search covered the portions of the 7th Arc near the Boeing maximum range curves.
@DrB
My conclusion is that ATSB literally does not know what assumptions were used by Boeing and do not now appear interested in learning what they were.
It would be interesting to know how the ATSB directed the search priorities. Did they meet every day and vote on it? If so, who were the voting parties? Did Dolan make a decision by himself? I ask out of curiosity, not to assign blame.
@Mick Gilbert,
You said: “ As I recall Richard had produced a 180 CMH at best holding speed solution that terminated at around 30.2°S 98.2°E on the 7th Arc.“
Your recollection may be wrong as Victor has responded. don’t think that any proposed CMH route with no auto-pilot changes post-19:41 has been demonstrated to match BTO/BFO/endurance/GDAS. I have searched extensively for one, but so far unsuccessfully. The route curvature at the 4th to 6th Arcs is huge, being the sum of the increasingly eastern magnetic variation and the strong westerly winds. The speed reduction needed to match the 6th Arc appears to be much larger than can be produced by any of the FMS speed modes. If anyone thinks an acceptable CMH route exists, please post the route specifics so that others may assess it.
@DennisW,
I was curious once, too, but knowing the ATSB decision-making processes does nothing to help find the aircraft.
@Mick
@DrB
Of course Victor has a 180 CMH path paper where there is a constant -100 ft/min descent rate
@IG members:
1: these images would have been of vital importance in July, 2014, when the JIT was trying to convince us that their models pointed to the s30 area (an area the IG concluded – strenuously and publicly – their models emphatically counter-indicated). In fact, any promising debris would have been rushed to the forefront, to combat the widespread criticism of this area. Did any of the officials whose minds you were working so hard to change ever point to these images in support of their (otherwise absurd) case? If not, why not?
2: what was the full process for assessing each collected image? What was the decision tree? Looking to compile the full list of people who scrutinized these photos, the full set of tests to which each image was subjected, and the follow-up actions taken, if any.
3: what official conclusions were reached on each image? Did the people who sent you these images indicate what THEY thought the probability was that each (or any) was a piece of MH370?
If I’m asking too much of you, please just send me the full names and contact info of the folks who gave you these images, and I will happily follow up with them directly.
Many thanks in advance.
@Brian Anderson: thanks very much for responding. Fear not: even though Duncan Steel intimated fairly strongly that the JIT was following the IG’s lead when it moved the search to s38, I myself have never made any such assumption. My work at present is merely to document who said what, when. The IG members have been – and continue to be – central to the science of MH370, and so their public statements deserve to be included in any such compendium. That’s all. Of course I’m not auditing the IG – I’m merely using the IG’s real-time statements to benchmark – audit, if you will – official search conduct.
Re: Acoustic pings at s21: great to hear we agree fully on the underlying science. But I am compiling real-time public statements. Please don’t go to any more trouble on my account – I appreciate the time you took to respond already – but could you at least point me in the general direction of public statements you made in real time which rubbished the claim? I’m guessing the Duncan Steel forum – and am happy to rummage around myself, regardless – but would deeply appreciate the general guidance. Many thanks.
Re: my other questions (re: claims surface debris waterlogged and sunk, and also drifted to Sumatra, and that the flaperon vindicated s38, etc): any public record of your real-time reactions to those now-discredited official claims?
It must seem I’m picking on you just for responding, Richard. That would be a real shame, because I truly appreciate your candor on the acoustic pings. I am just as interested in your colleagues’ public statements on those topics, so the above questions – are directed to them, as well
@VictorI
I have a request; could you add the flaperon-like piece with it’s coördinates in your figure 1 for context?
And add the original image(s) to the other images?
The Bernard Lagan article mentions a blue panel. Back then I and others manipulated the picture on it’s contrast, light and color.
It was concluded based on this that the panel was actualy white but largely slightly submerged with one corner-area sticking above the surface.
It therefore possibly could have been a high-windage piece traveling faster and further north than other pieces.
@victorl
Thanks for sharing this insightful analysis.
Some questions/observations:
(1) do you know the scale / (angular) resolution / angle of incidence of those images at the sea surface?
(2) the simulation results are very suggestive. Perhaps too suggestive?
For example,
(2a) did you tune the initial conditions to match the outcome as observed by the aereal search? The initial field appears quite spread out from the video, is this realistic? Suppose a nose-down impact with a concentrated debris field. Would it produce the same dispersion pattern on the 29th of march?
(2b) I think it would boost the significance of your findings substantially if you run another control study of many potential (including original southerly latitides) impact site debris along the 7th arc.
If these end up in totally different locations (like perpendicular distance) then it adds more weight to your hypothesis, and, incidentally, also to a more notherly impact site.
(3) how significant is this amount of manmade debris (not beingbmh370 related) in this corner of the SIO? is it usual to find that stuff there at any time, or not?
kr,
hendrik
@Hendrick: Welcome. To answer your questions:
1) I don’t know.
2a) I chose a 15-NM radius to select the seeding so I had a fair number of particles (31). Ideally, the study would be done with more closely spaced particles.
2b) I have asked David Griffin of CSIRO to do this. I would call my study a “poor man’s” drift analysis that made use of existing results.
3) Others may know the answer.
@Ge Rijn: You have the data to do your own comparisons. The “blue panel” is shown in images 5809 and 5810. From the data file I provided, the GPS coordinates are -32.476367,97.824467.
@TBill: The 180° CMH path with constant descent that I studied two years ago is not likely to have the required endurance for a constant speed of 310 KIAS.
@VictorI
Thank you. I will. I think it’s an important piece also that can not be left out of this context.
@DrB
“If anyone thinks an acceptable CMH route exists (INSERT- without pilot input after 19:41), please post the route specifics so that others may assess it.”
I am interested CMH paths but any such path from me would have pilot A/P setting changes after Arc5 to buck the winds and other strategies. Let’s face it, for all the gallant efforts, the unintended flight paths do not seem to explain what we are seeing. This is the elephant in the room.
@All,
A further set of images were released by the Australian Defence Forces following the FoI request by Twitter user @aussie500.
The images were poorly collated, encapsulated in PDF docs, some with original meta data, others not.
However, I have arranged these RAAF images as ‘collection‘ at Flickr. The RAAF AP-3Cs were tasked, on 24th and 26th April, to an area in the vicinity of S27.5º E99º. Please review and report back anything of interest.
@TBill said: I am interested CMH paths but any such path from me would have pilot A/P setting changes after Arc5 to buck the winds and other strategies. Let’s face it, for all the gallant efforts, the unintended flight paths do not seem to explain what we are seeing. This is the elephant in the room.
I’m not sure I understand this. In the last article, I showed that there are many paths with autopilot after 19:41 that match the satellite data, intended path or not. (I think the path was intended.) The question is how to choose which one(s) to search.
@Victor
Mainly I feel CMH is the most likely unintentional flight mode, so all the work has tended to rule that out (because CMH does not fit well). I am concerned that even intentional flight without pilot input after 19:41 may not be true. So one thing I like about ATSB search zone (32.6-36S) is I perceive it may capture more possibilities of pilot action before and/or after 19:41.
@TBill: I believe pilot input after 19:41 relaxes constraints and increases the range of possible impact points.
However, you made another observation. An unintended route to the SIO would most likely follow either a path of constant magnetic heading or track, since a path of constant true heading or track would require the HDG REF switch to be changed from NORM to TRUE. This switch is protected by a cover, and would have to be intentionally operated.
On the other hand, automated flight paths of constant magnetic track or heading do not match the satellite data due the curve to the east, which requires a reduction in speed to match the BTO data.
This (among other reasons) makes it more likely the path to the SIO was intentional.
@Ge Rijn,
Here is a software generated high resolution “edge detection” image of the “blue panel” photographed at 32.476367°S 97.824467°E.
The shape is strangely similar to that of an intact flaperon, but after spending many hours “playing” with it, I came to the conclusion it was a piece of fishing gear – probably rectangular, but twisted by the sea/swell condition to appear as it did.
@Mick @Victor
I have reconstructed over 700 possible MH370 flight paths using all navigation modes, altitudes, speed schedules and starting points on the 2nd Arc at 19:41 UTC, but using the GDAS air temperatures and wind speeds and directions, in order to see how well the flight paths fit the BTO and BFO data.
Victor is right that I favour CTT and LNAV flight paths. Especially LNAV to YWKS or 40S 100E, which end at around 30°S near the 7th Arc.
On 26th August 2017, I wrote a piece entitled Occam’s Razor Revisited. I postulated an early FMT, just before the 18:39 UTC SATCOM call, overflying waypoint ISBIX and continuing in CMH navigation mode at 180°M at flaps up holding speed. The Magnetic Variation used in this analysis is as of 1st January 2005. According to Boeing, the data available on MH370, was from 2005. The Service Bulletin regarding the Air Data Inertial Reference Unit (ADIRU) 2005 MagVar tables was 777-34A0138 dated 22nd November 2005. The CMH navigation mode results in a curved flight path, where the curvature depends on both the Magnetic Variation (MV) and the Wind Angle (WA). The curvature increases as the MV increases. The MV is around 1°W at 18:41 UTC and increases to around 12°W at 00:19 UTC. The Wind is initially from an easterly direction and later from a westerly direction and strengthening. This results in a higher WA in the latter part of the flight and increased curvature of the flight path. The flight path ended at 30.2°S near the 7th Arc. The RMS BTO Error was high at 10.6 km, but the RMS BFO Error was good at 2.7 Hz.
However, I made an error, which Mick kindly pointed out at the time, as I used the Boeing 777-200ER holding speed (flaps up) table for GE90-94B engines, instead of the RR Trent 892 engines.
@Richard
Is there a link to Occam’s Razor Revisited?
I also like CTH or CMH from ISBIX.
@DrB,
On November 7, 2017 at 8:19 pm you wrote: “The GDAS wind errors are given (using the only published value I could find) as 5.7 kts per horizontal axis (out of two possible horizontal axes). ”
Where did you find this number, and what does it represent specifically? Overall RMS? Near-surface? In what domain? When?
Bobby Ulich,
Thanks for trying to get clarity from the ATSB. R.E. Boeing’s performance limits, Figs. 2 and 3 of the Oct 2014 “Update” report, these figures make the most sense if they are computed using the actual GDAS (or equivalent) temps along with winds and a constant Mach at the assumed altitude. The reason that these figures are interesting is not because they provide useful constraints on the location of MH370 but rather that they provide useful constraints on fuel models. They are actual Boeing calculations covering a range of Mach and altitude.
Oh yes, for what it’s worth, of the seven routes depicted in the two figures, four require a Mach that is less than 0.73, the value that the DSTG took as its lower limit. That is one of the many reasons why anyone investing over $10 million in a future search (not me!) should ask pointed questions about how the current recommended search zone was defined.
@Victor: RE: “This greatly increases the probability that the impact was further north than the 25,000 sq km area recommended by CSIRO for the new seabed search.”
This should read “the impact was further outside the 25,000 sq km area “.
I cannot see in which way these images are sufficient evidence to suggest a particular location or even alter the location probabilty. The probability of the area you suggest by inductive reasoning is not higher than some other locations that were not searched. The scenario you suggest to end up at this location has its own inherent probability and assumptions too. Unless some photographic forensics turns up with possible identification of parts, such bias could mislead the search activities. The activity to recommend should be to further perform forensics on these photos before anything else be suggested. We need to be mindful of the inductive reasoning limits and implications.
@Oleksandr,
I could only find one published test of GDAS wind speed accuracy. It was a comparison with radiosonde data, and I think it was done in England. I believe the DSTG book references it. I was quite surprised there was not a NOAA report providing GDAS error data.
@sk999,
Yes, I agree that the Boeing maximum range calculations could be used as a test for alternative fuel models, if Boeing had defined all the external conditions they assumed, but they did not. Still, we already have public and leaked documents providing Boeing calculations of fuel consumption for a wide variety of conditions, and these are sufficient to achieve good accuracy.
@VictorI,
You said: “On the other hand, automated flight paths of constant magnetic track or heading do not match the satellite data due the curve to the east, which requires a reduction in speed to match the BTO data.”
This is incorrect. There is a CMT route I published about a year ago at Holding speed that matches BTO and BFO. There is sufficient decline in airspeed as the weight lightens to match both a CTH and a CMT route (which are nearly identical in this case). It is fully automated and requires no pilot inputs after 18:41.
It is incorrect to conclude that all routes which depend on direction (not destination) cannot satisfy BTO and BFO. I don’t believe any CMH route and autopilot airspeed mode can match the BTOs, but a CMT route can.
@sk999,
In my opinion, you don’t need any drift studies or details of Boeing maximum range calculations to define the new recommended search zone.
Basically, you only need two things: the 00:19 BTO/BFO, and a belief that the previous search would have seen the wreckage if it had been in the previously searched area.
@Andrew,
Properly fitting the BTOs/BFOs at 18:25-18:28 requires a reasonably accurate model of the turn rate at M0.84. I have been using a 25 degree bank angle limit, and allowing several seconds for the turn to reach that bank angle limit. The FCOM says the AUTO position on the MCP selector for the bank angle limit is 15-25 degrees, depending on TAS. I have two questions:
1. Do you know the relationship between TAS and the applied bank angle limit?
2. Is this same equation used by the FMC when in LNAV?
Bobby Ulich,
You state, “In my opinion …”
I care not a whit about your “opinion”, and in return, nobody should care a whit about mine.
You also state, “you only need two things: the 00:19 BTO/BFO, and a belief …”
Beliefs are what you learn in Sunday School and Church. They have no place in definining search zones.
@Barry Carlson
Yes I remember now I see that picture again. It was concluded (like you say also) it was probably some kind of pallet used on fishing-vessels. But it was never positively identified for the item was also not recovered.
I kept my doubts for the images (5809 and 5810) show details when zoomed in and contrast, light, color and sharpness adjusted that eerily resemble a flaperon. Especially image 5809.
In this image the ‘south’-short side clearly shows a curve along the submerged side suggesting a leading- and a trailing edge. It also shows sections in this side resembling what is seen in the real found flaperon.
Problem was- and is image 5810 does not show this curve anymore and the sections changed in this image. Hard to tell, could be water and light disturbance.
Further both images show three lines on the upper-surface that suggest sections resembling same sections as on the found flaperon; one near the ‘leading-edge’, one on ~1/3 and one on ~2/3 where in a real flaperon the trailing edge part starts.
I know it’s not by far any proof and it probably cannot be proven ever but the same goes for ‘fishing-gear’.
Imo it’s enough not to dismiss this piece in the context of new search effort.
Ulich,
You state, “… we already have public and leaked documents providing Boeing calculations of fuel consumption for a wide variety of conditions, and these are sufficient to achieve good accuracy.”
Bullfeathers. We have a mixed bag of documentation only partially covering a wide, multidimensional space. Paranoia is good. You can never have enough tests. You know that.
@DrB
” I don’t believe any CMH route and autopilot airspeed mode can match the BTOs, but a CMT route can.”
Now you got me wondering if CMT is intentional or not…in any case, I would say try CMH until 22S and then click over to CMT due to the wind.
@VictorI @all
This article states the new search is going to start in January;
https://www.perthnow.com.au/news/wa/wa-urged-to-be-alert-for-plane-debris-ng-b88673410z
Any value in this statement?
Anyway nice to read Blaine Gibson is back again.
@DrB
RE: “1. Do you know the relationship between TAS and the applied bank angle limit?”
The bank angle is limited to lower values at high TAS. In the cruise, for example, the bank angle is limited to 15° in HDG/TRK SEL with AUTO selected. I don’t know the exact relationship; it’s not documented in the manuals.
“2. Is this same equation used by the FMC when in LNAV?
No. The FMC also considers the available thrust limit and will limit the bank angle to ensure the thrust limit is not exceeded. If there is plenty of excess thrust available, the FMC will command a bank angle up to 25°, regardless of TAS.
@DrB you said “Basically, you only need two things: the 00:19 BTO/BFO, and a belief that the previous search would have seen the wreckage if it had been in the previously searched area.”
Dr Bobby’s, you are allowing your entrenched prejudices and preconceptions cloud your otherwise rational thinking processes. You are like many others on this forum – you are turning a blind eye to that humongous elephant that lurks in the shadows in your study (malicious, premeditated pilot action).
Your fuel model is an exceptionally valuable tool, yet it’s value is being called undermined and cheapened by irrational conformation bias (no, I haven’t swallowed a lexicon, and I’m not describing myself either, I’m just very angry and mystified)
The wreckage wasn’t found in the search area, agreed. The failure was down to the SSWG’s fatally flawed understanding of the problem, not down to the pilot’s failure to carry out his mission. The pilot did not fail in his mission. He did not end up flying a nonsensical autopilot mode that defies all attempts at identification. He flew in LNAV all the way to MEFE – your fuel model has proved it! Victor’s analysis of the SIM data shows unequivocally that the pilot was playing around with LNAV paths into the SIO, paths that didn’t use the standard waypoints, and yet this finding is also being tacitly ignored.
Then there’s the fiasco surrounding CSIRO’s recommended search areas. History will judge this to be one of the biggest distortions/manipulations of the data by a government body under political pressure.
The pilot did not fail in his mission.
Reassess the data from an unbiased perspective just for once.
@DrB
Further to my last post, @Don Thompson pointed out that there is a bank angle vs TAS graph in the 777 Training Manual used by the engineers.
The bank angle limit is 25° at speeds up to 332 KTAS, then decreases linearly with TAS, reducing to 15° at 381 KTAS and constant at 15° thereafter.
Oops, that last sentence should read, ‘decreases linearly with increasing TAS…”.
HB said: I cannot see in which way these images are sufficient evidence to suggest a particular location or even alter the location probabilty [sic].
If the first 25,000 sq km produces a null result, the search will continue elsewhere. Identified man-made debris increases the probability that the identified objects are related to MH370. We don’t know what that probability is. On the other hand, the satellite images that CSIRO is using to recommend its search area might not even be man-made.
The activity to recommend should be to further perform forensics on these photos before anything else be suggested.
As I said, I alerted the ATSB and CSIRO about the possible significance of the images. I am waiting to hear back.
I asked contributors here to recommend where OI should look next after the first 25,000 sq km. I don’t recall that you responded.
@DrB: My recollection is that your route required an overflown holding fix (with offset) in LNAV/VNAV mode to get the holding speed schedule. Andrew’s simulation results say a route discontinuity produces a path with constant CMH. Is my memory wrong?
@TBill
To shim in. It seems to me consensus is building on an intentional flight also after FMT. An intentional flight-path also after FMT with an open (uncontrolled) impact after fuel exhaust or a controlled intentional high speed descent and impact.
Imo if an intentional flight path after FMT is seriously considered then also an all controlled end-of-flight should be seriously considered but
including the possibility of a recovery after the steep descent and a glide after. And the possibility of a chosen specific crash area.
I think this is of great importance to make decisions on the width of the new search area(s).
@Barry Carlson
Regarding the flaperon-like item images 5809/10 I wanted to ask you if you know of more images of this item.
In my memory there was also one in the crest of a wave with one side sticking out further.
Sadly I lost all those images due to a crashed hard disk. Maybe you have them all (if I’m right..).
Then; did you consider the images were taken through an aircraft window probably causing additional disturbance in detail and perspective.
Image 5810 would be more affected for it was captured near the window edge.
Anyway I still think it’s a very interesting item. Shape and dimensions (and some details I mentioned) are very similar to an intact flaperon especially regarding image 5809.
I thought of one method of ruling out, this could have been a flaperon: Testing the floating attitude of an intact flaperon.
If it clearly does not float in a mainly flat attitude but also sticks out above the surface with it’s trailing edge it cann’t be a flaperon imo.
I sure would like to see this experiment happen. It could even explain a lot more. F.i. the barnacle growth all around.
In this case I have a possible other explanation of the trailing edge damage and some other damage which is visible on the found flaperon.
Just some thoughts to consider. Hope you (and others) will.
In case you missed the article, Blaine Gibson is continuing his search for MH370 along the shores of southern Western Australia with the help of Chari Pattiaratchi, who believes that debris that drifted towards South Africa could have returned to the southern shores of Australia due to an eastward current.
Although any recovered debris is not likely to help determine the point of impact along the 7th arc, the debris may give us a better understanding of the crash.
If something gets found now or later along those WA shores it would rather indicate a more northern (above 36S) impact area too imo.
It would be found in the rebound but not early arrived debris.
Any new found debris anywhere could help. Especially on the impact conditions I agree. But the longer time goes by drift-analizes on new found debris become more complicated and unreliable.
@Victor
I saw the article you linked above as well as a couple of others on the same subject. All of the articles referred to an Ocean Infinity January 2018 start date. I wonder where they are getting that information.
@DennisW
What predicts your Weibull-distribution now?
It’s been almost a year since new debris has been found..
@Ge Rijn
As far as consensus, we have to differentiate between all interested observers and MY gov’t officials. ATSB, IG, others. Bottom line no consensus in the official search, in fact strong preference for passive flight.
@DennisW: I don’t know where they get their information. I don’t trust anything the press writes on this subject, so I ignore it. However, I do have it from reliable sources that OI is planning to start the search in January. Of course, the search will only begin after a contract is signed, so there’s still time for the plan to break down.
@Ge Rijn
I have not looked at it since the last piece was found. I think Dec of 2016? Can’t recall exactly.
I think it is not meaningful at this point if it ever was. What Weibull and Poisson can’t incorporate are the search effort (which has gone to zero without Blaine) and the variability of the shoreline (the easy spots have been cherry picked).
@TBill
Here is the requested link to the piece entitled “Occam’s Razor Revisited” with the CMH 180°M from waypoint ISBIX dated 26th August 2017:
https://www.dropbox.com/s/hob5bvbz0nksry4/Occam%27s%20Razor%20Revisited.pdf?dl=0
@Richard: Is the 11 km RMS range error for the BTO calculated as a one-way or two-way error? That is, is the RMS BTO error 37 μs or 74 μs?
@TBill
Ocean Infinity is most that matters now imo. We have to feed them as much possible input as possible imo.
The IG in name of VictorI supports at least an intentional flight path after FMT if I understand him well.
The crucial disagreement in assumptions is on the end-of-flight scenario at this point imo.
The final BFO’s show a 8 sec. snapshot of an increasing high speed descent from probably high altitude. Suggesting a high speed impact near the 7th arc. The plane has not been found near the 7th arc where it was expected based on this data.
The found debris shows conflicting damage and other indications that also suggest a nose-up, ~wings level, relatively low vertical and horizontal impact speed is possible. In other words; a glide after the steep descent.
On this issue I think it will be impossible to reach consensus for it cann’t be proven unless full forensic debris examination is done and made public.
Anyway imo it would be a missed chance if OI does not take this possibility into account and will not consider to look at areas up to ~60 miles on each side of the 7th arc.
@Andrew,
Thank you for providing the details on the bank angle limit. I will incorporate those equations into my turn model. The SLOP maneuvers may not be affected because they are done in LNAV. At M0.84 at FL340 at +10C, would the thrust limit reduce the bank angle limit below 25 degrees? If so, can you guess what the bank angle limit would be?
@sk999,
You said: “You also state, “you only need two things: the 00:19 BTO/BFO, and a belief …”
Beliefs are what you learn in Sunday School and Church. They have no place in definining (sic) search zones”
You have missed my point. Let me explain further. In this case my “belief” is a considered scientific judgment that ATSB performed frequent and effective procedures to confirm that the sonar data would have detected wreckage in the search area, had it been present, with a probability close to 100%. These procedures included independent reviews of all sonar data to assess completeness and quality. In addition, they set up several sonar targets off the WA coast, and the system was verified to be properly operating at the start of every sortie to the search area. The contract was based on data quality and coverage, so Fugro knew that they would not get paid for unacceptable data. In my opinion, ATSB had an excellent data quality assurance plan. This provides confidence that it is very unlikely that the 120,000 sq. km search missed the wreckage. This conclusion is scientifically sound, insofar as I can judge. It is not an act of faith to accept this conclusion, since ATSB has provided both the data and descriptions of the processes used to collect and analyze it.
You also said: “Bullfeathers. We have a mixed bag of documentation only partially covering a wide, multidimensional space. Paranoia is good. You can never have enough tests. You know that.”
First, I very much doubt that you know what I know. Second, I have requested the Boeing assumptions from ATSB on multiple occasions. What have you done? Third, those assumptions will not be forthcoming, so do you suggest we do nothing because we don’t have all the information? I am moving forward with what we have. I spent six months creating a generalized fuel model that is useful. Let me know when you have done a generalized fuel model, and we’ll compare results. Sure, Boeing can do a little better than my fuel model, but the 1-2% errors in my model are close enough to winnow out the wheat from the chaff in route searches. As Sergey Gorshkov, Admiral of the Fleet of the Soviet Union, once said: “Better is the enemy of good enough.” I have always found that to be useful advice, and I follow it.
@VictorI,
You said: “My recollection is that your route required an overflown holding fix (with offset) in LNAV/VNAV mode to get the holding speed schedule. Andrew’s simulation results say a route discontinuity produces a path with constant CMH. Is my memory wrong?”
Your memory is not wrong on that point, Victor, but I think you may be generally conflating two results I obtained about a year ago.
One result was a CTH route starting at ANOKO (or up to 15 NM to the right side of ANOKO following a SLOP). That scenario involved a “failed” Hold at ANOKO. Andrew found in his simulator test, and using follow-up information, that an EOR or Route Discontinuity resulted in a constant heading mode, with the reference depending on the NORM/TRUE switch setting. Since none of us has come up with any reason for the PIC to change from NORM to TRUE, we assume that a CMH route would generally result from a route error. We also know that the Holding speed would have been set in advance before the route discontinuity would occur (due to failure to remove a route offset). Thus, there is a means to achieve a CMH route at approximately 181 degrees (depending on the crosswind at ANOKO) at Holding speed. However, to the best of my knowledge, no one has ever been able to match any autopiloted CMH route to BTO and BFO with no pilot inputs after 19:41. If you allow maneuvers (changes in course, speed, and altitude) after 19:41, you can fit almost any route and you can hit almost any point on the 7th Arc. That is why I have concentrated on finding solutions with no PIC maneuvers after 19:41.
The second route I found that matched BTO and BFO data was a CMT route at ~180 degrees. In fact, except for the 5th-6th Arc leg, The CTH and the CMT routes are virtually the same. The reason for this is that the crosswind deflection in this case closely mimics the magnetic variation. While I cannot explain the origin of a CTH route, I feel there is a possible explanation for a CMT route. If the PIC simply wanted to fly south, the easiest means of doing this is to set 180 degrees TRK HLD using the MCP. That gives you a 180-degree constant magnetic track (CMT). If this happened, the flight into the SIO was a result of an intentional act to set a direction, but not a destination. I believe it is also possible to set Holding speed without commanding the racetrack pattern, and I am sure someone will correct me on this if that is not possible. It does appear that a slower speed, and a speed that declines with time is required in order to fit the BTOs with a curved track like CMT/CTH/CMH. Possibly the PIC wanted to reduce speed and, at least temporarily, head south. Now that I have a fuel model and an accurate value for PDA, I will revisit that CMT route and see how well the endurance can be matched while maintaining acceptable BTO/BFO residuals.
@Rob,
You said: “He flew in LNAV all the way to MEFE – your fuel model has proved it!”
Actually, no. My fuel model demonstrates it is possible, not that it was done.
You also said: “. . . you are turning a blind eye to that humongous elephant that lurks in the shadows in your study (malicious, premeditated pilot action).”
Actually, no, I am not ignoring it. In my last paper I addressed this scenario specifically, and I gave my assessment of its positive attributes and its deficiencies.
Here is my overall perspective:
1. There are currently three scenarios that appear to match the BTO/BFO/wind/temperature/endurance data.
2. Scenario #1 is an extended loiter plus a geodesic to a far-south waypoint. One must guess the final waypoint in order to generate a specific 7th Arc location, so there are multiple solutions. The sim data do not provide a specific far-south waypoint.
3. Scenario #2 is a curved route starting due south in the vicinity of ANOKO. This route does not use waypoints, and it has a unique solution.
4. Scenario #3 allows maneuvers after 19:41. In this case, almost any 7th Arc location is possible, and all solutions depend on multiple assumptions.
5. The search strategy for Scenario #1 is to look at multiple specific areas.
6. The search strategy for Scenario #2 is to look at one specific location.
7. The only suitable search strategy for Scenario #3 is to cover the entire 7th Arc within fuel range.
8. The drift studies are useful in defining some broad arc limits, and the end point for Scenario #2 falls within the CSIRO recommendations, as do a few of the Scenario #1 end points.
9. The ATSB recommended new search area is compatible with an efficient initial search strategy for all 3 scenarios.
10. As Victor has suggested, it would be helpful to know specific locations for the first 2 scenarios within the proposed new search area so that those locations could be prioritized and searched first.
11. In my opinion, there is nothing that can be done to refine Scenario #3.
12. Victor and Richard have already proposed end points for Scenario #1 based on their waypoint assumptions.
13. I am continuing to refine Scenario #2.
Rob, I don’t see that my pursuing a particular route that seems to fit all the data is biased. In fact, I would argue that it is the one area that currently deserves more attention.
@DrB
Yes, nice summary above.
I used to get annoyed by posters on mostly the JW blog who are fond of saying that since the wreckage was not found in the original search area that the ISAT data must be wrong. I don’t get annoyed as easily anymore.
Ge Rijn: The notion that a live pilot first waited for fuel exhaustion to land in the water is contrary to any deliberate rational behavior. To assume the pilot then pushed the stick forward enough to reach a descent rate of 15,000 ft/min, and then pulled up to glide 60nm is even more ridiculous. What could possibly be the reason? It looks like nothing more than a desperate attempt to keep the soft landing myth alive. There are multiple lines of hard evidence and observations contradicting this fantasy. There is zero evidence to support it. A few facts:
1. There was no IFE logon at 00:21 as expected. A 60 nm glide would have lasted until at least 00:30 or later. How do you explain away the fact that there was no IFE logon at 00:21? There was more than enough fuel in the APU line to keep the AES alive until well after 00:21.
2. The right flap was retracted when it separated (probably prior to fuselage impact). That means there was no slow, flaps down water landing attempted. Your repeated claims that the damage indicates a slow water landing is not consistent with any known facts.
@Victor
@DennisW: I don’t know where they get their information. I don’t trust anything the press writes on this subject, so I ignore it. However, I do have it from reliable sources that OI is planning to start the search in January. Of course, the search will only begin after a contract is signed, so there’s still time for the plan to break down.
The notion of a reward based fee, still presents a fascinating consideration to me. Of course, my fascination threshold is quite low.
My latest thinking goes something like this. The value to Ocean Infinity is the marketing value associated with finding the aircraft. At least that is how I would look at it as CEO. If I could simply break even on the cost of the effort, I would regard it as a huge success.
So postulate that the aircraft is very very likely to be found in an area of X sq km which could be searched at a cost of $Y. X being subjective, of course, but arbitrarily make it as large as the area already searched by Fugro. Make the additional assumption that OI is twice as efficient as Fugro (seems reasonable), and you have a cost of $80M to search X. If the area X has uniform probability of discovery you have a 50-50 chance of finding the aircraft searching X/2. So reducing OI’s “bet” to a coin flip it would be more than reasonable to peg the reward at $40M. With a find in an X/2 area they recover their costs, and reap the marketing benefit. So it comes back once again to a $30M to $40M reward making some sense.
An unknown is how much of the $160M was actually paid to Fugro. Certainly not all of it. The $160M includes the cost of the air search and services to Fugro provided by Australia. Hence the shading to a span below $40M. It is also likely that the area X is not uniform relative to a find probability, and OI could derive additional benefit by an intelligent choice of search priorities.
“@Richard: Is the 11 km RMS range error for the BTO calculated as a one-way or two-way error? That is, is the RMS BTO error 37 μs or 74 μs?”
@Victor: The RMS BTO Range Error is actually 10.6 km or 35 µs as a one-way error.
BTO Range Error is the Predicted Aircraft Satellite Range (km) – the Calculated Aircraft Satellite Range (km).
For example, at 19:41 UTC in this scenario:
Predicted AS Range = ((‘BTO (us)’ ’19:41:02.906’−’Bias (us)’ ’19:41:02.906′) × Speed of Light (km/sec) ÷ 2 ÷ 1000000) − ‘Perth Satellite Range (km)’ ’19:41:02.906′ = 36,745.6 km.
Calculated AS Range = SQRT((POWER(‘Aircraft Satellite Rx (km)’ ’19:41:02.906′,2) + POWER(‘Aircraft Satellite Ry (km)’ ’19:41:02.906′,2) + POWER(‘Aircraft Satellite Rz (km)’ ’19:41:02.906′,2))) = 36,745.6 km.
BTO Error 19:41:02.906 = 0.0001 km
BTO Error 20:41:04.904 = -16.4949 km
BTO Error 21:41:26.905 = -2.3793 km
BTO Error 22:41:21.906 = 15.6426 km
BTO Error 00:10:59.928 = -2.9454 km
BTO Error 00:19:29.416 = -11.3691 km
@DrB
I would say Scenario #3 has two sub-scenarios (a) “SIMPLE CASE” – Flight still goes mostly straight (probably 180S) but heading and altitude/speed changes account for Arc5 to Arc6 distance and fuel management, and (b) COMPLEX CASES- involving huge mid-course turns like envisioned by some (eg; TimR to JAKARTA, or Ed Baker who sees a 90-deg East turn at about 20-22S to stay out of the cloud cover in order to see water surface).
@DrB: Your long comment proved my point. I said: “On the other hand, automated flight paths of constant magnetic track or heading do not match the satellite data due the curve to the east, which requires a reduction in speed to match the BTO data.” You said, “This is incorrect.” Yet, you have not produced a result which supports your statement.
You agree that the automated, CMH route will not satisfy the satellite data. You state that the automated CMT route satisfies the satellite data, but requires a holding speed schedule in which the speed declines with time. I submit there is no way to do this, as the holding speed for a path to the SIO requires VNAV, LNAV, an offset, and an overflown hold fix, and that would produce constant magnetic heading unless the HDG REF switch was deliberately changed to TRUE.
Until you provide proof otherwise, I will continue to claim that there are no automated flight paths of constant magnetic track or heading that match the satellite data. This means the only viable paths are either true track, true heading, or great circles. And this makes it very unlikely that the plane was unintentionally flown to the SIO.
@Richard: The way you define BTO error in km, the RMO BTO error is 73 μs for your CMH case. That is 2.5x times the observed standard deviation of BTO error observed in a large population. I don’t think that is likely.
I agree with you that CTT and LNAV modes are more likely.
@DrB In reply to my comment “He flew in LNAV all the way to MEFE – your fuel model has proved it!” you said “actually, no. My fuel model demonstrates it is possible, not that it was done.”
Yes you are right. I would have been more correct for me to say that it permits (or allows) it.
It allows the pilot to choose to fly at optimum speed and altitude along a great circle path that passes within 5 miles if the DSTG Bayesian hotspot’s centre, even though the DSTG did not have the advantage of a fuel model fine tuned to that specific aircraft in those specific atmospheric conditions.
It allows the pilot to fly in LNAV toward manual waypoint S40.50, E88.00 which as it turns out was evidently carefully chosen to be safely beyond the aircraft’s range that night, 173 miles beyond, but not excessively so. Inspection shows that S40.50 E88.00 was the natural choice to set a course to if you wanted to run out of fuel shortly after local sunrise.
The fuel model allows a range of other flight paths, but none are of these are as as simple and straightforward as the 0.82 LNAV to S38. The other paths are all come with conditions attached. The conditions become more stringent and contrived the further eastward you go.
Now if the fuel model had ruled out the M0.82 FL380/390 straight path to S38, we would all have been to breathe easier. We would have been handed a reason for excluding the pilot in control till the end scenario. Unfortunately that isn’t the. The set of ping ring we have to account for apparently wasnt told to avoid the taboo scenario of a pilot deliberately using his fuel and navigation system as efficiently as possible in order to fly as deeply as possible into the remotest part of the SIO, arriving just in time to take advantage of the early morning sunlight. The ping rings are sized this LNAV path, not for CMT or CTH path or a trip toward McMurdo, for that matter.
Rather inconvenient, that. If this were court of law, which it isn’t of course, the LNAV M0.82 making for S40.50, E88.00 would be enough to secure a conviction of murder, even in the absence of the murder weapon or the accused.
The 00:19 BFOS are no longer such a problem. No longer such a coincidence they coincidence with the SDU logon snapshot. It was no coincidence. The pilot timed a short steep descent to coincide with the logon. Brilliant!
@ALSM
There was no IFE logon at 00:21 as expected. A 60 nm glide would have lasted until at least 00:30 or later. How do you explain away the fact that there was no IFE logon at 00:21? There was more than enough fuel in the APU line to keep the AES alive until well after 00:21.
Great point. I had not heard that articulated before.
@DennisW wrote “An unknown is how much of the $160M was actually paid to Fugro. Certainly not all of it.”
All of it, rounded for convenience.
The value of Fugro’s two contracts with ATSB are published on the AusTender website: AU$13.2m for 2014 bathymetry survey; AU$154,363,087.50 as Prime Contractor for provision of Services for the search for MH370.
The March and April 2014 surface search was funded from a separate ‘pot’, primarily defence resources from Australia, NZ, USA, Malaysia, Korea, Japan, Vietnam, China, UK, & Singapore.
@Don
AU$168M comes to about $127M (USD). I see sightly different numbers bandied about, but $160M seems near the mean.
@DrB: I have good news for you. In the FSX/PMDG777, I found a way to achieve holding speed in CMT mode. If a hold fix is entered in a route in LNAV and VNAV mode, and TRK SEL is selected, when the plane overflies the point of deceleration for the hold, the target speed changes to the holding speed as if still in LNAV mode. (It appears that the plane has to be within a certain distance of the decel point on the path.) The plane would then overfly the hold fix, and the speed would remain as the holding speed (and change with weight).
@Andrew: Do you have any knowledge about this possibility in the real B777?
No IFE logon at 0021. Because it was load shed! RTM!
IFE logon occurred at 1828 because with an inoperative left high gain antenna, line of sight is not possible from the right high gain antenna. Load shedding of IFE was occurring during this extensive time. When electrical loads were restored and IFE was repowered, Satcom still had to wait until the right hga was exposed. This happened at 1825 when the aircraft rolled left for the turn at Nilam to Sanob. The aircraft then turned again at Sanob before overflying Banda Aceh and ending in southern Indian Ocean.
With APU on, aircraft is past Bayesian hotspot.
@Victor: RE: I asked contributors here to recommend where OI should look next after the first 25,000 sq km. I don’t recall that you responded.
By the way, don’t take my previous comment as a personal attack. Just wanted to highlight the issue.
Where Malaysia/OI should look?
I think this is very simple. The search should be systematic. The 25k sqkm is good enough as starting point.
Next, would be the possible locations (including Adaman Sea and Bay of Bengal) based on witness accounts disregarding Inmarsat data entirely, the location indicated by Georesonance and the locations based on the hydroaccoustic study.
Third, would be other locations on the 7th arc, starting where debris fields were identified. The reason it is third is because there in insuficient understanding from the debris analysis on how the plane came to pieces and this analysis compliant with ICAO manual is not complicated. This has been extensively debated and still inconclusive. That search should only start once all the data on the debris analysis becomes available. Also the discovery of more debris could bring additional clues. So we have to park this one for the time being.
It is obvious that pieces of the puzzle are missing or pieces have been introduced (intentionally or unintentionally) misleading the search. As such we have to assume that it may not be possible to find cohesion amongst all data available. Then the search should be systematic, ruling out scenarios one by one. This is the only way and yes budget could be a barrier to this.
@HB
Next, would be the possible locations (including Adaman Sea and Bay of Bengal) based on witness accounts disregarding Inmarsat data entirely, the location indicated by Georesonance and the locations based on the hydroaccoustic study.
Absolutely not realistic. There is no possibility the aircraft is far from the 7th arc.
@ALSM
If the flight path after FMT was intentional, one can logically assume this flight path was also controlled by a live PIC till the end including the end-scenario/impact.
If so he would have also controlled the fuel(consumption), range and destination. He would not have waited till fuel exhaust indeed. He would have controlled all to reach a certain goal/destination.
Possibly with a motivation to crash the plane without fuel. He could even have jettisoned excessive fuel at the end to reach this goal.
I did not mean to state the plane glided ~60 miles. I took this distance as a maximum width for a new search effort to be absolutely sure any possible glide-scenario will not be missed.
Probably ~40 miles on each side of the arc will be enough to reach the same but to be sure I suggest to take ~60 miles as a limit.
As earlier suggested the IFE log-on at 00:21 could have been absent due to the IFE-switch in the cockpit was being switch to OFF somewhere after 18:29. Also the ATSB stated this as a possibility.
@HB: Your proposal is completely unrealistic. The area is too large, it ignores the best evidence we have, and gives the least reliable evidence high priority. And the Georesonance technology is a fraud. It is physically impossible to do what they claim.
@Paul Onions: If you keep repeating what others have explained is not possible, you will be banned.
Ge Rijn:
1. “If the flight path after FMT was intentional, one can logically assume this flight path was also controlled by a live PIC till the end including the end-scenario/impact.” No. Not logical at all. The “intentional premise” is uncertain to start with, but even if true, the data suggests no one was altering the path after the FMT. It is therefor more logical to assume no pilot input after FE.
2. “If so he would have also controlled the fuel(consumption), range and destination. He would not have waited till fuel exhaust indeed. He would have controlled all to reach a certain goal/destination. Possibly with a motivation to crash the plane without fuel. He could even have jettisoned excessive fuel at the end to reach this goal.” This is nothing more than a pile of assumptions based on no facts whatsoever. These statements do not even meet the minimum requirements to call them “conjecture”. It is just nonsense.
“I did not mean to state the plane glided ~60 miles. I took this distance as a maximum width for a new search effort to be absolutely sure any possible glide-scenario will not be missed. Probably ~40 miles on each side of the arc will be enough to reach the same but to be sure I suggest to take ~60 miles as a limit.” The data I cited (facts) suggest that the POI was <10nm from the 7th arc. It would be irresponsible to waste limited search resources on a glide theory while so much of the true 7th arc ±10nm remains to be searched. There is not enough resources to search both.
"As earlier suggested the IFE log-on at 00:21 could have been absent due to the IFE-switch in the cockpit was being switch to OFF somewhere after 18:29. Also the ATSB stated this as a possibility." Why do that? What would be the purpose? Under your set of assumptions, no one was alive in the cabin. So why would a pilot do that?
@ALSM
It would be irresponsible to waste limited search resources on a glide theory while so much of the true 7th arc ±10nm remains to be searched. There is not enough resources to search both.
Totally agree. I remain a fan of giving Metron a contract. I don’t think they are smarter than the rest of us, but I think they would serve the useful purpose of legitimizing the search area. Plus that, they have the pedigree, which I think is important (but I am a bit of a wuss when it comes to spending money).
@Victor, RE: Your proposal is completely unrealistic. The area is too large, it ignores the best evidence we have, and gives the least reliable evidence high priority. And the Georesonance technology is a fraud. It is physically impossible to do what they claim.
Let me further explain the rationale behind this and I don’t think it is unrealistic. Not sure what you mean by ignoring best evidence. All evidence on the table is considered systematically.
Priority 1) 25000 sqkm is already agreed as priority based on best evidence
Priority 2) Areas derived from hydroaccoustic study, witness accounts, georesonance point to discrete points not to the entire ocean
Priority 3) I agree but that is why more clues are needed before searching. I also agree this area could also be further proritised.
Ignoring best evidence (I guess this refers to Inmarsat data). In my opinion, the evidence is not better than others such as witness accounts or debris and it is considered in 1 and 3. Yes, point 2) assumes that the Inmarsat data may not be relied on based on evidence that the Inmarsat CEO raised the possibility of data tampering.
Likewise Point 1 and 3 assume that witness accounts and other potential evidences such as hydroaccoustic study, georesonance, etc. may not be relied on. The approach is scenario unbiased and systematic.
Re Georesonance is a fraud (who says so? on what basis? and has the data been verified? they have their own reputation in O&G sector in APAC). I am not sure whether this has been discussed before but I have just found that there are a number of patents using that technology for the same Exploration purpose. Example of such patents here (http://www.google.com/patents/WO2008127211A2?cl=en).
I can’t post the manual, but this link might help. Load shedding of IFE from power panels P110 and P210 is on the list!
http://pubhtml5.com/uhqv/aanc/basic
@ALSM
You know imo it’s a lot more speculative to assume a clearly pilot controlled flight changes into a ‘ghost-flight’, after an intentional FMT following an intentional flight path, for no apparent reason at all and without any proof also.
The odds for something like that to happen is one in many millions.
Imo the BFO and BTO data cannot serve as proof of a not all pilot controlled flight after FMT.
A reason to switch off the IFE in the cockpit after 18:29 could be the PIC was aware the IFE came on line again when he de-isolated the left main bus. At that time he possibly could not be sure everyone in the cabin was unable to use the IFE anymore.
We have different opinions on this but I have absolutely no problem with searching along 7th arc first within 10nm each side up to 29S if necessary. This would take about ~10.000km2.
So enough to spare to search wider if nothing gets found there.
Especially if OI plans to search beyond the 25.000km2 if nothing gets found.
@Victor
I agree the CMH 180°M via ISBIX can be discounted.
As I stated above: “The RMS BTO Error was high at 10.6 km”.
To be clearer, I should have stated, that the RMS BTO Error is unacceptably high.
A CMH 180°M was just part of the exercise to exhaust all possible flight paths.
@All
Now go to the Movable-Type Scripts online Vincent calculator, scroll down to the Direct Solution box, enter 07 01 25N, 94 36 08E as the start point, bearing 186 48 25.66, distance 5305132.87 metres (2867.3Nm in old money), and see where it goes to. 0.0012 arc sec is inch at S40.50!
Our pilot entered S40.50, E88.00 as a waypoint to fly towards.
Now are you just going to sit there and tell me that’s a fluke?
The DSTG worked out that the 7th arc crossing should be S38.00 E88.30, without knowing about the waypoint, and S1.00 E93.60 for the 2nd arc contact point, while I get S37.88 E88.53 and S1.033 E93.65 respectively. Quite a good agreement between two entirely different approaches to the problem. No bad.
@Ge Rijn wrote “A reason to switch off the IFE in the cockpit after 18:29 …”
AMM Ch 24, Electrical Power – Operation – General states: “The [P5 panel] IFE/pass seats switch disconnects power to some IFE and passenger seat loads.”
The list of IFE components from which powered is removed includes: the Cabin Mgmnt Terminal (CMT), the Video Entertainment Players (VEPs), the Enhanced Video System Ctrl Unit (EVSCU), the Video Modulator Units (VMUs), and the Area Distribution Boxes (ABDs). But not the EPESC, the controller that manages the air-ground messaging function for the IFE system.
From S34º to S29º along the 7th arc is approx 730km, 20nm (+/-10nm) is 37.84km, an area of 27,623km². Do explain your “~10,000km²“.
@All
This is how it went, the plane flies out of radar range at 18:22. The pilot reconnects the LH AC bus and does a little shimmy in the form of a 12nm right offset in order to dissociate himself from N571, then at 18:29 he changes course for ANOKO, on a bearing of 268, staying at FL350 and at M0.84. He has figured in advance that if he waits until he’s 20nm out from ANOKO and then replaces ANOKO with S40.50, E88.00, he will end up just where he wants to when the fuel runs out. The FMT begins at 18:35.67, and just as the plane starts it’s turn, he begins to slow down to M0.67 to confuse anybody watching him on an Indonesian radar screen, or perhaps to gain a few minutes endurance. He waits for about 20 minutes before speeding up to his chosen cruising speed of M0.82 at FL350. He keeps going in the same direction until fuel is about to run out, but just before it does, he momentarily disconnects the AC bus again to get the SDU to reboot. One minute later, he is in a brief, steep descent. 20 minutes later, he hit the ocean at S39.25, E88.15, or thereabouts.
Let’s review the situation: he reached the 7th arc at S38, E88.50, but no trace of him was found on the seabed below. The only possible explanation; he had to have glided well south of the search area.
@All
Correction, chosen cruising speed of M0.82 at FL390, not FL350.
@DrB
RE: “At M0.84 at FL340 at +10C, would the thrust limit reduce the bank angle limit below 25 degrees? If so, can you guess what the bank angle limit would be?”
No, FL340 is well below the optimum altitude for the aircraft’s weight. There should be plenty of thrust available to allow a bank angle of 25°.
@Victor
RE: “Do you have any knowledge about this possibility in the real B777?”
I have not seen that particular scenario in practice, but VNAV does work independently of LNAV. If the aircraft is close to the FMC flight planned track in TRK SEL (or HDG SEL) mode and the LEGS page sequences normally, then VNAV should command a speed reduction at the appropriate point.
@Don Thompson
To me it was understood that the sat-phone from the cockpit works indepent from the IFE-system (explained by @Andrew some time ago) regardless if the IFE was switched to OFF.
But air to ground messages from the cabin would not be possible if the IFE-switch was OFF and an IFE log-on would not occure.
This possibility was also mentioned in an ATSB report.
You say now an IFE log-on would occure anyway, also with IFE-switch set to OFF?
With the ~10.000km2 I was counting from ~32.5S (about where the previous search ended) till 29S which I estimated around 350km and I was thinking in miles not Nm (which was an error I see now).
@Paul Onions: As @Andrew explained, the IFE head is not load shed.
@DrB, @Andrew: Here is an interesting article that discusses what could happen if VNAV is selected without LNAV selected. Although it doesn’t specifically discuss the case in hand, it does show that some unexpected behavior can occur.
@Richard
“A CMH 180°M was just part of the exercise to exhaust all possible flight paths (INSERT: assuming passive pilot after 19:41)” right?
What if Arc6/7 is optionally not included in the model, or is allowed to have greater error bars due to pilot actions after Arc5? That’s how I look at it…not expecting to see Arc6 fit, or rather, I am looking for pilot maneuvers that make Arc6 fit well.
@Victor
That “interesting article” link is not working for me.
@TBill: Try it now.
@Victor. It sounds like your holding speed scenario is analogous to my descent speed and speed restrict with an unimplemented descent. In other words, the speed profile follows what the plane would have done had it followed through on the programmed descent (or in this case, hold). From the reading that I did, VNAV on a 777 is a fickle thing with quite some controversy and discussion among experienced pilots (you’ll see it on the pilots’ forums with a little searching).
@Victor
RE: ”As @Andrew explained, the IFE head is not load shed.”
I now believe my earlier comments were wrong. AMM Chapt 24 – Electrical Power only states that ‘electronic seat equipment’ is loadshed by ELMS. However, the IFE power interfaces diagram in Chapt 23 – Inflight Entertainment System shows that a significant number of IFE head components are loadshed by the P110 and P210 power management panels.
@Victor
RE: ”Here is an interesting article that discusses what could happen if VNAV is selected without LNAV selected. Although it doesn’t specifically discuss the case in hand, it does show that some unexpected behavior can occur.”
I agree with the discussion in the article; it’s not usually appropriate to use VNAV PTH without LNAV, especially during descent. That said, it’s not normally a problem during cruise, although it could cause the behaviour discussed in DrB’s scenario.
“To me it was understood that the sat-phone from the cockpit works indepent from the IFE-system (explained by @Andrew some time ago) regardless if the IFE was switched to OFF.”
Yes: the SATVOICE service to the flight compartment is entirely separate from the cabin services for telecommunications.
“But air to ground messages from the cabin would not be possible if the IFE-switch was OFF”
Yes: the passenger seat equipment is not available. However, core IFE systems remain powered and functional.
“… and an IFE log-on would not occure.”
The EPESC is powered whatever the state of the ‘IFE/Pass’ switch. The EPESC manages IFE air-ground messages, including its BITE and Log On correspondence.
“This possibility was also mentioned in an ATSB report.”
Yes, ATSB listed four possibilities for the absence of the IFE originated correspondence (update to the MH370 search area definition, 3rd Dec 2014, pg 10)
“You say now an IFE log-on would occure anyway, also with IFE-switch set to OFF?”
Yes, and why assume any change in IFE state between 18:25 and end of flight? The IFE/Pass switch has no effect on the EPESC.
S29º to S32.5º = 500km, +/- 18.52km = 18520km²
@Don Thompson
Curious why the ATSB stated this possibility about the IFE back then while you state with no doubt the IFE log-on would occure also if the IFE switch was set to OFF.
I suggested a reason to switch off the IFE after 18:29 in a previous comment. But if an IFE log-on still should occure at 00:21 anyway, this has no value to the end-scenario regarding a possible glide after the steep descent.
I took 68 miles for one degree latitude which makes ~110km.
This makes 385km between 32.5S and 29S. I only took the 10Nm which @ALSM suggested also for miles.. Corrected I come to ~15.000km2.
I think this number can leave the point that I was trying to make to @ALSM standing.
Ge Rijn: When defining the width of a search area, the uncertainty of the 7th arc location and effect of altitude uncertainty must be added in. The 7th arc is only known to ±5.3 nm/99%. https://goo.gl/9e1V7b
The 7th arc location at 35k vs 0 feet differs by ~5 nm. Given the estimated time of FE (00:17:30) and the rate of descent at 00:19:37, it is reasonable to assume the plane was well below 35k feet at 00:19:37, but no one knows how far below. Thus, it is reasonable to add, say, 3-4nm to the arc location uncertainty due to altitude uncertainty. Thus, in order to search a band of ±10 nm from the true 7th arc, the total search width needs to be ~34 nm.
Assuming, for example, an OI search budget of 50,000 km^2 (14,577 nm^2), the budget would cover 429 nm along the true arc ±10 nm. If OI were to include the remote possibility of a steep descent followed by a pull up and max L/D glide, the same budget would only cover about 150 nm of the arc.
@ALSM
Where should the Arc7 centerline be assumed to be – FL350 divided by 2?
I misplaced my calcs but I am thinking the green CSIRO recommended (Fig. 3 prior thead) is about +- 25-30 nm from the the centerline, and the centerline I assume is FL350 Arc7
TBill: As I noted, no one knows the altitude at 00:19:37, but is was certainly well below 35K by 00:19:37. All things considered, 15-20 k ft is a reasonable estimate.
@TBill
You ask “What if Arc6/7 is optionally not included in the model, or is allowed to have greater error bars due to pilot actions after Arc5? That’s how I look at it…not expecting to see Arc6 fit, or rather, I am looking for pilot manoeuvres that make Arc6 fit well.”
In the link below are all the LNAV tracks from 19:41 UTC to 00:11 UTC between 168°T and 185°T.
https://www.dropbox.com/s/wuk3nbom58arac3/LNAV%20GC%20Latitude%20vs%20Track.png?dl=0
As you will see, there are a number of database waypoints and entered waypoints that satisfy the BTO and BFO Error limits.
If you stand the problem on its head and ask what speed schedule (with Pilot input) will reduce the BTO Error to almost zero, then you get the following results:
https://www.dropbox.com/s/o8tcmtdzm3kao2e/LNAV%20GC%20Latitude%20vs%20Track%20Speeds.png?dl=0
The BFO Error still favours a MH370 End Point around 37°S, which has already been searched. However, solutions up to around 30°S are still within the BFO Error limits.
@Paul Smithson
RE: ”From the reading that I did, VNAV on a 777 is a fickle thing with quite some controversy and discussion among experienced pilots (you’ll see it on the pilots’ forums with a little searching).”
I don”t know that ‘fickle’ is the right word to describe VNAV. Like any automated system, VNAV is very predictable if the pilot understands how it works. Unfortunately, some aspects of VNAV behaviour are not well documented and many pilots don”t understand its finer points because they’ve never been taught. The scenarios we have been discussing here tend to compound the ‘mystery’ because they are well outside the normal operation of the aircraft.
@Andrew. Perhaps not the right word but pilots fairly new to the 777 (particularly those coming from Airbus) were sounding off about the unexpected things that VNAV was doing (or not doing). It sounds as if even experienced hands find that 777 VNAV has its oddities. Of course I am quite unqualified to pass any opinion at all – its all hearsay to me!
@Paul Smithson
I agree – VNAV does cause confusion, especially for those new to the Boeing ‘way’. There are a number of differences between the Airbus CLB/DES modes and Boeing’s equivalent VNAV modes. For example, the Airbus autoflight system does not allow the CLB or DES modes to be engaged unless NAV mode (the Airbus equivalent of LNAV) is engaged. That avoids the problems described in the article Victor posted, but it also results in automatic mode changes of which the pilot must remain aware. Believe me, pilots converting from Boeing to Airbus have just as many problems as those going the other way! As I noted previously, those problems are easily overcome with effective training.
@Paul Smithson
I’ll amend the last sentence of my previous post:
“As I noted previously, those problems are easily overcome with effective training, something that is sorely lacking in some airlines.”
@Andrew: I don’t know whether this is feasible, but PC-based simulator tools are perfect for learning things like the intricacies of VNAV. There would have to be some assurance that the fidelity of the simulator at the systems level is better than what enthusiasts use, which is fairly good in general, but not good enough for training.
@Victor
I absolutely agree. Autoflight management is an area that is not taught particularly well in airline training programs and there are plenty of traps for players, young and old. Conversion and recurrent training footprints have been reduced to the bare minimum in an effort to contain costs and they tend to focus on abnormal and emergency procedures at the expense of the ‘normal’ operation. PC-based simulators could be a cost-effective addition to the training ‘toolbox’, but I suspect that validating the software against the aircraft’s actual behaviour would prove to be a stumbling block. The regulators would probably insist on the involvement of the aircraft manufacturers and that would no doubt mean big bucks.
@Andrew
I suspect that validating the software against the aircraft’s actual behaviour would prove to be a stumbling block.
It is huge stumbling block. I would routinely no bid FAA solicitations (without even looking at them in detail) for the reason that they are a complete pain in the ass to work with. My only interaction with government agencies was supplying “standard products”.
All,
I have posted a small note giving my final results from an analysis of Figs. 2 and 3 in the “Flight Path Update” from Oct 2014. As usual, follow the top link at my index:
https://docs.google.com/document/d/14hleZyx1pUPL44yaeHKt6jnSQ3DbgRq2zibbKkFLq2c/edit?pref=2&pli=1
Bottom line – when the ATSB says, “Maximum Range Cruise”, what it realy means is “Maximum Endurance.” (Also as usual, I could be completely wrong.)
Aside: When the ATSB (which I really thing is the DSTG) derived both “data error optimized” and “contrained autopilot optimized” routes in the Oct, 2014 Update, it combined them by finding the intersection of the probability distribution functions. That is curious, because in comparable situations (such as the search for the wreck of the Central America), investigators instead computed a PDF that is the weighted sum of the individual input PDFs. In the case of the Central America, the wreck was found at the edge of the 2-sigma contour defining the search zone, which has been derived from a weighted sum. Had the searchers for the Central America followed the procedures of the DSTG, they would likely have missed the wreck completely. [The fact that the Central America was carrying upwards of 26 tons of gold bullion is irrelevant to this analysis.]
@sk999
Had the searchers for the Central America followed the procedures of the DSTG, they would likely have missed the wreck completely.
Yet another good reason to get professionals involved. A contract to Metron would be my first priority as an official of the ATSB or any government agency involved in this search. Hard to understand why they are screwing around.
@DrB,
Re: “I could only find one published test of GDAS wind speed accuracy. It was a comparison with radiosonde data, and I think it was done in England.”
I suspected that. I have seen some RMS estimates over the Atlantic of 3-5 m/s; long time ago I have also compared GDAS with ground weather stations data. In my experience GDAS often underestimated peak winds, but it was still better than ECMWF. But I don’t see how these exercises can be related to the winds in the Indian Ocean at FL300-350. It is simply wrong to assume similar RMS errors for at least several reasons:
1). Winds at high altitudes are generally much stronger than at the ground, and hence I would expect larger errors.
2). There are a lot more measured data available at the ground, which can be assimilated by GDAS or similar systems, thus reducing RMS errors. This affects overall RMS statistics.
3). The middle of the SIO is not England.
In other words, your approach based on the use of RMS of 5.7 kts is totally wrong.
Re: “I believe the DSTG book references it.”
DSTG book does not reference GDAS. DSTG states: “The wind data used in this study was provided by the Australian Bureau of Meteorology from the ACCESS-G model”.
In addition, DSTG mentions the standard deviation of 5.684 kn of wind components they used in the simulations, but I am not very clear how they derived it. But this number is surprisingly close to yours 5.7 kts. Are you sure you are not confusing estimated GDAS errors with the DSTG study?
Re: “If anyone thinks an acceptable CMH route exists, please post the route specifics so that others may assess it.”
I suspect such a route does exist. As you know the use of GDAS results in remarkably small BTO and BFO errors at the altitude of 4.5 km, corresponding to the magnetic heading 172-174 deg – I have posted it. And the wind is responsible for the good fit. But the endurance criteria is not satisfied at this altitude. I suspect that GDAS might be inaccurate at only 4 km higher level, where fuel would be sufficient to reach the 7th arc.
@Andrew,
Re: “The bank angle limit is 25° at speeds up to 332 KTAS, then decreases linearly with TAS, reducing to 15° at 381 KTAS and constant at 15° thereafter.”
Is it about automatic bank angle or bank angle limit? I was going to ask similar question as @DrB, but I was hesitating in light of your comment with regard to the triangles. Should not the automatic bank angle depend on altitude?
Also, based on FCOM, bank limit can be exceeded by a pilot, correct?
One more question: are you aware of any standard maneuver involving turns of 60 or 120 degrees?
DennisW,
You say, “Yet another good reason to get professionals involved. A contract to Metron would be my first priority …”
Realistically, when you deal with small companies like Metron, there’s usually only one or two people key who really matter. If one of the key people leaves, the company is no longer of use. I presume you have encountered this situation before.
In this case, the key person seems to be Lawrence Stone. Having said that, it is not clear that either he or his coollegues could bring more to the table than has already offered up by the DSTG. (Just my opinion – not a belief).
Finally, let us remember — there are no silver bullets.
@Oleksandr
“Is it about automatic bank angle or bank angle limit? I was going to ask similar question as @DrB, but I was hesitating in light of your comment with regard to the triangles. Should not the automatic bank angle depend on altitude?”
I’m not sure what you mean by ‘automatic bank angle’. When the BANK LIMIT selector on the MCP is set to AUTO, the bank angle limit depends only on TAS.
“Also, based on FCOM, bank limit can be exceeded by a pilot, correct?”
Yes, the pilot can select a higher bank limit via the BANK LIMIT selector.
“…are you aware of any standard maneuver involving turns of 60 or 120 degrees?”
There is a 60° dogleg manoeuvre that is sometimes used to lose time in cases where an aircraft is required to cross a specified point at a specified time and a speed reduction isn’t advisable or can’t produce the required crossing time. The pilot could turn 60° (left or right) away from the track and fly a constant heading for the amount of time that needs to be lost (less a correction for the turn), then turn 120° in the opposite direction and fly a constant heading for the same time interval to intercept the original track.
@Oleksandr
Further to the above, the BANK LIMIT selector only affects the bank angle limit when TRK SEL or HDG SEL is engaged.
@Don Thompson
Are you the author of this article: 9M-MRO Satellite Phone Connections?
It’s very informative.
http://031c074.netsolhost.com/WordPress/wp-content/uploads/2014/09/SDU-HGA-Recovery-Draft.pdf
@Don Thompson
Thanks for confirming that the IFE switch selected off would not prevent IFE logon. For one reason or another, that question hasn’t been settled until now.
@Paul Onions @Don Thompson @others
Interesting article.
I think this is very interesting:
‘An assumption is that Calls A and C were routed to the flight deck audio system and Call B to the cabin crew station. The Call B failure may be linked to the ABC ‘Four Corners’ commentary that the
IFE (i.e., the cabin communications system – CCS) was “interfered with”. The cabin crew satphone would be connected via the CCS whereas the flight deck circuit terminates in the flight deck audio.’
Imo this indicates my assumption of the IFE-switch being deliberately switched to OFF shortly after the re-log-on between 18:29 and 18:39 could be right.
If proven this would hold very important consequences ofcourse.
I would like to hear educated comments on this.
And I like to hear more confirmation on @Don Thompson’s statement an IFE log-on would occure at 00:21 also when the the IFE-switch was set to OFF.
Not that I don’t believe him but such an important statement needs more confirmation as far as I’m concerned.
to add..
I never knew there were actually three sat-phone calls.
If the second call was made to the cabin without response it could mean two things I can think of; the IFE-switch was set to OFF in the cockpit before 18:39 or there was no one consious in the cabin left to answer the call.
@sk999
Metron is not as small as you think they are, and they have been around since 1982.
My motivation for engaging them:
1> They are a private entity not some organization sucking on a government teat.
2> They have the pedigree relative to solving problems of this type.
3> Another opinion would be very interesting and potentially useful.
The group here has some really bright people that have done a lot of great work, and continue to do great work. However, we are not sitting in the same room every day in front of a white board. I don’t think the Metron people as individuals are a cut above many of the folks posting here, it is simply a matter of having access to ALL the information in a focused and directed environment.
ALSM
I guess/hope OI is not restricted how to fill in their 25.000km2 or 50.000km2 ‘budget’.
Any search I guess will start beginning from the 7th arc working outwards but it all depends on OI’s search strategy.
F.i. I can imagine they first start to search the CSIRO/Griffin hotspots around ~35S. Then finish the areas that were not searched from ~34.5 till ~32.5. Then start a search along the 7th arc North including Victor’s hotspot at 29.7S within a +/-15Nm.
If nothing gets found there they will have a lot of ‘budget’ left to search wider and/or at other places.
I mean; 25.000km2 or 50.000km2 don’t have to be straight stretches.
Flexibility is key I think. I hope OI will have the freedom to act this way.
to add..
That’s what hampered the previous search imo.
They never adjusted the search area according new information.
After the flaperon find it soon became increasingly clear the crash area had to be north of 36S. New finds like the Roy-piece only confirmed this.
But nothing was changed.
I hope OI has the flexibility to adjust their strategies according all information that is available and forthcoming.
@Rob
” then at 18:29 he changes course for ANOKO, on a bearing of 268, staying at FL350 and at M0.84. He has figured in advance that if he waits until he’s 20nm out from ANOKO and then replaces ANOKO with S40.50, E88.00″
Rob I am not yet with you all the way to 40S, but you are suggesting something I might see in the simulator cases, and that is the idea of “turn anticipation”.
@TBill
The idea of turn anticipation started with Victor’s with on SIM data; the turn before reaching DOTEN. It was a case of serendipity, pure and simple. I had been having problems fitting ISBIX in with a M0.82 great circle path. Then I got hold of a more accurate position for the 2nd arc, which basically settled it – I had to forget about ISBIX. With the more accurate 2nd arc position and Dr Bobby’s range figures for high altitude M0.82, I was, by simple iteration, able to home in on a rather promising looking path that passed a few miles east of ANOKO. I went back to the other end of it and just as a long shot began to carefully extend it southwards, beyond the 7th arc to see if anything promising shows itself, and then, like wow!, I could hardly believe what was happening, it was slowly converging on S40.50, E88.00. My initial path was not quite on, it was a minute or so out in longitude, but close enough to be tantalising. So I started from the bottom and worked back up. I started from exactly S40.50, E88.00, and found it fitted in perfectly with a turn of 7.5nm radius, beginning 20nm in front of ANOKO. Then I worked it back from the top downwards as a check, and it was an inch out in longitude at S4.50, E88.00. The perfect path. I passes DSTG hotspot by less than 5nm.
For a pilot wanting to leave as little trace on how he arrived at the chosen path, this method would be ideal. If the end waypoint is only a few miles south of anticipated fuel exhaustion, it is a very accurate technique. A few miles out at the FMT end has little effect on the accuracy at the southern end.
@TBill
Just a minor point, the first try was a few seconds out in longitude, not minutes. And why the need for accuracy at the southern end? The obvious driver there is the preference for arriving shortly after sunrise. The big question to be addressed now is how can a steep descent be squared with an extended glide, assuming there was an extended glide. If there wasn’t, then the debris should be near the 7th arc. What are the chances the debris was missed? I am not entirely convinced there was an extended glide.
@Andrew,
Re: “I’m not sure what you mean by ‘automatic bank angle’. When the BANK LIMIT selector on the MCP is set to AUTO, the bank angle limit depends only on TAS.”
Yes, I meant AUTO on the MCP bank limit selector. I have 2 questions:
1). Would the bank limit in the position AUTO be equal to the actual bank angle during a turn, or the actual bank angle is computed based on some other considerations, subject to the selected maximum?
2). Does the bank limit in the MCP AUTO mode depend only on TAS as you described in your response to @DrB? I would expect altitude/air density to be taken into consideration as well.
Re: “There is a 60° dogleg manoeuvre…”
Interesting. My major idea is as before: a wrong exit from some kind of a maneuver, comprised of 60° and 120° turns, may result in the heading of around 172°. Just of note that Prostyakov-Ulich’s ‘feature’ found in the IR NPP Suomi image 18:55, also exhibits a turn of approximately 120°, but I found it incompatible with the radar alerting maneuver I proposed earlier.
DennisW,
After looking a little more into Metron (metsci.com), I will concede your points. On the home page, under News and Announcements, three of the first five articles involve Searches, and two of those reference MH370, even though Metron is not directly involved. Larry Stone, currently Chief Scientist, is in the thick of it and is basically begging to be asked to participate. While I do not see the ATSB bringing them in, perhap OI would be amenable to doing so.
I am not very familiar with Metron or related companies. It looks like Metron was a spin-off from Wagner Associates, and both resemble a third company (actually a non-profit, which I recently received an invite to visit), IDA. I would describe them as “beltway bandits” – living off contracts with DOD, DHS, etc, although Metron seems to have non-government business as well.
@sk999
I think most of Metron’s business does fall into the “beltway bandit” category.
A lot has been written about the power of collaborative (actually at the same location) work. Most of it by psychologists. A fairly recent work, “Scale”, by physicist Jeffrey West has some interesting things to say about it. West pondered if there was a critical size for cities. He postulated that the infrastructure (roads and mass transit) would eventually get to the point where work efficiency was negatively impacted by commute time. What he found is that so far (up to a couple years ago) that a turning point has not been reached. The GDP per person is still increasing with city population size. So even wasting an hour commuting each way to work is more than offset by productivity gain of people working together.
There is also evidence that companies locating new divisions in less dense urban areas did not achieve the productivity gains they thought they would. Productivity actually decreased.
We have the recent moves by Yahoo and IBM eliminating the policy of allowing work from home. They found that the productivity gain associated with requiring employees to come to the “office” to be significant.
Sorry for the long ramble, but I found it all very interesting.
@sk999
It would be great of OI invited Metron to participate in the reward based extended search. Stone would probably take a few weeks to model the idea. 🙂
There are some interesting video interviews on the Metron website here: http://metsci.com/Media-Interviews
@VictorI,
You said: “@DrB: Your long comment proved my point. I said: “On the other hand, automated flight paths of constant magnetic track or heading do not match the satellite data due the curve to the east, which requires a reduction in speed to match the BTO data.” You said, “This is incorrect.” Yet, you have not produced a result which supports your statement.”
Actually I have produced this result for a constant magnetic track last January HERE.
At that time, it was not clear to me how the Holding speed could be obtained when flying a TRK HLD with the NORM/TRUE switch set to NORM. You have subsequently shown this can be done using a PC-based simulator (thank you for investigating this).
I expect that once a deceleration to Holding speed begins in VNAV (about a minute or two before the racetrack pattern begins in LNAV at the Holding fix), that Holding speed is maintained in VNAV even if a track or heading is subsequently selected and held using the MCP, thereby disabling LNAV. What else could The FMS possibly do other than keep the same speed setting?
I have posed specific questions on this possibility below.
The PDA needed for this particular CMT route was about 2% lower than what we now know the average PDA to be, and that estimate was done with a crude fuel model a year ago. I will repeat the route fit with my new fuel model and see what altitude and endurance results.
You also said: “I submit there is no way to do this, as the holding speed for a path to the SIO requires VNAV, LNAV, an offset, and an overflown hold fix, and that would produce constant magnetic heading unless the HDG REF switch was deliberately changed to TRUE.”
You have since potentially demonstrated that there is a way to do this. By the way, I don’t think it is necessary (to produce Holding speed in non-LNAV navigation) to have (a) an offset, nor (b) a route error, nor (c) to “overfly” a Hold fix, nor (d) to set the NORM/SWITCH to TRUE.
Here is what could have happened (expanding on your PC simulator result):
1. A Hold is set up for an upcoming waypoint.
2. There is or is not a route offset (it doesn’t matter in terms of obtaining Holding speed by this method).
3. Several minutes before the Hold fix is reached, VNAV switches to Holding speed and begins to decelerate.
4. Subsequently, TRK SEL is enabled, and 180 degrees is dialed into the MCP. The NORM/TRUE switch is unchanged, being set to NORM.
5. Then TRK HLD is enabled.
6. This could be done BEFORE, or even AFTER, the Holding pattern fix is reached, regardless of whether or not an End of Route error occurs due to an uncancelled route offset.
7. The result is that LNAV is disabled and a CMT route ensues due south (magnetic).
8. VNAV maintains the best Holding speed, which declines with weight, and also with decreasing temperature in the south temperate latitudes. The combination of these two effects is sufficient to match the BTOs with a curved track.
9. In this scenario the direction, but not the destination, intentionally leads to MEFE in the SIO.
@Andrew,
@Mick Gilbert,
1. In a B777 will the scenario described above produce a CMT course with Holding speed?
2. Are there other scenarios (besides the one above and the previous one with the route error due to an uncancelled offset when entering a Hold) that would also produce Holding speed with a course not controlled by LNAV?
@Mr Onions,
Yes, I did author that discussion piece. It is thoroughly out-of-date, only a few days after I drafted it Inmarsat published their JoN paper which included previously unseen RxPwr parameters which clearly showed that the bursts received by the GES were in order.
@Ge Rijn, et al.
The EPESC, Enhanced Pax Entertainment System Controller, is not isolated from power by the ‘IFE-Pass’ switch but ELMS does have the ability to load shed the EPESC. However, concerning ELMS & load shedding, the EPESC was powered at 12:50 prior to the MH370 service when 9M-MRO was operating only on its APU generator.
DennisW,
Good ramble. Regarding the power of being at the same location, you don’t have to convince me. Been there, done that mutiple times, almost always to positive effect.
Actually, not always. One of the innumerable training exercises I was forced to take was intended to demonstrate how people work better as a group rather than individually. We were presented with a scenario, then asked to rank a set of items in order of importance. We first did it indiviually, then came together as a group and repeated. The expectation was that the group would do better than any individual. We uniformly did worse. The instructor was flummoxed. I guess that root cause was that the group members were “highly opinionated.”
I do not expect that to happen within Metron. What if Metron and the DSTG were put togther? It could get interesting!
@sk999
Training exercises are generally viewed negatively by skilled folks. I’ve had several similar experiences. My fantasy of getting into a conference room (white boards and walls draped with relevant stuff) with people here for a few weeks is incredibly appealing. It is about the only thing I miss from the working life. Don’t miss sitting in a coach seat 30 hours a month.
RE: “Would the bank limit in the position AUTO be equal to the actual bank angle during a turn, or the actual bank angle is computed based on some other considerations, subject to the selected maximum?”
The BANK LIMIT selector determines the maximum bank angle that will be commanded by the AFDS during a turn. The actual bank angle will depend on the size of the turn. For example, if the pilot commands a turn through 90°, the AFDS will command a bank angle equal to the bank limit. For a smaller turn, through 20° say, the AFDS will command a lower bank angle. I assume the commanded bank angle is determined by an algorithm that takes into account the roll rate, turn size, etc, up to the bank angle limit.
RE: “Does the bank limit in the MCP AUTO mode depend only on TAS as you described in your response to @DrB? I would expect altitude/air density to be taken into consideration as well.”
According to the manuals, the AUTO bank limit depends only on TAS. I don’t know the logic behind it, but it’s common to other Boeing types. I suspect it’s based on the simple premise that higher TAS is more likely at higher altitudes, where the manoeuvre margins are more limiting.
@GeRijn
@Paul Onions
Following on from Don’s comments above, the IFE is not load shed on the ground when the APU is the only power source, even when the fuel pumps and electric hydraulic pumps are selected to ON/AUTO for engine start. That suggests the APU has the capacity to power the IFE in-flight without load shedding.
@Andrew
‘That suggests the APU has the capacity to power the IFE in-flight without load shedding.’
I did not doubt that actually. My doubts were on the possibilty of an IFE log-on at 00:21 when the IFE switch was set to OFF after 18:29.
I would like to see/hear more confirming information on this issue.
@Andrew
Don Thompson’s article suggests there were actually three sat-phone calls; A, B and C.
The second (B) at 18:39 was to the cabin crew sat-phone which was unanswered too. His article states; ‘The Call B failure may be linked to the ABC ‘Four Corners’ commentary that the IFE (i.e., the cabin communications system – CCS) was “interfered with”.’
Which implies imo the IFE switch was deliberately set to OFF between 18:29 and 18:39 for a complete IFE log-on was established between 18:25 and 18:29.
Only other reason I can think of is the call went unanswered for there was no one left in the cabin able to answer the call at 18:39 anymore.
Unless also this information from his article is out-dated too.
@Oleksandr,
Here are two papers dealing with global data system wind errors:
https://drive.google.com/file/d/1HOeH4KmxPmcXcbNE3iMIAZxMsiJ24v_8/view?usp=sharing
https://drive.google.com/file/d/1XUa3qNvLo4bHDRf72Ilrd1ghFZxh7smP/view?usp=sharing
The first paper shows about 3-4 m/s error (in both U and V) at 400-1000 hPa over the eastern Atlantic, using dropsondes as truth data. The GDAS data were on a 2.5-degree grid every 6 hours.
Figure 2 in the second paper shows 2.5-4.0 m/s RMS errors with radiosonde truth data at 100-900 hPa over China compared to NCEP FNL wind speed (which is based on 1-degree GDAS data every 6 hours).
I expect that actual wind errors from GDAS do vary somewhat with location and with altitude. However, all the sources I can find demonstrate errors in the ballpark of 3 m/s, or roughly 6 knots RMS. The ACCESS-G system used in Australia incorporates the same types of observational data as used in GDAS, and I would expect the ACCESS-G accuracy (stated by the DTSG as 5.7 knots RMS) to be similar to that of GDAS, as it appears to be based on the other papers.
In summary, all the wind error data I can find suggest about 6 knots RMS in U and in V. That is what I use in my route fitting, but it could be anywhere from 5-8 kts.
@DrB
RE: “In a B777 will the scenario described above produce a CMT course with Holding speed?”
Yes. Some additional points:
1. The speed reduction does not occur when the aircraft is flying an offset path, as I found in the simulator earlier this year.
2. The speed reduction is commanded about 30 seconds before the holding fix.
3. I don’t see a need to engage TRK HLD after engaging TRK SEL with 180° selected.
RE: “Are there other scenarios (besides the one above and the previous one with the route error due to an uncancelled offset when entering a Hold) that would also produce Holding speed with a course not controlled by LNAV?”
None that spring to mind. As I said above, in the simulator the speed did not reduce to holding speed when the aircraft was flying an offset path. That seems to leave only one possible scenario for a southerly CMT at holding speed, ie overfly or the holding fix in TRK SEL with 180° selected.
@Ge Rijn, Mr Onions
At 18:39 the initiated two C-channel connections. One was immediately cleared by the GES & the other held for a minute, remained unanswered and cleared by the GES.
At 23:13 a single C-channel connection was initiated.
Subsequently:
Inmarsat’s ‘Search for MH370’ JoN paper provided RxPwr data recorded by the GES during the flight. These data suggested nothing anomalous with the AES transmissions.
The FI listed originating calls to the AES.
The expanded Log, published here, describes the called terminal ID.
The note is out-dated, set it aside.
@DrB
@Andrew
DrB,
Regarding your questions, I’m largely, if not completely, unqualified to answer. However having kicked a few holding scenarios around with Andrew just the other day I will make a couple of observations. If VNAV is engaged and the autothrottles are armed then commanding a hold at Present Position (PPOS) should set Best Holding Speed as the target speed immediately. So long as no contrary speed commands are issued (eg via an MCP speed intervention) the FMC should then stick to Best Holding Speed. Thus, having commanded a hold at PPOS, if you make a heading or track command via the MCP then you’ll have a CMH or CMT path at Best Holding Speed.
Regarding your second point, a question for Andrew;
If you have an active route in LNAV and you command a hold at PPOS, does that generate a Route Discontinuity between PPOS and the next active waypoint? If so, what happens when you exit the hold? Might that be another way to end up on a CMH at Best Holding Speed?
@Mick Gilbert
@DrB
RE: “If you have an active route in LNAV and you command a hold at PPOS, does that generate a Route Discontinuity between PPOS and the next active waypoint? If so, what happens when you exit the hold? Might that be another way to end up on a CMH at Best Holding Speed?”
Yes, the FMC does create a discontinuity between the PPOS hold and the next waypoint. If you exit the hold and take no other action, the aircraft will maintain a CMH in LNAV when it reaches the discontinuity. If I’m not mistaken, the speed will increase to the ECON CRZ speed.
@Mick Gilbert
The reason the speed increases in the above scenario is because you would need to activate EXIT HOLD to leave the hold in LNAV. The speed then reverts to the ECON CRZ speed as the aircraft leaves the hold. In DrB’s scenario, the aircraft leaves the hold in TRK SEL/TRK HLD, but the FMC still thinks it’s holding because EXIT HOLD was not activated.
@Ge Rijn,
I will update/rewrite an analysis of the SATVOICE connections & post a link here.
Concerning the datalink traffic initiated by the ‘IFE’, they are initiated by the EPESC. If the EPESC is powered operating, it initiates the virtual connections over the datalink. I will reiterate that the ‘IFE-Pass’ switch does not have any effect on supply of power to the EPESC.
@Andrew
Thank you for the clarification regarding the different target speeds arising from the different scenarios.
Now that we agree, that it’s possible that the IFE logon at 00:21 did not occur because of load shedding when on APU only, let’s turn our attention to why it logged on at 18:28.
For IFE to logon at 18:28, it would require a re-establishment of power after any load shedding, and a line of sight connection between the satellite and the aircraft’s antenna (Left or Right High Gain Antenna).
MH370 was equipped with dual High Gain Antennas (HGA), one mounted flush on the left side and one mounted on the right side of the fuselage.
http://031c074.netsolhost.com/WordPress/wp-content/uploads/2014/09/SDU-HGA-Recovery-Draft.pdf
Referring to the antenna coverage chart on p4 of Don’s paper, it shows the reception coverage relative to the aircraft. The outer ring represents the relative horizontal bearing from the aircraft’s nose in degrees. It is not compass direction. Thus 0 is the nose of the aircraft, 90 is the right wing, 180 is the tail, and 270 is the left wing. The inner circles, labelled on the x-axis, represent the angle below the aircraft’s vertical axis. Thus 0 is straight up, and 90 is toward the wingtip.
As MH370 flew from Mekar to Nilam, it was tracking 296 degrees True. The compass direction and elevation to the satellite at this time was, 262 degrees True at 52 degrees elevation (above horizon). This equates to a relative bearing of 326 degrees (clockwise from the nose of the aircraft) and 38 degrees below the aircraft’s vertical axis. Plot this on the chart in Don’s paper, and it’s slightly closer to the centre than point 3, well within the left HGA coverage area.
But what if the Left HGA is inoperative and the Right HGA is serviceable? Let’s look at the coverage. To look at right side coverage, continue the arc from relative bearing 45 to 120 around the centre to complete the circle. Thus, on a relative bearing of 326 the Right HGA can only see about 20 degrees past the vertical on the left side of the aircraft. In other words, it’s 18 degrees short of seeing the satellite. Thus, Satcom and IFE could be ready to log on again following load shedding and restoration, but can’t because there is no line of sight reception with the satellite.
If the aircraft rolls 18 degrees left, to start a left turn at Nilam to Sanob, then the aircraft’s vertical axis is tilted 18 degrees towards the satellite, and BINGO, the right HGA now has line of sight and satcom logs on at 18:25. At that moment the aircraft is still tracking towards Nilam. The cold start logon continues with BFO overshoot, as discovered by Dr Ulich. At 18:28 the aircraft turned left again at Sanob and over flies Banda Aceh, since the crew had programmed the autopilot for a diversion to Banda Aceh airport before passing out. The crew weren’t alive when the Satcom logged on at 18:25. And they weren’t alive at 18:40 to answer the satellite phone call. The aircraft continued south at ECON speed at 35000 feet with its APU on, until the aircraft ran out of fuel. Our inflight experiment showed that at an end of route, LNAV remains engaged but captures Heading True. Because the APU was already on, the autopilot remains engaged even after both engines have flamed out. The APU is using residual fuel. Our sim experiment has the Satcom momentarily depowered long enough to reboot when the Left engine spools down and the APU (which was already on) repowers the Left Main AC bus. Normal switching with left engine operating does not lead to reboot. With the autopilot still engaged and both engines failed, the aircraft porpoises further south beyond 40 nautical miles past the Bayesian Hotspot. IFE is load shed. When the APU stops, the aircraft ends in a dive.
To confirm our simulator experiment; Left engine failure leads to reboot of satcom, can anyone get access to the satellite communications logs for an Emirates B777, registered A6-EGA, that manually shutdown the left engine at 10:21UTC on 7Sep2016.
https://www.atsb.gov.au/media/5772230/ao-2016-113-final.pdf
Apologies to Victor for the lengthy post.
@Paul Onions: Please answer the following questions, some of which have been posed now several times.
1) You claim there are two HGAs. That is false. There is a single HGA with separate left and right antenna arrays.
What failure could possibly lead to one array inoperable and the other operable?
2) You believe that the SATCOM would re-boot when the power to the left bus switches from the left IDG to the APU. The SATCOM is specifically designed to “ride-through” a momentary loss of power. You are relying upon the PMDG777 model to accurately characterize this sub-second switching transient, which is an unrealistic expectation.
Do you have any evidence that the SATCOM manual is wrong, and there is insufficient bus capacitance to ride-through the voltage dip when the power to the left bus is switched?
3) You claim the plane would follow a path of constant true heading after a route discontinuity. Unless TRUE rather than NORM was selected via the HDG REF switch, the path would be a constant magnetic heading.
Why do you believe a pilot switched the HDG REF from NORM to TRUE?
4) You claim the APU remained operating after the second engine flamed out, and the aircraft glided for 40+ NM. @Andrew and @Don have determined that the power to the IFE head is not load-shed when the APU is the only power source for the AC busses.
Why was there no log-on to the IFE server after 00:19?
@Victor
You claim the APU remained operating after the second engine flamed out, and the aircraft glided for 40+ NM. @Andrew and @Don have determined that the power to the IFE head is not load-shed when the APU is the only power source for the AC busses.
Thanks for that. I’ve been following along, but not contributing since I know next to nothing about the aircraft or aircraft systems. I was sure I read that same conclusion. I thought I was having a brain issue.
@Don Thompson
Thanks for clarifying. I look forward for your re-written articleon on this issue.
Re VI’s statement at 9:58 above:
“…and there is insufficient bus capacitance to ride-through the voltage dip when the power to the left bus is switched”
To clarify, all the certified avionics in aircraft (see DO-254/DO-178/DO-160) are required to have the means to continue to operate through short power source interruptions, such as a transfer from Ground Power to APU, or ME to APU. The “bus” itself does not provide this function. Each “box” on the bus provides its own “capacitance”. In fact, it is literally capacitors in the AC to DC power supplies within the AES that provide the function. When AC power is removed, the AES power supplies DC outputs do not fall immediately. It can take several seconds before all the regulated DC outputs start to drop. This is designed in for several reasons, not just to assure continuous operation during a brief power source change. For example, it also gives any device with logic/memory time to shut down gracefully when power is switched off.
@ALSM: Yes, when I was referring to bus capacitance, I meant the internal DC bus of the SATCOM after voltage rectification. The capacitance is sized so that the voltage dips by an allowable amount as current continues to flow from the capacitors to the device.
@DrB: You are over-simplifying what occurs for VNAV without LNAV. What occurs is related to how the active leg changes when not in LNAV. For instance, if the decel point to a hold is reached and then the plane is taken off the route with HDG/TRK SEL so that the holding fix is not reached, the speed again increases to ECON after some distance. I don’t have the time to work out all the cases, but I believe the speed will not remain at holding speed unless certain conditions are met.
@VictorI,
You said: “@DrB: You are over-simplifying what occurs for VNAV without LNAV. What occurs is related to how the active leg changes when not in LNAV. For instance, if the decel point to a hold is reached and then the plane is taken off the route with HDG/TRK SEL so that the holding fix is not reached, the speed again increases to ECON after some distance. I don’t have the time to work out all the cases, but I believe the speed will not remain at holding speed unless certain conditions are met.”
Is this what the PC simulator does, and are you sure the B777 also increases the speed to ECON cruise after a TRK HLD is commanded before the Hold fix is reached? Are you aware of any Boeing documentation of this case?
If you are correct, then if TRK HLD is executed AFTER reaching the Hold fix, the Holding speed would be maintained afterward and would not revert to ECON cruise, right?
@Andrew,
Thanks for answering my questions. I was confused by one comment you made: “I don’t see a need to engage TRK HLD after engaging TRK SEL with 180° selected.”
Perhaps I am misunderstanding how the MCP works. I thought to fly a track or heading you did the following:
1. Set MCP toggle switch to Track or Heading.
2. Push the Select knob in.
3. Rotate the Select knob so that the displayed number of degrees is what you desire.
4. Push the square button below the Select knob, and that button label changes from SEL to HLD, and this action causes the AFDS to match the commanded course.
Is this correct?
@Mick Gilbert,
Thanks to you also for your suggestion.
@VictorI,
@Andrew,
Let me see if I have got it right: (1) If a Hold at an upcoming fix is executed with no offset present, and AFTER the Hold fix is reached a track or heading is manually commanded using the MCP, the MCP course will be executed and the speed will remain at Holding in VNAV. (2) If the track/heading is executed after the deceleration begins but BEFORE the Hold fix is reached, the airspeed will initially remain at Holding but will subsequently revert to ECON cruise after some time interval.
Do you agree with these statements? Of course, what I really want to know is how one could end up in a CMT route with Holding speed, and is there more than one way to do this?
I don’t quite follow the discussion of the Hold at Present Position case, except that it can lead to a route discontinuity when exiting the Hold, and this most likely leads to a CMH route which does not appear to fit the satellite data.
@DrB: First, HLD and SEL are two different modes. In SEL, a turn is executed to align and then hold the direction in accordance to the displayed value. When HLD is selected, the current direction is maintained, independent of what is displayed. SEL is entered when the SEL knob is pressed. HLD is entered by pressing the HLD button. (SEL mode is a little dangerous because once selected, the plane will start to turn to the displayed direction. I believe the FO is assigned to keep the displayed direction close to the actual direction so that when SEL is entered, there are no surprises.)
Regarding maintaining hold speed while not in hold, that will occur if the decel point and hold fix are overflown in VNAV, regardless of when LNAV mode is exited. As I said, it is complicated because the active leg is advancing without being in LNAV, and there must be some criteria for that to occur, for instance, a maximum deviation from the path. I don’t have the time to explore all the nuances, nor can the PMDG777 model be 100% relied upon. Maybe @Andrew knows off hand.
@Andrew,
I found a (poor) description of the MCP controls on Smart Cockpit. I was missing the fact that there are actually two separate switches for Track/Heading – one toggling pushbutton for HLD at the top of the MCP and one (the middle rotary one) for SEL. Now I understand that pushing the SEL button causes the roll mode controls to go to the displayed number of degrees. So you don’t need to use the HLD switch at all, although you could do so without changing the roll mode. Now I understand your comment.
@DrB: one toggling pushbutton for HLD at the top of the MCP and one (the middle rotary one) for SEL.
The HOLD button is below the SEL knob, not above. The HDG/TRK button is above the SEL knob.
So you don’t need to use the HLD switch at all, although you could do so without changing the roll mode
If not in HOLD mode, for instance, in SEL or LNAV, then pressing the HOLD button changes the roll mode to HOLD. In HOLD mode, the direction displayed in the window has no effect. They are distinct modes.
@DrB,
Re: “The first paper shows about 3-4 m/s”.
It actually states 3-5 m/s, i.e. up to 9.7 knots.
Re: “Figure 2 in the second paper shows 2.5-4.0 m/s RMS errors with radiosonde truth data at 100-900 hPa over China compared to NCEP FNL wind speed (which is based on 1-degree GDAS data every 6 hours).”
Note wind speeds in this paper. I think for our purpose it would be more appropriate to use relative RMS errors in the wind speed, rather than absolute values. Given that the magnitude of U-component exceed 20 m/s at FL300 and above, I would expect 2-2.5 times larger errors than those you cited, i.e. up to 15 knots or so. It is a lot bearing in mind that U-component affects BTO errors in the CMH or CTH modes.
Re: “However, all the sources I can find demonstrate errors in the ballpark of 3 m/s, or roughly 6 knots RMS.”
The fist slide you shared states that RMS errors were in the range 3-5 m/s, so your conclusion is not true. Also, the use of relative RMS would probably be more appropriate compared to the absolute RMS. You cannot expect the same absolute errors at 7 m/s and 20 m/s wind speeds.
Re: “The ACCESS-G system used in Australia incorporates the same types of observational data as used in GDAS, and I would expect the ACCESS-G accuracy (stated by the DTSG as 5.7 knots RMS) to be similar to that of GDAS, as it appears to be based on the other papers.”
I don’t know ACCESS-G. I don’t know its performance. But how can you state that GDAS performance is similar to ACCESS-G? Do you know what data are assimilated in both of them?
As I mentioned before, I am not very clear how the DSTG derived 5.7 knots. What data did they use: where, what range, when? What other papers do you mean? The only two cited by you?
Re: “That is what I use in my route fitting, but it could be anywhere from 5-8 kts.”
I doubt that the use of 5.7 knots is a correct approach. I actually don’t know what the correct approach should be for the CMH or CTH modes bearing in mind that BTO errors are sensitive to U-component in these modes.
@DrB
Yes, as you discovered (and Victor also described), TRK HLD (or HDG HLD) and TRK SEL (or HDG SEL) are distinct modes. When the ‘HLD’ modes are engaged by pressing the HLD switch, the aircraft maintains the current TRK or HDG, regardless of the heading or track displayed in the heading/track window. When the ‘SEL’ modes are engaged by pressing the SEL switch, the aircraft turns (if necessary) and maintains the heading or track selected in the heading/track window. The HDG/TRK reference switch at the top of the MCP changes the reference between HDG and TRK.
@Victor
RE: “Regarding maintaining hold speed while not in hold, that will occur if the decel point and hold fix are overflown in VNAV, regardless of when LNAV mode is exited. As I said, it is complicated because the active leg is advancing without being in LNAV, and there must be some criteria for that to occur, for instance, a maximum deviation from the path.”
The maximum distance from the lateral path at which waypoints will automatically sequence is 21 nm. According to the FCOM:
The FMC will not sequence the active waypoint when more than 21 nm off the active route and not on an offset route.
I suspect that distance may be less in the case of deceleration points that are not displayed on the LEGS page, but it’s not specified in the FCOM. I think it’s safe to say that VNAV will command a speed reduction to best hold speed if the aircraft passes within a ‘few’ miles of the holding fix, regardless of the lateral mode that is engaged.
Ocean Infinity’s Seabed Constructor with its AUVs has just departed Port of Spain for Durban. If this is MH370 related, presumably it will recrew there and move directly to the search area for a full search swing. That’s 20 days sailing at 13kn to Durban and 10 more to the search area.
@VictorI
A point I still like to adress in this topic is your motivation to explain the CSIRO argument of debris above 32.6S:
‘For latitudes north of 32.6S, debris should have arrived in Madagascar, Mozambique, and other locations in Eastern Africa, before the discovery of debris was reported. But this could easily be explained by the delay between when debris arrived near a location, and when it is discovered and reported to authorities.’
First; any debris arrived ofcourse before the debris was discovered and reported.
Second; various drift studies show that debris form the region north of 32.6S would arrive at Reunion and Madagaskar up to a year earlier than the first (flaperon) find.
I think this delay is not easily explained.
I can imagine on average two months or so between arrival and discovery but not a whole year or more. And since beginning of 2017 nothing is reported anymore. This should be telling also.
Imo this still is the big problem with the latitudes north of 32.6S.
And the farther north the more problematic this becomes.
@Ge Rijn said: various drift studies show that debris form the region north of 32.6S would arrive at Reunion and Madagaskar up to a year earlier than the first (flaperon) find.
When I look at the debris paths using the CSIRO model for “flaperon-type” debris originating at 30S, the debris begins to arrive at Reunion longitudes in October 2014, but the vast majority arrive much later. I don’t see how you can use the single discovery of the flaperon in July 2015 to say the arrival was too late to consider an impact at 30S latitude. On the other hand, I believe the discovery of “Roy” in December 2015 is a much better discriminator, as it suggests that debris likely arrived along the shores of more northern parts of Eastern Africa well before “Roy” and “No Step” were discovered.
Not to mention that debris consistent with an impact at 30S might have been seen in March 2014. From all that we know, I think the area around 30S qualifies at a “hot spot” to consider in the new search.
@Ge Rijn,
You fail to understand that MH370 debis include not only 9M-MRO fragments. Also you seem to be unable to grasp a difference between statistical modelling and deterministic modelling. Finally, as I and others unsuccessfully tried to explain to you: early arrival is not a problem due to the lack of awareness, remoteness or complicated accessibility of shores, failure to report suspected objects; only late arrival is indicative. A proper search has never took place.
@Richard Cole: Thank you for the update. Things seem to be aligning for a search starting in January 2018.
@All: I will ask one last time. Now that we’ve had a opportunity to discuss the new search, and we’ve had the opportunity to consider more information (the images and drift data), please consider responding to this question:
If the search of the initial 25,000 sq km is unsuccessful in finding the wreckage, where should OI search next? Please note that is not necessary to re-affirm support for an impact site within the initial 25,000 sq km area.
@Victor
I remain in favor of the 28S to 30S area on the 7th arc for the reasons I stated answering this question previously. I would not search at all in the 25,000 km^2 area now viewed as most likely by the ATSB/CSIRO. I think the debris photos taken from the search aircraft are more convincing than the satellite photos.
My understanding is that no debris was ever recovered and examined. Is this still a true statement?
@Oleksandr @VictorI
Your own ‘statistical’ drift-studies and animations show arrival on Madagaskar shores as soon as july 2014 and many hundreds- even thousands of particles beached on Madagaskar from the ~30S region by november 2014.
If you mean this is a difference between ‘statistical modelling’ and ‘deterministic modeling’ (whatever is the difference..) than your kind of drift modelling has no value imo.
Then you keep a blind eye to your own results and tell others they don’t grasp it. Which is the easy way out.
I ask you to take a look at this issue with an open mind.
THe Roy-piece can be explained by the much faster currents around the tip of southern Madagaskar along the Mozambique and S.A. shores.
Delays of one year or more between beaching and finding on African shores and islands is just unrealistic imo. More than a billion people live there and many in coastal regions.
The debris reported till now is found in a time window of ~1 1/2 years. Nothing has been reported since.
To me this is also indicative of an arrival window of at max one year after the flaperon find.
Please think of it.
@DennisW: I am not aware of any recovered debris that was photographed in the air search.
@Ge Rijn: Please consider proposing a search area beyond the 25,000 sq km based on what we know and what you believe is likely.
to add..
The flaperon and Roy-piece were full of barnacles when first found.
Being the pieces found closest and most far from the 7th arc they on itself define an arrival window of most other pieces imo (without barnacles).
They were found soon after arrival.
@Victor,
Re: “If the search of the initial 25,000 sq km is unsuccessful in finding the wreckage, where should OI search next?”
My recommendations remain as before.
1. I would not recommend searching in the “initial area”. I would tag it as the least priority area.
2. As @Dennis, given all the information we have by now, I favor 28-30S.
3. Intersection of the 7th arc with the acoustic signal recorded by the Curtin University.
However, I would suggest starting from #3 because it is a relatively small area. I think 50 km x 40 km area would be sufficient to address the uncertainties in the input acoustic data and the final BTO.
@Ge Rijn,
“your kind of drift modelling has no value imo.”
I am done on commenting your silly nonsense. Sorry, my patience ran out.
@VictorI
I will and I’ll take some time.
@Oleksandr
That’s a pity you don’t even consider my arguments. You still stick to your ‘Curtin boom’ as it seems. Very narrow imo. A waste regarding your obvious statistical talents.
But it’s all in the pursuit of finding a solution I know.
@Oleksandr: Please provide the coordinates for the crossing of the acoustic event so it is not subject to my interpretation.
@Oleksandr
You replied to me;
“your kind of drift modelling has no value imo.”
You know that’s out of the context I said this.
I really value your drift-modeling. That was not the point.
Please keep arguments and statements in the context it was meant for.
Re: the HA01 acoustic event
From “Rewinding the waves…”, The signal observed at HA01 provided an estimated location E1: Bearing=301.4 ± 0.4°, range=1900 ± 200 km from HA01, centred at −23.662°, 96.676°, recorded at 01:34:40 UTC (eventsource between 01:11 and 01:16). Curtin observed the same signal, but did not use energy exchange between gravity and acoustic waves to estimate the distance from HA01. Instead, Curtin used a weak signal from RCS to estimate the distance from HA01 (4961.6 km). However, there were large error bars on the Curtin distance estimate. At the extreme SE end of the Curtin uncertainty region (Figure 11), the range is near the same range estimated by Kadri et. al. The bearing was essentially identical (301.6° +/- 0.75°.)
If we assume the signal was associated with the MH370 impact and the bearing is accurate, but the range is uncertain, then it is worth considering the intersection of the 7th arc and the 301.5 degree bearing as a possible hot spot to search, particularly if it aligns with any of the simulator derived theories, etc.
The location of this point is: 26°46’46.46″S, 100°42’37.01″E (1565 km from HA01) Note that Kadri’s estimated range at the low end (1700 km) is within 135 km of this point.
I forgot to point out that the “eventsource between 01:11 and 01:16 time” does not match the approximately 00:20 estimated time of impact.
Could it take an hour for the hull to impact the sea floor?
If we assume the signal was associated with the MH370 impact and the bearing is accurate, but the range is uncertain, then it is worth considering the intersection of the 7th arc and the 301.5 degree bearing as a possible hot spot to search, particularly if it aligns with any of the simulator derived theories, etc.
The McMurdo Station path falls within the bearing range of the acoustic event.
@ALSM @Oleksandr @VictorI
We have been through this before in lenght long time ago.
In the many hundreds (if not thousands) of seismic events recorded each year the impact of an airplane hull with sinking speed on the ocean floor won’t be registered let alone be noticed as something outstanding or significant.
By the way a high speed impact then would be excluded for the hull would have been shattered in pieces. Leaving no impact on the ocean floor at all.
It’s just picking your choice of the many seismic (very small) events that also occured around 8-3-2014 just like they did throughout the whole year before and after, and still do.
Just wishfull reasoning without any substance imo.
@Victor you said “@All: I will ask one last time. Now that we’ve had a opportunity to discuss the new search, and we’ve had the opportunity to consider more information (the images and drift data), please consider responding to this question”
Victor, I’m sure you won’t mind if I pass on this one. I always was party pooper, anyway.
@VictorI
You asked:”If the search of the initial 25,000 sq km is unsuccessful in finding the wreckage, where should OI search next? Please note that is not necessary to re-affirm support for an impact site within the initial 25,000 sq km area.”
The fact that objects or even potential debris fields have been observed from air but were not collected and identified imo undermines the claim that the surface search in certain areas has been effective. As a consequence, results of drift studies without the restrictions based on surface search “results” should be leading in defining a new search area. I’m in the process of studying the different drift models and results (for example I have asked prof. Pattiaratchi more info on the UWA models and calculations). From what I have seen so far regarding the CSIRO work: It seems to indicate that the 36 – 32.6 area should be extended north from 32.5 to roughly 30 degrees south. This would include the UWA “hotspot” mentioned earlier this year.
@Niels: Suppose the debris seen on March 29 was the only debris seen. (I don’t know this to be true.) Would that change your thinking?
@VictorI
We have the Pleiades images (March 23rd), the potential debris field mentioned in the Bernard Lagan artcle (March 28th), and the images you have posted (March 29th). I find it really hard to know on which observation to put (more) weight.
@Victor,
Re: “Please provide the coordinates for the crossing of the acoustic event so it is not subject to my interpretation.”
My approximate coordinates are 100.1E, 27.5S. Approximate – because I digitized Duncan’s plot to derive delay, and also RCS coordinates from the publicly available plots. I have not considered bearing sourced from HA01. The uncertainty is also subjected to the assumed temperature, but generally it appears to be within 0.2 deg (each side from the indicated point).
@Niels: So we’re all on the same page, which images from March 28 do you believe show a debris field? I am aware of the “blue panel” and fishing gear.
@ALSM,
Te: “I forgot to point out that the “eventsource between 01:11 and 01:16 time” does not match the approximately 00:20 estimated time of impact.”
It matches. Consider seabed impact.
Re: “Instead, Curtin used a weak signal from RCS to estimate the distance from HA01 (4961.6 km).”
This is false. Signals at HA01 and RCS were of similar magnitudes; slightly lower at HA01, as could be expected due to the distance. Later Curtin added one more station, Scott Reef, and estimated location using triangulation. They did not estimate the distance of 4961.6 km as you described.
Re: “Could it take an hour for the hull to impact the sea floor?”
We had discussed this ad nauseam, as you like to say.
Why not? Take into account entrapped water. Also consider the time it took the water to fill cavities after the impact, before the hull started to sink. I think 1 hour is very reasonable timing.
@ALSM
“Could it take an hour for the hull to impact the sea floor?”
That’s a potentially important observation.
I gather from the ARA sub search that YES they think the sub could have drifted quite a distance before hitting bottom. Also don’t forget there are other pressure vessels on the aircraft that could implode (escapes my mind at the moment what they are).
I suppose the whole MH370 could have drifted away from Arc7, if we are unlucky.
@TBill,
What is a potentially important observation? I have just checked JW website – we have been discussing seabed impact and its possible relevance to the Curtin boom for 2 years.
@Oleksandr,
You said: “I don’t know ACCESS-G. I don’t know its performance. But how can you state that GDAS performance is similar to ACCESS-G? Do you know what data are assimilated in both of them?”
To quote from the DSTG-referenced ACCESS paper available HERE :
“Observations currently used by ACCESS include:
Land surface stations, ships and drifting buoys: surface pressure, 2m temperature, 10m
winds and 2m relative humidity.
Radiosondes: temperature, wind and relative humidity
Pilot and profiler winds
ATOVS radiances (HIRS, AMSU-A and AMSU-B)
AIRS radiances (62 channels)
Geostationary AMV winds including locally calculated 6 hourly cloud drift winds from
MTSAT-2
Scatterometer winds (ASCAT)
Aircraft (AMDAR) winds and temperatures”
These are the same types of observational data used in GDAS (you can look those up if you want). That’s why I said the errors are likely to be similar.
@Andrew,
@VictorI,
Thanks for the clarifications. One last (?) point: Andrew said: “I think it’s safe to say that VNAV will command a speed reduction to best hold speed if the aircraft passes within a ‘few’ miles of the holding fix, regardless of the lateral mode that is engaged.”
Please correct me if I am wrong, but in a normal Hold situation the deceleration would begin a minute or two BEFORE the Hold fix is reached. That means that the deceleration to Holding speed would begin maybe 8-15 NM before the Hold fix is reached. I think you (Andrew) are also saying that the Holding speed would not REVERT to ECON Cruise (and VNAV would stay in Holding speed) so long as the aircraft “passes within a few miles of the holding fix”, even if a different lateral navigation mode (such as TRK SEL or TRK HLD) were begun BEFORE the Hold fix was reached. Am I understanding you correctly?
To summarize, if a Hold were programmed and executed, the commanded speed would begin to slow to best Holding BEFORE the Hold fix was reached. If a TRK SEL or a TRK HLD were subsequently commanded after the deceleration began (indicating that current best Holding is the new speed target), the Holding speed would be maintained (and it would not revert to ECON Cruise) so long as the aircraft passed within a few miles of the Hold fix, even if the TRK SEL / TRK HLD began BEFORE the Hold fix was reached. If TRK SEL/TRK HLD began AFTER the Hold fix were passed, Holding speed would also be maintained in VNAV. Did I get it right this time?
Oleksandr,
You ask, “I am not very clear how the DSTG derived 5.7 knots. What data did they use: where, what range, when?”
As is made clear in Bayesian Methods, it used the wind and temperature data from the ACARS position reports. We have those numbers available for the MH371 flight of Mar 7, 2014. A straightforward analysis shows that the rms differences between the ACARS data and the GDAS model are 4.5 knots for the E-W component, and 4.6 knots for the N/S component. Biases are small (-1.1 and -0.3 knots respectively). The number of data points is 71.
In the citation that Bobby Ulich made to the ACCESS-G data sources, there is reference to “Aircraft (AMDAR) winds and temperatures”. These are similar data from a large number of aircraft around the world that are ingested into global weather models. Here is a reference that explains how the system works in more detail:
https://amdar.noaa.gov/FAQ.html
You also state, “Note wind speeds in this paper. I think for our purpose it would be more appropriate to use relative RMS errors in the wind speed, rather than absolute values. Given that the magnitude of U-component exceed 20 m/s at FL300 and above, I would expect 2-2.5 times larger errors than those you cited, i.e. up to 15 knots or so.”
I think what you are saying is that one would expect that, as the wind velocity increases, the errors should do so as well. That was certainly my expectation, but it is not born out in practice, either by the MH371 weather data or in a previous study I carried out 2 years ago comparing GDAS with ECMWF; the rms errors are largely independent of absolute wind speed. I also found no particular bias in the mean velocties, so relative and absolute RMS errors are going to be comparable.
@sk999,
Re: “As is made clear in Bayesian Methods, it used the wind and temperature data from the ACARS position reports.”
From DSTG book: “Parameters of the model were determined using the process described in section 6.1.1 based on logged aircraft flight data (i.e., using the predicted wind value as the nominal
set point).”
I am not clear what flight(s). Same questions as before: where, what altitudes when, range, how many data points?
I also did comparisons of data from GDAS and ECMWF, but about 4 years ago. GDAS was always a bit more accurate, particularly with regard to peak wind speeds. But in my experience, there was a strong dependence of errors on magnitudes, in contrast to your experience. My comparisons were all at ground level; I can’t say anything about FL350.
@Richard Cole. Seabed Constructor ETA Durban is listed now as 27th Dec 1000 local time, so proportionately 9.6 knots.
Economical speed moderate seas is 10 knots: page 4/8 of downloaded brochure from:
http://www.swireseabed.com/vessels/sbc
Speed made good of course depends on current.
A couple of days there. Your 10 days direct transit to start would extend to 13 at 9.6 knots. Start at 11 Jan therefore at the earliest. The season ticks away.
They must be confident or else will search regardless.
….will start search regardless.
@DrB,
It is obvious that various models use the same kind of instruments, e.g. surface stations, buoys, etc. to obtain measured data. It is not obvious whether GDAS uses the same data as ACCESS-G to generate gridded data. Especially in the remote SIO.
@DrB
RE: “Please correct me if I am wrong, but in a normal Hold situation the deceleration would begin a minute or two BEFORE the Hold fix is reached. That means that the deceleration to Holding speed would begin maybe 8-15 NM before the Hold fix is reached.”
In my experience, the speed reduction commences much closer to the holding fix. As I noted in my earlier reply, the speed reduction typically commences about 30 seconds before the fix, depending on the difference between the descent speed and the hold speed, and whether the aircraft is in level flight or descending. The FMC aims to arrive at the holding fix at the holding speed.
“I think you (Andrew) are also saying that the Holding speed would not REVERT to ECON Cruise (and VNAV would stay in Holding speed) so long as the aircraft “passes within a few miles of the holding fix”, even if a different lateral navigation mode (such as TRK SEL or TRK HLD) were begun BEFORE the Hold fix was reached. Am I understanding you correctly?”
Yes. VNAV doesn’t care about the lateral mode. As long as the aircraft passes reasonably close to the holding fix in any lateral mode, the FMC will behave the same way it would if the aircraft flies directly overhead the fix in LNAV.
“To summarize, if a Hold were programmed and executed, the commanded speed would begin to slow to best Holding BEFORE the Hold fix was reached. If a TRK SEL or a TRK HLD were subsequently commanded after the deceleration began (indicating that current best Holding is the new speed target), the Holding speed would be maintained (and it would not revert to ECON Cruise) so long as the aircraft passed within a few miles of the Hold fix, even if the TRK SEL / TRK HLD began BEFORE the Hold fix was reached. If TRK SEL/TRK HLD began AFTER the Hold fix were passed, Holding speed would also be maintained in VNAV. Did I get it right this time?”
Mostly correct. TRK SEL or TRK HLD could also be engaged before the deceleration point. As long as the aircraft flies reasonably close to the holding fix, the aircraft will still slow down to the hold speed if VNAV is engaged.
@Sk999,
Would you be able to generate and post scatter plots of ACARS MH371 data vs GDAS wind components? As long as you already work with these datasets, it should not take much efforts.
@Andrew: This is getting into the weeds, but although the FCOM says “The FMC will not sequence the active waypoint when more than 21 nm off the active route and not on an offset route”, I think the deviation in track is also important. For instance, I used HDG SEL to leave the path, and returned to the path right at the active waypoint, but at a 90° angle to the path. The FMC did not sequence to the next waypoint.
“Would you be able to generate and post scatter plots of ACARS MH371 data vs GDAS wind components?”
No time now – off on yet another trip, once again trusting that the pilot will not dump the plane into the SIO. Surely you know how to do it – all the data are public.
@Victor. “Not to mention that debris consistent with an impact at 30S might have been seen in March 2014. From all that we know, I think the area around 30S qualifies at a “hot spot” to consider in the new search.”
The prime candidate as MH370 flotsam was the object with the full flaperon shape. This now has been adjudged unlikely. I would add that later removal of its trailing edge would be hard to explain. Also its flat-in-the-water appearance conflicts with the French and CSIRO floating images of its rear being above water, whichever way up it was. I doubt the added weight of the trailing edge, less buoyancy, would have the whole awash.
Of the 50 items sighted on 28th March, the ATSB summarises in its Operational Search Update, “Examination of images and recovery of items by surface vessels did not identify any items considered associated with MH370.”
What we do not know is what items were recovered, or identified as unassociated without any recovery, or when (ie the intervening drift), though there were 6 ships on task.
Whether all these were in that area is unclear since there had been 5 at 42˚-46˚S on 27th March. These would have had to transit. However numbers on task reached 9 on 1st April.
Overall it appears that with the above summary in mind, the only support offered for this latitude by this flotsam and the 16 items spotted on the 29th is based just on uncertainty; so currently is weak.
Separately on the post 25,000 sq km search, I proposed one along the 7th arc of constant width, partly because of the penalties of stopping, transiting and restarting. I omitted to mention that a 180˚ turn will entail substantial penalties also.
A hot spot would warrant the penalty of turning to widen the sweep there during initial 7th arc sweeps. However IMO there is none. Still, if search speed/season, budget and Malaysia permit, search along the medium-altitude 7th arc to 11 miles each way (2 runs to 28˚S and back) thence would make practical use of what then would amount to a hot line.
Any capacity left over could be devoted to widening what was then seen as the warmest spots, there or at its extremities. As to left overs, there has been a claim of up to 900 AUV line km/day. At 1.6 km wide that is 1400 sq/km/day. See p4,
https://www.sut.org/wp-content/uploads/2016/07/Jan-Ingulfsen-distribution-approved-PPT-at-v2-Ocean-Infinity-Perth-distribution.pdf)
The acoustic location to me is not even warm: uncertainty as to accuracy first, then the noise itself. If the aircraft sank whole, drag and an average 3 knots sinking speed (5000m in an hour), would provide little impact. (In passing, a nice study would be needed as to underwater glide distance). Fragmented, the engines and undercarriage would have little momentum: I much doubt the lost (Jan 2016) depressor registered when it hit the bottom.It was heavy and quite streamlined, though still attached to the tow vehicle.
As to implosions, the wreckage would not sink with much air entrained, eg in fuel tanks; and there would be crushing and forced displacement of the remainder, not implosion. The oxygen cylinders would implode before reaching the bottom but pretty small and low energy, including the compressed bubble of oxygen released. The size and shape of the San Juan makes that submarine quite different and that might have been an explosion anyway. (The Independent today said, “The navy has said the vessel’s captain reported that water entered the snorkel and caused one of the submarine’s batteries to short circuit. An explosion was later detected around the time and place where the San Juan last made contact.” Hydrogen maybe.
So far as I know lithium batteries in MH370 will not cause an explosion. possible that the eh
@DennisW. “My understanding is that no debris was ever recovered and examined. Is this still a true statement?”
There might have been some amongst the following, page33:
https://www.atsb.gov.au/media/5773565/operational-search-for-mh370_final_3oct2017.pdf
This says, “During the surface search a number of floating objects in the drifted search areas were reported by aircraft, including wooden pallets and fishing equipment. None of these items were assessed as being associated with MH370. Over thirty items of debris were recovered by surface search vessels. All of these items were considered unlikely to be associated with MH370.”
@Victor, about the 1400 km/day maximum search rate, aside whether this is for a period without AUV rotation, this begs the question as to processing rate on board, which the OI documentation mentions and the hold up in downloading from the AUV, assuming the memory is not physically exchanged.
@David
Constructor has used speeds of 12-13kn in the last few months when moving to and from its operational areas. As you say, the fuel cost of the long journey to Durban may be prohibitive. Constructor is now making a steady 10kn, consistent with its ETA.
Also, higher speeds would give an arrival time in Durban just before Christmas. The lack of available flights for incoming crew and non-availability of port facilities over the holiday may just make it too difficult to complete a turn-round before the 25th, so an arrival on the 27th makes more sense. I still expect them to use higher speeds after Durban.
@Victor
RE: “…I think the deviation in track is also important.”
No doubt. I don’t know the logic, but I assume the aircraft would need to be tracking in a direction somewhat less than 90° away from the flight planned track and 21 nm or less from that track. If the aircraft crosses the active waypoint at 90° then perhaps the FMC sees the distance to the next waypoint increasing and consequently it does not sequence.
I compared the ACARS data from the MH371 flight with the GDAS data for the route.
I found an average Temperature at Altitude error of -0.92°, an average Wind Direction at Altitude error of -4.9° and an average Wind Speed at Altitude error of -0.63 knots.
However, I was interpolating between the times of 00:00, 03:00 and 06:00 and adjusting for temperature following the step climbs to 38,000 feet and 40,000 feet.
@VictorI
Victor, my reference to a potential debris field observed on March 28th was based on the following quote from the Lagan article (I haven’t been able to find a photograph showing a high density debris field near the blue object):
“/Wing Commander Rob Shearer, who piloted the Orion, described it as “a bit of a frustrating day”.
He told The Times: “We’re looking for that one piece of evidence that ties objects in the water to the event, the potential accident. All we got was a lot of small contacts in the area.”
Wing Commander Shearer agreed that the large blue object picked up by his team did look “promising” and that the number of objects pointed to some kind of debris field.
“There was a lot of objects, but we don’t know if that was from the event or it could have been a whole bunch of stuff washed off a fishing boat or a container ship,” he said. “That’s what I mean by looking for that one definitive piece of evidence.”\”
The quote illustrates well the problem with the surface search: results seem to be inconclusive. Hence my suggestion to mainly rely on drift studies of collected debris.
@Niels: Thank you for that information. Based on the quote, there were some individual contacts on March 28, but no debris field was found. On March 29, there were two debris fields located, both visible in images, and drift analyses show they could have come from the same impact point. Certainly experts should be able to characterize whether the debris fields were consistent with an aircraft crash. They were either analyzed and deemed not relevant, or they were ignored.
@VictorI
Considering all information I can and based on my assumption the flight was all planned and thus intentional with a specific end-point/area included, together with the possibility of a pull-up after the steep descent and a following glide, I have my priority-search area between:
32.2S and 33S latitude
96.5E and 97.5E longitude Ob-trench
Further I go along in great lines with what @David also suggested;
-start searching the previously not searched areas between ~36S and ~32.5S including the CSIRO/Griffin hotspots according the previous search area bounderies.
-then make some sweeps along the 7th arc from ~32.5S till ~28S including your (and others) hotspot area(s).
-if nothing gets found make choices of where to search wider also depending on new information that possibly will come while the search is ongoing.
-I think flexibility is key
@Andrew, @DrB: I think the easiest way to enter an autothrottle mode where hold speed is maintained is to choose the present position (PPOS) as a hold fix while in VNAV but not LNAV (e.g, HDG/TRK SEL or HDG/TRK HOLD). The active waypoint will not automatically sequence past the hold waypoint.
Thank you for your responses. As OI will be starting the search next month (January 2018), there is not much time to provide input to the search plan, which is why I again posed the question.
One thing is clear: The images from March 29 (and possibly March 28) need further analysis to determine if they are relevant to MH370. Also, we need clarification about the other images from the surface search. Were there other images not made available to us from other search planes that captured candidate debris? As the null result of the surface search has been used as one of the factors for not searching the seabed north of 32.6S, we need to better understand what was possibly seen during the surface search.
@VictorI
As far as I know the ‘suitcase’, the ‘debris with lines attached’ and the ‘cargo package’ were not recovered. At least never read anything about it.
Those pieces remain therefore unexplained and possible MH370 debris.
As is the ‘blue panel’ which actually is a white panel almost completely overwashed with water on the picture which gives it its an overall blue color.
I think this piece could be important for it fits shape and sizes of a flaperon.
Trailing edge and other damage on the found flaperon could possibly also have been caused by battering on rocky shores for some weeks.
Reunion has more than 90% rocky shores.
One way to find out if this possibly could have been a flaperon is to conduct a floating test with an intact flaperon. If it floats flat like in the picture this could explain the barnacle growth all around till it reached Reunion.
In the end I agree with @Niels. It’s all too uncertain. The latest drift analyzis are the best lead together with the latest Inmarsat data interpretations.
@VictorI
BTW you realize you suggest confirming the search is going to re-start in January by OI?
@GR: “I think this piece could be important for it fits shape and sizes of a flaperon.
Trailing edge and other damage on the found flaperon could possibly also have been caused by battering on rocky shores for some weeks.””arrived too early”?
I happen to agree with this statement [with heavy emphasis on “might”!] But aren’t you now arguing against your statement, just a couple days ago, that northern sites on 7th arc should be discarded because the flaperon?
…arrived too early? [sorry last message posted itself while I was typing!]
@Ge Rijn said: It’s all [the RNZAF images] too uncertain. The latest drift analyzis are the best lead together with the latest Inmarsat data interpretations.
The drift analysis (based on the recovered MH370 debris), Inmarsat data, and debris images all present uncertainty, but in different ways. The Inmarsat data predicts an impact along the 7th arc, but over a broad range of latitudes. If we accept the 7th arc, then the drift analysis also predicts an impact over a broad range of latitudes. The debris images (both the French satellite images and the RNZAF images) can be used to predict the impact along the 7th arc with a much higher level of precision. However, the uncertainty lies in whether the images show debris from MH370. I don’t think you can disregard the images just because it is uncertain the debris is from MH370. In fact, based on what we know, the precision in estimating where along the arc these objects were on March 8 makes them important.
@sk999,
Nice work, Steve. Your comparison of GDAS with MH371 wind data appears to confirm that the GDAS wind error are not significantly larger than the 5.7 kts used by DSTG; that is a very direct and useful conclusion.
I am also very curious about the GDAS temperature error. Did you compare it also? If it is about 1-2C RMS, it is not entirely negligible in terms of predicting TAS and fuel flow. I have not been allowing for temperature errors in my fits, because the effect on airspeed is small compared to the tailwind error. However, the effect of temperature error on endurance is not so small (2C = 0.7% in fuel flow and endurance). One could allow for this by simply increasing the allowed endurance error, since we don’t have enough BTO and BFO data to fit a temperature error for each leg between arcs.
@Richard,
Would you please provide the RMS errors in temperature and wind in your comparison of MHG371 and GDAS. Thanks.
@Ikr
Your comment;
‘But aren’t you now arguing against your statement, just a couple days ago, that northern sites on 7th arc should be discarded because the flaperon?’
No I don’t think so. I argued debris from ~29S would arrive much to soon on Reunion and Madagaskar. The delay between beaching and discovery would be too great imo especially considering the flaperon with all its barnacles.
I was (and am) thinking of a crash-area between ~32S and ~33S. That would suite arrival times and discovery times much better imo.
@Andrew,
@VictorI,
In the case I call Scenario #2, in which no specific destination was intended, LNAV was not used for the post-FMT SIO route. This scenario requires a southern course to be set before 19:41, possibly before or even during a holding pattern. In this scenario, there appear to be only two possibilities (to be discussed below) that may satisfy all the observations (of BTO, BFO, wind, temperature, and endurance).
A common feature of these two cases is a speed reduction from the military radar track speed. This speed reduction is required to match the known endurance, and the 18:40 BFOs also imply a reduced speed at that time.
In addition, a speed profile that decreases with increasing time and with decreasing aircraft weight is required for these curved routes to match the BTOs. The only speed profile shown to match this curved route is Holding (so far I have been using Flaps Up). However, this implies a Holding pattern was commanded between 18:29 and 19:41, since there does not appear to be another means to set Holding speed in VNAV except by commanding a Holding pattern to be flown in LNAV.
Also, a constant magnetic heading (CMH) route does not appear to match the BTOs, with any available speed profile, because it is too curved. A straight CTT route near due south is essentially the same as a great circle route, which has previously been investigated under Scenario #1. That leaves only CTH and CMT as possibilities for Scenario #2, and these routes have similar curvatures.
Here are the two remaining cases identified for Scenario #2:
(Scenario 2a – CTH) – We previously investigated the case of an End of Route error caused by a lateral offset not being removed before a Hold fix was reached. This results in a constant heading course, with the reference normally being magnetic (with the NORM/TRUE switch set to NORM). Because no one has any plausible explanation why that switch would have been changed to TRUE at a location near the equator, we have therefore discounted this case (2a) as being highly unlikely.
(Scenario 2b – CMT) – A more recent development is the possibility that a Hold was commanded, and subsequently a CMT course was set using the MCP. In this case, it is also possible for VNAV to reduce the airspeed to best Holding and to maintain that speed profile until fuel exhaustion occurs, even though LNAV is no longer controlling the course and there is no route error. The Holding fix could be an upcoming waypoint or the present position (PPOS). With PPOS the deceleration in airspeed would begin immediately when the Hold is enabled. For a Hold fix at an upcoming waypoint, the deceleration would begin so that Holding speed would be reached at the same time at the waypoint is overflown, and this would require roughly 30 seconds of deceleration.
In Scenario 2b, the MCP course could be set in one of two ways: (1) the PIC sets the HDG/TRK switch to display TRK degrees (if the switch was not already in that position), dials in the desired number of degrees (i.e., 180), and pushes in the SEL switch, or (2) the PIC sets the HDG/TRK switch to TRK (if needed) and pushes the HLD switch in. Method #2 is the simplest means of exiting/avoiding the racetrack pattern and assuring continuation of the present southern course.
An immediate question arising from Scenario #2 is why would a PIC select a Hold and then exit the Hold by setting a fixed course? This question is common to both Scenarios #1 (extended loiter) and #2, although the answers might be different.
For Scenario 2b-1 in which the course is dialed in, I think there would be a greater likelihood of a number like exactly 180 degrees to be used, so, if that is the best fit to the satellite and weather data, it might be favored over the TRK HLD case (2b-1). Scenario 2b-1 is implemented by simply pushing the HLD button. In that case a CMT in the neighborhood of 180 degrees is possible, but the exact course does not need to be an integer number of degrees, and it will be offset from the current true track by the current magnetic variation.
We know that in Scenario #2 the Hold could not have been flown for an extended period of time; otherwise the BTOs and the endurance cannot be matched at Holding speed. The brief Hold (or perhaps no Hold at all) could have occurred either before 18:40 or after 18:40. If we interpret the 18:40 BFOs as indicating a southbound course, that still does not tell us if the MCP course was set before or after 18:40. For courses near 180 degrees, the 18:40 BFOs imply a reduced speed at that time. This, in turn, argues for the Holding speed to have been (mostly) achieved by 18:40, and this argues that the Hold may have been enabled before 18:40.
Victor, you said: “I think the easiest way to enter an autothrottle mode where hold speed is maintained is to choose the present position (PPOS) as a hold fix while in VNAV but not LNAV (e.g, HDG/TRK SEL or HDG/TRK HOLD). The active waypoint will not automatically sequence past the hold waypoint.“
Are you saying that the MCP course could be set first (thereby disabling LNAV control), followed by setting a Hold at PPOS, and this would keep the MCP course but change the speed to Holding? If this is true, then the order of the two commands (TRK HLD and Hold at PPOS) is immaterial if they are executed closely together in time.
Andrew, is there a situation where you would want to use Holding speed but not fly a racetrack? Please consider the cases of a constant flight level and a descent to a lower altitude.
@DrB asked me: Are you saying that the MCP course could be set first (thereby disabling LNAV control), followed by setting a Hold at PPOS, and this would keep the MCP course but change the speed to Holding? If this is true, then the order of the two commands (TRK HLD and Hold at PPOS) is immaterial if they are executed closely together in time.
Yes, HDG/TRK SEL followed by a PPOS Hold would set the speed to holding speed. Regarding the order, remember that if in LNAV mode and a hold at PPOS is executed, the plane will start to follow the racetrack pattern, so the order does matter. The order of a closely spaced TRK HOLD and Hold at PPOS doesn’t matter if the mode was not previously LNAV.
I will add that the more I study various scenarios between 18:22 and 19:41, the more I realize it is almost impossible to assign a high probability to any one. And as I have said previously, it really doesn’t matter which path we think is the correct one if it terminates within the 25,0000 sq km, as that area will be searched first.
@VictorI
I certainly don’t disregard those images. I think I showed this in my ‘search strategy proposal’.
I only agree with @Niels those images are too uncertain to put more weight on compared to the drift analyzis and other data.
Those observed pieces could be most telling but we just don’t know.
It sure is interesting they were floating there in the remote SIO at the time MH370 disappeared.
@Ge Rijn: As I said, there is uncertainty with all the data. What the images offer is locational precision, which the drift and Inmarsat data do not, although if the objects are not from MH370, the precision is useless.
@Victor
As you know, I have a lot of respect for many of the posters here. Not intending to name names, but when Richard Godfrey or Don Thompson posts something it captures my attention. Many others as well.
I think your recent post above along with the Richard’s drift analytics deserves priority over CSIRO’s work. If it were my money, I would but the AUV’s in the water at 30S.
@DrB
RE: “…is there a situation where you would want to use Holding speed but not fly a racetrack? Please consider the cases of a constant flight level and a descent to a lower altitude.”
Holding speed implies maximum endurance. In normal operations, it is only used when there is a requirement for delaying action such as ATC holding in a holding pattern. The only logical reason to use holding speed outside a holding pattern is if the pilot wishes to remain airborne as long as possible on the remaining fuel, without trying to maximise the range. Otherwise, MRC would be a better choice.
RE: “If this is true, then the order of the two commands (TRK HLD and Hold at PPOS) is immaterial if they are executed closely together in time.”
If LNAV is engaged in the hold at PPOS case, the pilot would need to immediately engage TRK HLD or TRK SEL after executing the hold. Otherwise, the aircraft will start to turn and enter the hold.
My first comment above should read:
“The only logical reason to use holding speed outside a holding pattern is if the pilot wishes to remain airborne as long as possible on the remaining fuel and fly a long distance, without necessarily trying to maximise the range.“
@DrB, you said @ December 3, 2017 at 1:16 pm
“(Scenario 2a – CTH) – We previously investigated the case of an End of Route error caused by a lateral offset not being removed before a Hold fix was reached. This results in a constant heading course, with the reference normally being magnetic (with the NORM/TRUE switch set to NORM). Because no one has any plausible explanation why that switch would have been changed to TRUE at a location near the equator, we have therefore discounted this case (2a) as being highly unlikely.”
I think it depends on what you define or regard as plausible, and who is in the cockpit making the decisions at that time.
A B-777 rated pilot, may, or may not have a reason to set TRUE at or near the equator.
On the other hand, any other person, say a yachtie, or anyone else with any knowledge of navigation, and / or if they had ever looked at any nav chart of the IO, would be familiar with the isogogonals, and know full well that mag var in the IO is a BIG issue. Selecting true woud be high on their list of simplifying their nav problem, or nav strategy (depending on their intent).
If we assume that it was Z and the intent was to vanish, then setting true at this point, is an obvious additional PLOY, to me at least.
Having just flown though, and now heading out of, an area of little mag var, the difference in actual path flown with the switch in either position, up to this point, are minimal, and probably undetectable, by any ivestigation, post fact.
When crossing the equator and heading off into the Io though, the situation vastly changes.
He would assume, that the investigators (both oficial and unoficial) would subsequently assume, that the switch would have remained in norm, because the official investgation at least, would, and has, assume, that the flight was a ghost flight by this stage, so there is no one alive to switch it anyway, so they will assume that it wasn’t switched.
He knows that switching to true now will produce a more westerly straight path, (away from land) than the left swinging – and increasingly easterly diverging path, that NORM would produce (towards land, ie, Australia).
It would be one more feather in Z’s briliant and meticulous planning cap, to throw the investigators further off the scent, particulalrly if he was confident that there was no other tracking information available to the investigators, ie, he did not know about the Inmarsat pings, or their implications, as we now know them. I think it is reasonable to assume he did not know about that.
So, I think it would be unwise, to actually discount Scenario 2a.
Give it a lower probability perhaps, put it on the back burner, but discount it, no.
@DennisW said: If it were my money, I would but the AUV’s in the water at 30S.
David Griffin has seen the blog post and has agreed to do a drift analysis for the debris in the images. Considering that I used his drift model results to evaluate the impact location of 29.7S latitude, I would be surprised if he arrives at a different conclusion. Hopefully, he’ll be able to say more at a later date. I have a lot of respect for his work. It would be helpful if imagery experts also analyzed the photos, if that hasn’t already been done.
@DrB said;
“Because no one has any plausible explanation why that switch would have been changed to TRUE at a location near the equator, we have therefore discounted this case (2a) as being highly unlikely.”
Maybe this? If the flight was all planned with the goal to let the plane vanish in the SIO leaving as less possible evidence, the PIC would have anticipated the possibility the plane could be found anyway some day and also the FDR. Entering an end-waypoint would be recorded and would give conclusive proof of deliberate intention by the PIC.
If only the switch was set to TRUE around the equator (the PIC would have known his destination was the SIO) this would be recorded also but when the FDR gets found it would still leave anyone guessing what happened after FMT and why.
@DrB, @Ge Rijn: If the intention was to divert the plane to the SIO (which I believe is the most likely scenario), the A/P modes that best accomplish this are CTT and LNAV, flying at a mileage-efficient cruise speed and altitude after 19:41. I’ve (somewhat patiently) listened to many theories here involving hypoxic-induced errors, but the tag-team hypoxia required for bad decisions between 17:21 and 18:40 is very unlikely.
@Ge Rijn said: BTW you realize you suggest confirming the search is going to re-start in January by OI?
As far as I can tell, the only way the search will not re-start in January is if the Malaysian PM decides to not proceed. OI’s actions are consistent with starting in January. My recommendation is to continue to disregard news reports other than statements attributed to OI.
@Victor
“I believe is the most likely scenario), the A/P modes that best accomplish this are CTT and LNAV”
I would add CTH. I am working (very slowly) on a case to try to show, with the winds, CTH from ISBIX is M0.81 whereas CTT is more like M0.84 and they end up at the same spot. CTT or LNAV has to fight the wind, right?
I also perceive that using CTH mode, gives a potentially “perfect” BTO/BFO match for Arc3,4,5 – in other words it almost seems like you can see the wind impact.
@VictorI @TBill
I also think a CTH-route after FMT should be considered on those arguments too (@TBill).
In my thinking it could also bring the plane near enough to a specific destination/area without leaving conclusive clues in the FDR why it ended up there (if the FDR ever gets found).
A steep descent after FE would suggest an uncontrolled high speed impact.
A pull-up and glide after the steep descent won’t show on the FDR for there was no power anymore to the FDR.
It would be a perfect strategy to disguise responsibility and what really happened when the FDR ever gets found.
Probably my imagination is running wild a bit..
This from the Times of London. The article focuses on the “blue panel” from March 28. Of course, my blog post was mainly looking at the debris on March 29:
‘Missing flight’s debris spotted by The Times’
Bernard Lagan, Sydney
Experts monitoring the search for the Malaysia Airlines jet that went missing in the Indian Ocean nearly four years ago have suggested that pictures first published in The Times three weeks after it disappeared showed wreckage from the lost plane.
The significance of the images may have been missed by Australian search organisers who failed to find any surface debris from Flight MH370, which disappeared in March 2014 with 239 people on board.
A group of independent scientists, engineers and aviation experts has published a paper which suggests that the debris photographed was ignored or wrongly identified by search controllers.
The Australian government has long rejected accusations that it missed or ignored potential debris in pictures taken by the 22 aircraft that searched millions of square miles of the Indian Ocean in the six weeks after the jet vanished. An international £113 million search has proved unsuccessful but the government has insisted that none of the photographed objects was “assessed” as being connected to it.
The pictures were taken over the Indian Ocean from a New Zealand air force Orion maritime patrol aircraft on March 28 and published the next day in The Times, which had a reporter on the search plane. The leader of the surveillance crew on the Orion, which included airmen seconded from the RAF, told his men he believed that they had found a debris field while searching for MH370, which had gone missing 20 days earlier on March 8, 2014. The plane was on an overnight flight from Kuala Lumpur to Beijing when it mysteriously turned around and headed into one of the world’s remotest areas.
Australia’s Maritime Safety Authority, which co-ordinated what became the world’s biggest aerial search, issued a press release saying that the New Zealand search aircraft and four other planes had sighted many objects on the same day. It said: “The objects cannot be verified or discounted as being from MH370 until they are relocated and recovered by ships.” Few if any of the objects floating about 1,200 miles west of Australia, were retrieved but the authorities later said they were confident that none was from the missing jet.
The claim has long puzzled members of the MH370 Independent Group, a collection of scientists who are investigating the mystery and monitoring the search. The paper’s author, Victor Iannello, an American engineer who has published several articles on the disappearance, has suggested that a large, rectangular panel photographed from the Orion and published in The Times could have come from the missing aircraft. He said that other military search plane photographs, some taken in the same place the next day, and obtained by his group’s members, included objects that resembled a suitcase, a panel with wires and a cargo package.
It was 500 days before the first confirmed piece of wreckage from the jet washed up on the island of Reunion. Several other items have washed up on islands in the eastern Indian Ocean and on the coast of Africa.
A new underwater search is likely to start in January. The Malaysian government is expected to sign a contract before Christmas with an American company that will be paid a multimillion-dollar success fee if the missing plane is found. It will search a newly identified 10,000 square mile section of the Indian Ocean floor believed by Australian government and other international experts to be MH370’s most probable crash zone.
And here is an article from The Australian.
First some corrections: I never said that I lead the IG, and I never said the search of the first 25,000 sq km was “misguided”. What I told the journalist in an email was “We don’t know if the objects in the images taken on March 28 and March 29, 2014, are pieces of MH370. The images from March 29, and in particular the two images of the two debris fields, deserve further attention.”
‘Photographs of debris a possible link to MH370’
Rhian Deutrom
Aviation experts have called on authorities involved in the search for the missing MH370 plane to re-examine photo¬graphs captured by the Royal New Zealand Air Force during the initial surface search in March 2014.
Authorities are plotting locations for the next subsea search, to be conducted by private company Ocean Infinity next year, in a bid to locate the remains of the plane, which was carrying 239 passengers and crew when it disappeared over the Indian Ocean.
US-based engineer Victor Iannello, who heads a group of scientists and aviation experts known as the MH370 Independent Group, said photographs of debris in the Indian Ocean were disregarded by original search teams and “demand a new level of attention in light of the prospect for restarting the search”.
Dr Iannello said proposed plans to search a 25,000sq km area of ocean were misguided, pointing instead to an area just north of the site, where items were spotted by aerial cameras in the first 21 days after impact.
“Many believe that … it is possible that debris was missed during the surface search,” Dr Iannello said.
The photographed items include two debris fields containing small fragments, an object resembling a suitcase, a large metal panel, an item containing wires and a cargo package.
“As far as we know, none of the objects identified in the surface search on March 29, 2014, were recovered by ship, so the relationship to MH370 remains unknown,” Dr Iannello said. “As only 21 days elapsed between the impact of MH370 and the discovery of the debris, it should be possible to backtrack the objects to March 8 to determine a potential point of impact.”
A spokesman for the Australian Maritime Safety Authority, which assisted in the initial surface search for survivors, said any photographs taken from the air were referred to the expert working group for investigation.
“During the search for MH370, many objects were observed by aircraft in the search areas,’’ the spokesman said.
None of the objects recovered “were confirmed to be related to MH370”, he said.
@VictorI
Yes it was actually me who put more focus on the ‘blue panel’ so you can blame me for that one..
Great to see the media still pick up on MH370 and the work you are doing with- and on your blog is reqocnized so seriously.
@Ge Rijn: No, I think Bernard Lagan was already very interested in the “blue panel”, as evidenced by this video.
@VictorI
I mean in the comment section I tried to put more focus on the ‘blue panel’ shown in the video with Bernard Lagan. I requested to add a picture of it to your topic to put it in context and @Barry Carlson chimed in on it.
Further it has been ignored mostly on this topic which is a bit of a pity I think.
But it has not been overlooked. I guess it´s on the backburner which is fine with me also.
This story will now get repeated over and over, errors and all.
Hi Victor: The Fake News, hype and headlines are annoying to say the least. Best ignored. It is right to reconsider the aircraft and satellite photographed debris for clues. It is happening in multiple places. Thank you for drawing attention to the March 28/29th debris in particular. And thanks to Brian for obtaining the “raw NZ data”.
Some number of comments back there was discussion of seabed impacts. It’s very unlikely that the CTBTO hydroacoustic arrays will record anything of a seabed impact, for the same reason surface events do not translate into the deep sound channel. Too much energy is lost at the water-seabed interface & dissipated before any refraction occurs at the upper DSC boundary.
The example of the lost Argentinian submarine shows how acoustic events, initiated on a sloping continental shelf, do translate into the deep sound channel & travel considerable distances.
Ocean Infinity has moved its ship to the Indian Ocean without a signed contract. Presumably, if this didn’t happen now there would be inadequate time left this season.
https://www.thedailybeast.com/mh370-search-is-back-onthis-time-with-robots
@Richard Cole @Victor
Is there any news on Fugro or the” local Malaysian company” getting prepped showing their willing, able and ready, while waiting for any contractual finalization. Or is it only OI that have been closer to a deal?
@Joseph Coleman
I did look at the Fugro boats a few days ago and none were in the area or moving towards the area.
My experience of government-type contracts is that negotiations are started with the winner of the bidding process, until a contract is agreed or the process fails. If the latter they move to the second-placed acceptable bid. I am sure there are many counter-examples.
@Joseph Coleman: I believe only OI has offered the “no-cure, no-fee” contract type. I do not believe that the other two companies are under serious consideration at this time. It’s likely OI or nobody.
@Richard Cole @Victor
Thanks for comments it’s good to see, that efforts are underway to continue search, let’s hope that this time it’s a step further to finding answers.
@VI [Times article] “The pictures were taken over the Indian Ocean from a New Zealand air force Orion maritime patrol aircraft on March 28 and published the next day in The Times, which had a reporter on the search plane. The leader of the surveillance crew on the Orion, which included airmen seconded from the RAF, told his men he believed that they had found a debris field while searching for MH370.”
So we have a phenomenon that drew the attention of several trained observers and a London Times reporter. And mention that the concentration of debris was seen by 4 other patrol aircraft. And discarded as a negative/non-positive by experts in photoreconnaissance [I presume].
I’m guessing that the decisions was based on the photos we see in the archive [hand held,, decent digital SLR, through aircraft windows, at some altitude and speed].
A lot of questions come to mind: where the still photos backed up by other sensing [photo or radar] mounted on the Orion? Where patrol aircraft allowed to circle for better looks, or were they charged to stay on pattern? could they/did they drop beacons on promising concentrations? Perhaps this was reported at the time, but I don’t recall hearing.
Just my thought — that we normally assume that photo evidence is objective and trustworthy, but eyewitness, not so much. Contrary to this, the still photos don’t necessarily document what drew the observers’ attention — a summation of these seeming to be concentrations of objects that just looked ‘different’ perhaps. Or something the commander or other observers remembered from previous missions?
Bottom line, I’d agree with others here that even a 10% or so probability that these observations were related to MH370 is sufficient to define a hot spot..
@ikr
My guess, and it is only that, is that OI had no intention of starting a search before January 1. Simple tax code weighs against it. Gambling losses from year x cannot be deducted against gambling winnings in year x + n. So any losses (expenses) incurred in a no cure no fee contract in prior years would not be deductible against gains in in tax year x+n. Basically under the terms of the contract OI would be foolish to start a search in 2017. None of the expenses would be recoverable unless the wreckage were found in 2017. There are more than a few subtleties associated with the OI approach. Basically a no cure no fee contract is a pretty unusual way to do business.
I have no idea what the state tax codes are where OI is incorporated. California is basically the same as the feds – you get raped.
@Victor
It is my opinion, shared by informed others, that the OI effort will be a fiasco. The conditions in the search area are not at all amenable to the OI approach which is basically more in tune with scanning a sheltered area. We shall see, but I am not at all hopeful of a positive result even if OI gets started which is by no means a given. You are an optimist. Nothing wrong with that except that it is wrong.
@VictorI,
I disagree with you on several points.
You said: “If the intention was to divert the plane to the SIO (which I believe is the most likely scenario), the A/P modes that best accomplish this are CTT and LNAV, flying at a mileage-efficient cruise speed and altitude after 19:41.”
The simplest means of sending the plane into the SIO is to enter a CMT route before 18:40, not to spend an hour circling near an air force station before heading south toward some LNAV waypoint. There is no evidence for a “failed negotiation”, other than the fact that such an odd, extended-loiter route may match the very limited satellite data.
You also say there are as many LNAV solutions as there are far-south waypoints, but none of these matches the recovered data points, so that general similarity may just be a coincidence.
Then you said: “I’ve (somewhat patiently) listened to many theories here involving hypoxic-induced errors, but the tag-team hypoxia required for bad decisions between 17:21 and 18:40 is very unlikely.”
Do you want a gold star for listening to other ideas besides the one you promote? Other people do that, too, you know.
I submit EVERY proposed cause is highly unlikely. This was certainly an unusual, highly improbable event. The contortions required to explain a “failed negotiation” followed by mass murder and suicide are extreme, to say the least. That may also be true, in your opinion, for a cascading sequence of on-board failures. I am not promoting any particular cause for the SIO route; I am just trying to find those few routes which match the available information.
Lastly, why do you conclude that a southbound FMT before 18:40 must be due to hypoxia? I don’t know why a FMT occurred, and I certainly don’t know whether or not hypoxia was involved, but hypoxia has happened before and cannot be discounted completely. Who knows even if either of the pilots made the FMT? It appears that a trained pilot was navigating until shortly before the FMT (because an airway was followed), but you don’t need a pilot after that, just a crew member who knows how to operate the MCP (and I recall reading they all did).
Let’s face it. There is no “rational” explanation for MH370 – not suicide, not murder, not a simple failure of on-board systems, – IF you work under the assumption that the aircraft was navigated by a single “rational” pilot (even if bent on suicide), or even by a pair of “rational” pilots. If you allow more complex scenarios where the pilots are incapacitated (not necessarily by hypoxia), and the crew then attempts to fly the plane, you could get a flight path which appears “irrational.”
@Oleksandr,
You said: “@DrB, It is obvious that various models use the same kind of instruments, e.g. surface stations, buoys, etc. to obtain measured data. It is not obvious whether GDAS uses the same data as ACCESS-G to generate gridded data. Especially in the remote SIO.”
Yes, my point was simply that if the two global-data systems use the same types of data (not necessarily the exact same data), their errors are likely to be similar.
@ DrB ….. “The contortions required to explain a “failed negotiation” followed by mass murder and suicide are extreme, to say the least …..”
This was not the way it was planned to turn out.
Irrespective of the outcome of negotiations Captain Zaharie intended to land in Indonesia, release the passengers and put himself at the mercy of the Indonesian authorities.
There were two stages: fly out into the Andaman Sea and come back towards Indonesia while negotiations commenced. When out of radar range listen out for confirmation of success and proceed to a landing in Sumatra. In the event negotiations were not finalised come round below Sumatra waiting for confirmation. The latest point before turning up for a landing on Java was Cocos Islands. As it turned out that was too late and MH370 ended up ditching almost in sight of land near the 7th arc.
The Indonesians were aware of what happened, in fact the Indonesian Police Chief let slip this fact to some senior airline officials at an aviation seminar 6 months later.
@ventus45,
You said: “I think it depends on what you define or regard as plausible, and who is in the cockpit making the decisions at that time. A B-777 rated pilot, may, or may not have a reason to set TRUE at or near the equator. On the other hand, any other person, say a yachtie, or anyone else with any knowledge of navigation, and / or if they had ever looked at any nav chart of the IO, would be familiar with the isogogonals, and know full well that mag var in the IO is a BIG issue. Selecting true woud be high on their list of simplifying their nav problem, or nav strategy (depending on their intent).”
I agree that a non-pilot could be more likely to use TRUE rather than NORM.
You also said: “So, I think it would be unwise, to actually discount Scenario 2a. Give it a lower probability perhaps, put it on the back burner, but discount it, no.”
I think we are in agreement. I said “. . . discounted this case (2a) as being highly unlikely.” I did not intend to infer that the probability was zero, just that I thought it was very low (but this can also be said for all the proposed scenarios).
@Andrew,
Thanks for providing your additional insight into the TRK HLD/PPOS scenario.
Now I have an off-the-wall question for you. I was wondering if there is any way the Right/Left turns we now construe as a SLOP at 18:25 could have been the beginning portion (1/2 loop) of a Holding pattern. I can’t see how to make it work unless the racetrack is aligned perpendicular to and then parallel to N571 and is roughly square (12 NM on a side). Can one do a racetrack like that, where the first two legs are a Right turn to a perpendicular to N571, then a Left turn parallel to N571, etc., and the legs are of about 12 NM length each? I find it interesting to contemplate a Hold (at NILAM?) very soon after power is restored to the SDU and possibly to the left bus.
@TimR
We both know what happened, I put DrB’s rant in the “nerd” category.
@Andrew,
I don’t know if it is easy or even possible to fly a Holding pattern like the one I described above (with an entrance at 297 degrees true on N571 and “sides” at 27/297/207/117 degrees), but I will note the following:
1. The 18:25 BFOs are consistent with a 27 degree track first leg after the first right turn.
2. The 18:27/28 BFOs are consistent with a second leg (perhaps longer than 12 NM) at 297 degrees.
3. The 18:40 BFOs are consistent with a 207 degree leg (the second time around the racetrack).
Is it possible that the 18:40 phone call occurred during a Hold (near NILAM either at the PPOS or by using a NILAM fix) that began circa 18:24? If that actually occurred, then the FMT to exit the Hold could have been well after 18:40. An interesting aspect of this scenario is that such a Hold could have been programmed as early as 18:23-18:24, and only one action (TRK SEL or TRK HLD, for instance) is needed later on to exit the Hold and head into the SIO while maintaining best Holding speed.
@DrB
Interesting question you posed to Andrew, I have been looking at something similar. Nearest I can get to something approaching an offset-like manoeuvre is a parallel entry to a hold at present position but with an inbound leg that is the reciprocal of the track; it would look something like this. Like you, I was wondering if you could end up offset to your previous track by 12-13 nm at 1825-1828 UTC and then end up on a southerly heading about 14 minutes later.
I hasten to add that a hold at PPOS on a reciprocal track is by no means a “standard” manoeuvre.
@DrB
RE: “Can one do a racetrack like that, where the first two legs are a Right turn to a perpendicular to N571, then a Left turn parallel to N571, etc., and the legs are of about 12 NM length each?”
A pilot could theoretically fly such a pattern by using TRK SEL to turn the aircraft onto the appropriate track at the appropriate time; however, I doubt very much a pilot would do so. Such a pattern would be non-standard and why bother when it’s so much easier to plug a holding pattern into the FMC and let the autopilot fly it in LNAV?
@Mick Gilbert
Nice drawing, but the entry is always flown on the same side as the holding pattern to keep the aircraft inside the ‘protected’ airspace!
@Andrew
There you go, I knew I should have run that by you first.
@DennisW wrote “the OI approach which is basically more in tune with scanning a sheltered area”
During the just passed months of Sept and Oct Ocean Infinitiy’s trials had Seabed Constructor working in the open mid-Atlantic, north of the Azores. Not at all sheltered out there. Other areas included in the vicinity of UK Scilly Isles, off west Ireland coast, and off South Carolina: within 100nm of shore but hardly “sheltered”.
(Tracking info courtesy of Marine Traffic, Vesselfinder, AIS sources)
@DrB: You’ll have to show me where I proposed the failed negotiation scenario. In fact, if you check, you’ll see that although I think it was possible, there are reasons for me to doubt this occurred. What I do believe is POSSIBLE is that there was a hold and descent before flying south. I have long ago come to the conclusions that is impossible to propose a flight path between 18:22 and 19:41 that has any reasonable probability of being correct. That’s why I start at 19:41.
Why the descent/hold interests me is there are many LNAV and CTT paths that satisfy the satellite data that terminate located north of the 25,0000 sq km that will be searched. As OI is preparing to start the search with the 25,000 sq km and then will move from there if the wreckage is not found, I feel is helpful to consider paths that end north of the 25,000 sq km. Some here are doing that.
And of course, some of us are willing to consider the relevance of the simulator data.
@Ikr
I have to come back to you for you were right I seemed to suggest areas north of 32.6S better be discarded, which is not what I stated though.
I said north 0f 32.6S becomes the more problematic the more north you go cause of the delay in arrival and discovery times.
I actually had my previous comments to @Oleksandr and @VictorI in mind about the ~29S latitude when I wrote that comment. Imo this problem is really problematic at and above this latitude. Not impossible (I’m not in a position to state something like that).
@VictorI
On the revelance of the similator data I like to mention Jeff Wise made an intriguing point in his latest topic imo.
He mentioned the final attitude of the plane at 45S in the simulated flight showed a pitch reflecting an almost perfect angle for a glide after fuel exhaust.
What is your opinion on this?
@Victor
Seems like there are 3 basic options for 18:25 to 19:41 based on fuel considerations:
(1) Early FMT to 34-40S
(2) Early Hold to save fuel for non-active-pilot SIO journey 26-32S
(3) Evasive flying around the Andamans until 19:41 for piloted SIO journey
I am assuming option 3 is possible, but that active-pilot case has been less studied. Option 3 presumably needs a slow-down (after Arc5) to account for fuel remaining until Arc7.
@DrB
Victor told you he has no idea what happened between 18:29 and 19:41.
Dr Bobby, Let me come to the rescue:
18:29 (coordinates N07.18, E95.68) Pilot replaces IGOGU with ANOKO and cancels N571 offset, aircraft flies toward ANOKO on bearing 268deg. Maintains FL350.
18:35.67, 20NM before reaching ANOKO, pilot replaces ANOKO with manual waypoint coordinates S40.50, E88.00 (S40:30:00, E88:00:00) Aircraft begins FMT toward manual waypoint calculated to be at least 170NM beyond fuel exhaustion. FMT radius taken as 7.5NM, turns through 81.2deg at 1deg/sec. The pilot had evidently practiced a similar procedure on his simulator on 2nd February 2014.
18:36 A few seconds after starting the turn south, the pilot reduces speed from M0.84 to approximately M0.67*, maintaining FL350. The 23 minute period at reduced speed M0.67 was probably intended to extend the endurance a few minutes, to make sure fuel exhaustion would coincide with sunrise. When the fuel eventually ran out, the Sun was slightly less than 6 degrees above the eastern horizon. * The BFO values recorded at 18:39/18:40 during the first telephone attempt indicate that the aircraft was flying south at about M0.67 at the time.
18:37 FMT completed, at position N07.0236, E94.6022 (coordinates N07 01 25, E94 36 08), bearing 186.8071.
Now try this for yourself: Go to the Movable-Type Scripts online Vincenty calculator, scroll down to the Direct Solution box, enter 07 01 25N, 94 36 08E as the start point, enter bearing 186 48 25.66, distance 5305132.87 meters and see where it goes to. The precision is impressive: 0.0012 arc sec is one inch at S40.50! If you back the distance off by 293832.87 meters ,ie to 5011300 meters then you get the 7th arc crossing point at S37.884 E88.516. You have just recreated the actual flight path taken by MH370.
19:01 Pilot ends 23 min reduced speed phase, increases speed to M0.82 and at same time begins climb from FL350 to FL390. M0.82 at FL390 is then maintained until fuel exhaustion.
19:41 Aircraft makes grazing contact with 2nd arc: distance covered between 18:29 and 19:41 is 552.3NM , being 56.50NM from 18:29 to start of FMT, plus 11NM for FMT , plus 484.82NM from FMT until 2nd arc crossing, 460.25Kts avg GS, 1.8Kts avg TW, TAS458.45Kts = Mach 0.78 (for Mach1 593) 2nd arc tangent contact S1.0333, E93.65.
Aircraft proceeds toward manual waypoint S40.50, E88.00
2nd to 3rd arc: 486.42NM, 486.42Kts avg GS, 1.6Kts avg TW, TAS484.82Kts = Mach 0.82 (Mach1 591Kts, ) 3rd arc crossing S09.1148, E92.6897
3rd to 4th arc: 492.83NM, 489.81Kts avg GS, 4.6Kts avg TW, TAS485.21Kts = Mach 0.821 (Mach1 591Kts) 4th arc crossing S17.2965, E91.6713.
4th arc to 5th arc: 484.12NM, 484.77Kts avg GS, Zero avg wind, TAS484.77Kts= Mach 0.82 (Mach1 591Kts) 5th arc crossing S25.3217, E90.579.
5th arc to 6th arc: 698.18NM, 467.37Kts avg GS, 13.5Kts avg HW, TAS480.87Kts= Mach 0.82. (Mach1 586.5Kts) 6th arc crossing S36.8681, E88.7062.
6th arc to 7th arc: 61.5Nm, 434Kts avg GS, 33Kts avg HW, TAS467Kts = Mach 0.81. (Mach1=577)
This path has the following features:
1) Close to minimum BTO errors.
2) Compliance with DSTG Bayesian analysis
3) Close to maximum range for M0.82, FL390
4) Synchronization with local sunrise
5) Manual waypoint S40.50 E88.00 in order to achieve feature 4)
Conclusions;
1)Because aircraft was not found within the original search area, BFO at 00:19 requires reappraisal.
2)Drift studies can be consigned to the paper bin.
@Don
During the just passed months of Sept and Oct Ocean Infinitiy’s trials had Seabed Constructor working in the open mid-Atlantic, north of the Azores. Not at all sheltered out there. Other areas included in the vicinity of UK Scilly Isles, off west Ireland coast, and off South Carolina: within 100nm of shore but hardly “sheltered”.
You forgot a link to any results. Including:
1> imagery
2> sea state
3> scanned area per unit time
I would provide a link to my collection of “big foot” photos, but Ami plans to monetize them after my passing.
@TBill
I would add an option 4.
The plane started a descent after ~18:25 and then flew straight on towards 10N/90E. The descent was captured with the 18:40 sat-phone call.
The plane made a Major Turn back towards Sumatra somewhere not that far before 10N/90E and made another less drastical turn to the right, before entering Indonesion radar space, towards the SIO.
Imo this is more plane and simple without loiters or other complicated race/holding/warning tracks and so on.
My best guesses of where to search (not based on my own analysis but rather my favorites of models put forth by others):
First of all, something happened that caused the PF to adjust the plans starting at 18:22 Then there were a series of turns (not a single FMT) and the flight headed South at:
1) 180° CTH but was blown to the east due to the winds
2) fixed to 45S 104E (the final points on ZS’s Sim)
3) McMurdo
and search the area where each of these three paths crosses the 7th Arc.
@TBill-
As you can see from above, I like your 180° path. My sequence of events say it was chosen by someone other than the person who started the initial diversion.
One thing I would suggest is you should include the fuel endurance in your models. It appears that many here accept the endurance to about 00:17. Using this value in a model helps eliminate some of the whacko routes.
@Andrew,
@Mick Gilbert,
Thanks for commenting on the possibility of a Hold near NILAM. If I understand correctly, then the problem is the initial right turn followed by a left turn in the opposite direction, since normally all the turns would be in the same direction. Is that correct?
I am not generally a fan or more complex solutions, but what if there was a right SLOP followed by a PPOS Hold so that the subsequent racetrack is to the left side of the offset track and is followed in a counterclockwise direction with all racetrack turns being to the left?
@Rob
You state “2)Drift studies can be consigned to the paper bin.”
I missed your evidence for this bold statement!
@Richard
Hi Richard. Nice to speak to you again. I initially misread your post, I thought it said it was my proposed flight path that could be consigned to the paper bin.
The short answer is I don’t need to provide evidence. The identification of manual waypoint S40.50 E88.00 obviates the need for any further evidence. It was always apparent to me that CSIRO’s revised drift studies were commissioned for a specific reason; to get the ATSB out of a fix when their search failed. The search failed only because the ATSB weren’t permitted to extend the search downrange of the hotspot until they found the wreck site.
This search was hamstrung by overriding political constraints. My only hope is that OI aren’t gullible enough to be taken in by the political duplicity. After all, it’s their money they are going to be spending.
That’s about the long and the short of it. Think about it: there is only one set of BTO rings. And it’s this set, and only this set that permits the aircraft to be flown efficiently at high altitude to rendezvous with local sunrise, with minimum BTO errors toward a precise simple, easily remembered manual waypoint a hundred or so miles beyond the fuel limit.
The last time we spoke, I was having difficulty explaining why it was I had the temerity to suggest that the Captain diorlld it. Things have moved on a lot since then. Most people (well most rational people) now accept that the Captain did it. That’s still your problem, you simply refuse to accept that the Captain did it.
You are going to be able to put off the inevitable for much longer.
@Ge Rijn: I produced videos some time ago showing how the simulated plane was hand flown between 45S1 and 45S2, and then what the flight would look like with no pilot inputs after 45S2. I’m not sure what that proves. We don’t know how the plane was flown in the simulation after 45S2. Even if we did, be it a glide or a steep descent, what does this tell us about how MH370 was flown after fuel exhaustion? What we know is there was a steep descent of around 15,000 fpm.
It is interesting to observe the ongoing discussions as each of the protagonists continue to push their favourite hypothesis/track solution. Seems each are becoming more and more adamant that theirs is the [only] correct solution.
The most adamant amongst them is perhaps Rob. A couple of comments, Rob. Where is the evidence, any evidence, of someone wishing to terminate a flight to coincide with local sunrise? Grazing contact with the second arc? I don’t think so. The sequence of BTO pretty clearly illustrates that the aircraft crossed the 2nd arc from outside to inside, and then again from inside to outside. The first crossing can be estimated at about 19:53. Reappraise the BFO at 00:19? Well, only because it doesn’t fit your hypothesis.
@Rob
Your proposed scenario is also not compatible with the lack of an IFE logon during your extended glide.
@VictorI
On your comment;
“We don’t know how the plane was flown in the simulation after 45S2. Even if we did, be it a glide or a steep descent, what does this tell us about how MH370 was flown after fuel exhaustion?”
If one believes the simulated flight is connected with the real flight than I think it’s at least intriguing the final similator point shows an attitude similar to a glide-attitude.
I see it doesn’t prove anything about how MH370 was flown after fuel exhaustion.
Paul Allen’s Petrel announces discovery of the USS Ward [and several Japanese vessels] in the Surigao Strait. Used AUV to locate and document. Surely shooting fish in a barrel compared with MH370. But I wonder where the Petrel goes next?
@Brian Anderson
“The sequence of BTO pretty clearly illustrates that the aircraft crossed the 2nd arc from outside to inside, and then again from inside to outside. The first crossing can be estimated at about 19:53”
Brian that’s an interesting statement. Can you say a little more about that? Is there a path (eg; Victor’s for VOCX to NZPG?) that shows the Arc2 crossing behavior you are talking about? How can the first crossing be 19:53 when 19:41 was the Arc2 time?
@DrB
RE: “If I understand correctly, then the problem is the initial right turn followed by a left turn in the opposite direction, since normally all the turns would be in the same direction. Is that correct?”
Yes; if a hold is entered into the FMC at a waypoint or at PPOS along N571, the default pattern would be right-hand with 1.5 minute legs (above FL140). The inbound course would be aligned with the direction of flight along N571 (eg 296°M between MEKAR and IGOGU) and the aircraft would enter the hold via a ‘direct entry’. In other words, the aircraft would simply turn right and enter the hold without the need for a parallel or offset entry and would continue making right turns in the hold.
The FMC default values for inbound course, turn direction (ie right or left) and leg time/distance can all be overridden by the pilot, which may affect the type of entry that is flown. For example, Mick Gilbert suggested a hold at PPOS where the inbound course is reversed as shown in his diagram. Here, he depicts an offset (or ‘teardrop’) entry into the hold where the aircraft initially turns right and flies an offset track for a short time before turning left to enter the hold.
RE: “I am not generally a fan or more complex solutions, but what if there was a right SLOP followed by a PPOS Hold so that the subsequent racetrack is to the left side of the offset track and is followed in a counterclockwise direction with all racetrack turns being to the left?”
That’s certainly possible, but I question the logic. If the intention of the SLOP was to avoid traffic on N571, then a left-hand hold would put the aircraft head-on to any west-bound traffic on the airway.
Woops, a little typo pointed out by TBill.
The first crossing was of course 19:41. The time 19:53 is my estimate for the time at which the track is closest to the satellite.
@ Andrew,
Consider a RIGHT-turn clockwise Hold after the SLOP was established. With the default racetrack parameters, that avoids any collision risk because one would never be closer than 12 NM to N571. In this case the 18:40 phone call would occur on the southbound leg on the easternmost side of the pattern.
What is your estimate of the total time for one lap of the racetrack pattern (assuming default parameters), and what is the relative length of the “parallel” (to N571) sides to the “perpendicular” sides? Are they equal, creating a “square” pattern?
I will see if a match to the 19:41-00:11 data is possible for a Hold exit in the vicinity of NILAM with a variable delay time. It may be possible to determine if a significant Hold duration occurred there.
Brian Anderson says:
“The time 19:53 is my estimate for the time at which the track is closest to the satellite.”
What is your position for the point of closest approach to the satellite at 19:53 ?
@Ge Rijn
You said, “… the final similator point shows an attitude similar to a glide-attitude.”
Does it?! You need to bear one thing in mind when interpreting the sim data; for data points that show evidence that the airplane has been “dragged” (viz all of them except 5N); the attitude, heading and airspeed data most likely relates to what the airplane was doing before it was dragged rather than what it was doing at the point it was dragged to. There is also the matter that some parameters, that include but may not be limited to, vertical speed and pitch, have “transient” values that are probably not wholly relevant to the airplane’s condition either immediately before or after being dragged. Accordingly, drawing an inference along the lines that you have done is highly conjectural.
Bearing the conjectural nature of interpretations in mind, the data for the final data point, 45S2, shows a ground speed of around 195 knots and shows that the airplane’s altitude had been adjusted down from around 37,500 feet prior to the save. Ground speed is the same as True Airspeed with zero winds. A TAS of 195 knots at 37,500 feet under ISA conditions is around 104 knots Indicated Airspeed; that is well below the stalling speed of the airplane and most assuredly not what you would expect to see in a stable glide.
@DrB
@Andrew
DrB, the times in seconds in that updated diagram are roughly what you’d see for a standard hold; 1 minute and 30 second inbound and outbound legs and a smidge under 3 minutes for each turn for roughly 9 minutes to complete one hold. That’s all based on a hold being executed at FL350 in ISA+10 conditions at a weight of 200T, with a 15 knot wind (direction doesn’t matter). The holding speed would be 253 KIAS, or 441 KTAS.
@ventus45,
>What is your position for the point of closest approach . .
Ha !! That is the $64,000 question. If I knew the answer to that we would have solved this mystery about 3 years ago.
@Mick Gilbert: I don’t think you have interpreted the speed data correctly for the 45S2 data set. If you look at the video I provided in the simulator post, I show how the plane was likely flown between 45S1 and 45S2. At the end of the video, the altitude and speed are falling, but still well above stall. (This business about optimum pitch for a long glide is a far reach.) The video ends with a speed of about 180 KIAS. I then created a second video showing how the plane would have flown after the manual altitude change but with no pilot input during the simulation. The second video starts at about the same 180 KIAS.
With Regard to the proposed NILAM hold.
(https://www.dropbox.com/s/8pvormvgea9tw93/Hold%20at%20Present%20Position%20with%20Reciprocal%20Inbound%20Leg.pdf?dl=0)
Could it be, that the “hold” was planned to be conducted safely BELOW any possible traffic on N571, and, out of sight of all PSR’s ?
An examination of http://aip.dca.gov.my/aip%20pdf%20new/AIP%20SUPP/AIPSUPP%20200836.pdf and http://aip.dca.gov.my/aip%20pdf/ENR/ENR%203/ENR%203.3/Area%20Navigaton%20Routes%20(RNAV).pdf suggests the route is for Upper Limit FL460 lower limit FL275 with lateral limits of 20 Nautical Miles.
This would imply, that the only way to plan a “safe” hold anywhere near NILAM would be below FL275.
Perhaps the theorised descent post MEKAR was specificlly for this reason ?
This would require recalculating the TAS for the hold at lower level(s), that might give a better fit to the BTO/BFO’s ?
In which case, the fact that an apparent left hold circuit was established could mean that the intent initially was to hold there, below FL275, and below all PSR radar coverage, to conduct negotiations, then return to KL (WMKK) ?
See https://www.dropbox.com/s/dnedqibzsz668aw/Dropbox%20-%20Nilam%20Hold%20-%20.gif?dl=0
I forgot to add the frequencies (P) 133.4Mhz, (S) 132.55Mhz.
Do they match with the frequencies found in the simulator files ?
@DrB
RE: “What is your estimate of the total time for one lap of the racetrack pattern (assuming default parameters), and what is the relative length of the “parallel” (to N571) sides to the “perpendicular” sides? Are they equal, creating a “square” pattern?”
As Mick Gilbert stated, the total time for a complete hold is about 9 minutes; 1½ minutes for each ‘straight’, and about 3 minutes for the 180° turn at each end.
The length of each straight is 10.7 nm, assuming a holding speed of 253 KIAS (441 KTAS) and a 14 kt headwind component during the inbound leg of the hold, resulting in a groundspeed of 427 kt on the inbound leg. The diameter of the turn at each end of the hold is 13.1 nm.
@Victor
Without knowing what the corresponding data sets looked like for those video recreations it’s difficult to know whether they match the 45S1 and 45S2 data but I think that I understand what you’re getting at. If the altitude of the simulation airplane is altered in the Map screen then IAS is the constant and the simulation program makes the necessary adjustments to the TAS and Ground Speed values; I can see how that would work.
Do you recall how you derived ground speed from the data?
So glad to see the geometry and timing of the course between 18:25 and 19:41 being discussed thoroughly. The ideas I had about this have all been brought up by others but many have I think, not been discussed in quite these terms before.
East of Sumatra, altitude is key because I believe the mountains create a radar shadow which shields from the facility at Lhokseumawe. That may or may not be material to any choice of course.
@DrB
@Andrew
@Mick
@Ventus
Re: avoiding air traffic on N571 (UAE343), one piece of data we apparently do not have (but ATSB may know) is the path of UAE343. Seems like UAE343 was right on N571, but the FlightAware *estimated* flight path seems to show UAE343 offset approx. where we think MH370 was. Joseph Coleman had posted the Flight Aware data a few blog articles back.
Just some raw thoughts on a maybe (??) possible route between 18:22 and 19:41;
Between the cellphone contact at 17:52 and last radar contact at 18:22 the plane covered ~200Kn.
The distance between last radar contact and LAGOG is also ~200Nm which makes an hour flight reaching LAGOG at ~18:52 from 17:52.
That leaves 51 minutes till 19:41 to produce the 2d arc handshake. Which makes about ~350Kn (if assuming the overall speed of the plane stayed more or less the same).
Assuming MH370 flew straight on along N571 (with maybe an offset and descent) after ~18:25 till LAGOG and made the FMT shortly after LAGOG towards the SIO heading/tracking 96E, I see (quite roughly) after 51 minutes (and ~350Kn) with FMT just past LAGOG, MH370 would produce the second handshake at 19:41 and pass the second arc outside-in around ~0/93E.
Could this make any sence? Just had to share.
@Rob
A very strange statement from Rob: “That’s still your problem, you simply refuse to accept that the Captain did it.”
You sound more than a little confused.
I am on record as authoring a paper together with Victor entitled “Possible Flight Path of MH370 towards McMurdo Station, Antarctica” by Victor Iannello and Richard Godfrey dated 24th August 2016 and published on 25th August 2016 (please see linked .pdf). This paper connects the Captain’s home simulator with the flight reconstruction of MH370. Of course, this is only circumstantial evidence, so I wanted to back it up with evidence from the floating debris analysis, which you wrote off without … even remembering you wrote in the conclusions of your post dated 6th December 2017 at 10:43 am “Drift studies can be consigned to the paper bin.”
https://www.dropbox.com/s/au5w29f7eyescei/2016-08-25%20MH370%20Path%20Towards%20McMurdo%20Station.pdf?dl=0
..I switched Kn and Nm some times I see. It’s all Nm ofcourse..
@ Mike Chillit
https://mobile.twitter.com/MikeChillit/status/938154668265484288
The graphic shown by you, in the tweet above, is a copy of Fig.3 from the paper “Possible Flight Path of MH370 towards McMurdo Station, Antarctica” by Victor Iannello and Richard Godfrey dated 24th August 2016 and published on 25th August 2016 (please see attached .pdf).
https://www.dropbox.com/s/au5w29f7eyescei/2016-08-25%20MH370%20Path%20Towards%20McMurdo%20Station.pdf?dl=0
You claim, in the tweet, that we copied the location in Fig. 3 from your book “The Plane at Batavia” published 9th May 2017.
Since we published 9 months earlier than you, Mike Chillit, I conclude that if anyone copied, then you copied from us.
This is not the first time we have caught you plagiarising.
It is a shame that Ben Sandilands has passed away, otherwise he would have another Chillit story about plagiarism.
@Ge Rijn
This falls under my category #3 the pilot making maneuvers at speed prior to 19:41. The reason I assume Victor is not endorsing that idea (I had suggested out towards DOTEN per Sim cases) is fuel consumption considerations requires a Hold prior to 19:41 to save fuel. There is an exception is that we could say Victor’s Hold to save fuel actually occurred later in the flight, after Arc5. But most of the modelers have been trying to match the path assuming inactive pilot after 19:41.
If the pilot was inactive, Victor/Richard may have nailed it with their McMurdo path.
@Andrew,
@Mick Gilbert,
Thank you for being patient in answering my many questions on the possibility of a Hold near NILAM.
I will model the following cases to see if the CBTOs/CBFOs can be matched from 18:22-18:41:
1. A right SLOP followed by a direct entry, right-turn Hold. The right SLOP and right-turn Hold get 9M-MRO completely away from N571, so no altitude change would be needed to reduce collision risk.
2. An offset/teardrop entry, left-turn Hold with no preceding SLOP. I will also explore the possibility of a simultaneous descent (to reduce collision risk), as suggested by ventus45.
In both cases, I will use the default 1.5-minute straight legs, and I will look at both a NILAM fix and a PPOS fix. We’ll see if any of these can match the satellite data.
Case #2 above seems pretty unlikely to me because it is more complex in that it does not use the default entry method, nor the default turn direction. The simplest method to loiter there seems to me to be the default Hold pattern at PPOS coupled with a descent below FL280, but this alone cannot match the CBTOs/CBFOs. So either there was no Hold at all near NILAM, or there was a combination SLOP and Hold, or there was a complex Hold.
@all
It is getting pretty dull in here. Or is it just me?
At Don, sorry but Ami is adamant about the big foot photos.
Question for the group:
We know of several studies and papers on MH370 debris drift analysis predicting a POI between ~S25 and S40. Richard Godfrey (S30, Feb 2017 and other dates) and other IG members have published work in this field (Henrik had an early paper suggesting S34). CSIRO has published 4 papers, most recently suggesting ~S35. Chari Pattriaratchi had the earliest estimates I know of (first predictions predating the flaperon, now predicting ~S32), Metron paper (S34)dated August 24, 2015, Glomar/ECMWF paper, V. J. García-Garrido1 et.al. November 2015, and several others. The question is: Of all these studies, do any of them stand out today as the most useful in terms of providing advice on the best place for OI to look?
@ALSM
Where is Henrik today? I think I pissed him off and he went away.
@TBill
On your comment; ‘The reason I assume Victor is not endorsing that idea (I had suggested out towards DOTEN per Sim cases) is fuel consumption considerations requires a Hold prior to 19:41 to save fuel.’
I think that’s my problem. I don’t see why it would save fuel to loiter around for some time or put the plane in a racetrack o.s.l.t.. And I don’t see any logic in actions like that from a PIC between 18:22 and 19:41.
I just think this time and distance of 79 minutes and about~550Nm (just raw estimate..) has been flown as straight and simpel as possible.
I don’t see why flying a certain distance and time in circles would save fuel against the same distance and time flown straight.
I rather think flying around in circles uses more fuel.
@ALSM
My preference is for a MH370 End Point around 30°S for the following reasons:
1. 30°S is a good fit to the satellite data for a LNAV flight path to an ultimate waypoint of YWKS or 40S 100E.
2. 30°S is a good fit to the fuel range for a LRC speed schedule at around 35,000 feet.
3. 30°S accounts for air temperature and winds predicted en route.
4. The drift analysis points to around 30°S near the 7th Arc, which is a good fit to the average historical drift data from the GDP Programme and all the locations of the 25 floating debris finds (majority flat panels with little windage).
5. The aerial photographs provided by the RNZAF on 29th March 2014 point to around 30°S, as Victor has demonstrated above.
@ALSM
Not suprising I preferre Pattriarachi’s ~32S I think.
Mainly based on arrival times of the flaperon and Roy-piece (both with plenty barnacles).
– 30S and more northern latitides are more problematic imo because general arrival times of the flaperon are predicted (far) too soon before the discovery time.
-~35S has a kind of same problem but with the Roy-piece. This model imo indicates an arrival time after it was found.
– I think therefore it’s a better bet to chose the middle at ~32(.5)S.
This can explain both arrival and discovery times better and can fit all other data too.
– The flaperon-like ‘blue panel’ image was taken at -32.476367/97.824467.
@ALSM
To add; and I think all reverse drift studies (like Geomar) can be discarded. I mean the ones that take the found debris locations as a starting point. Those switch reality asif the plane crashed there, resulting (ofcourse) in huge (im)possible crash areas that have no value at all in locating a specific crash area.
@ALSM
I am not qualified to answer who has the best drift model, but I sure do feel like 32S and vicinity is being avoided due to degree of difficulty of Broken Ridge region.
I’d like to the planned search zone go a bit wider than 32.6S to 32S at least.
@Ge Rijn
To add; and I think all reverse drift studies (like Geomar) can be discarded. I mean the ones that take the found debris locations as a starting point. Those switch reality asif the plane crashed there, resulting (ofcourse) in huge (im)possible crash areas that have no value at all in locating a specific crash area.
That is a ridiculous concept. Forward drift studies only show what is possible not what is probable.
@DennisW
IMO this is the basic flaw in those reverse drift studies; they switch reality asif the plane crashed where the debris was found.
They create such huge ‘possible and probable’ areas that the meaning of ‘possible and probable’ doesn’t have any meaning anymore regarding efforts to define a specific crash area.
Forward drift studies on the other hand show what is possible and probable if you take the Inmarsat-data and found debris locations and times as facts.
@DennisW
To show you an illustrating example of a reverse drift study (á la GEOMAR, @Brock McEwen and others):
http://thehuntformh370.info/content/location-mh370-reverse-drift-study-based-debris-found
Please tell how this kind of studies contribute to define a more specific crash location.
@Ge Rijn
I am not going to even bother discussing it with you.
There are two areas of high probability that have been previously outlined in detail.
Having come round below Sumatra and turned up at Cocos Islands towards Java MH370 ditched short at between 9°S to 10°S near the 7th arc.
There was a possibility of disruption on the flight deck before Cocos Islands.
This could have meant there was no turn at Cocos Islands leading to overflight and continuing Southeast to around 21°S to 23°S near the 7th arc.
Two drift studies undertaken by professional organisations in independent countries free of confirmation bias gave a high probability MH370 was between 8°S and 23°S at the 7th arc.
In addition drift trajectories from ATSB buoys released much further North have washed up in West Africa where MH370 debris has been found.
@TimR: The two drift modelers that I believe have studied this problem more closely than anybody else are Chari Pattiaratchi and David Griffin. Based on their published results, the timing and location of the recovered debris are not consistent with an impact along the 7th arc at latitudes as far north as 23°S.
Re: ATSB – Operational Search – 3 October 2017, page 101, Figure 75.
Figure 75 is virtually illegible.
Was Figure 75 ever published in any legible detail ?
Does anyone have a better image, and / or the coordinates of the grids ?
@Victor
David Griffin. Based on their published results, the timing and location of the recovered debris are not consistent with an impact along the 7th arc at latitudes as far north as 23°S.
In CSIRO study Part 3 Griffin stops modeling above 28S. I have seen nothing from CSIRO discussing latitudes higher than 28S.
A tiny amount of debris from the preferred location of 35S does reach WA. A degree above or below 35S shows substantial debris reaching WA. As implied the 35S location is quite unique because of the claimed narrow Westward current there.
At 30S and above no debris is shown reaching WA.
@DennisW: I was referring to the timing of the arrival of debris in Eastern Africa.
@TimR
On your comment;
‘Two drift studies undertaken by professional organisations in independent countries free of confirmation bias gave a high probability MH370 was between 8°S and 23°S at the 7th arc.’
Imo you give a good example of the value those reverse drift studies have in predicting a ‘high probability’ crash area. The distance between 8S and 23S is about 1700km.
Without doing any drift study at all it’s save for anyone also to predict a crash area between 40S and 25S or between 23S and 38S.
It only has no value at all in determining a more specific crash area useable to find the plane.
@TimR
To give an illustrating example:
It’s about the same as stating; ‘after a deep independent scientific study we have concluded there is a high probability the plane crashed somewhere between Vancouver and Los Angeles’.
@DennisW
“I am not going to even bother discussing it with you.”
That’s a pity. You probably have no good arguments left against what I stated. And your demeaning attitude shows again. Highly productive..
@Ge Rijn
You probably have no good arguments left against what I stated
What you stated goes against the analysis that was done relative to AF447 as well as the work of Geomar and other highly qualified scientists. I am in no position to be critical of anyone’e drift analysis, CSIRO, Geomar, Richard Godfrey’s,… Not my bag, and not your bag either.
What is my bag is how analytics are constructed. Forward drift analysis requires the arbitrary selection of starting points. A much earlier private discussion I had with Hendrik relative to this approach did not end well either. Ask yourself a simple question. What if the Inmarsat data did not exist? How could you possibly do a forward drift analysis? By using the Inmarsat data you fall into the well-known anchoring bias trap. That is allowing other data (which is irrelevant to how debris drifts) to influence your analytics. By ignoring the Inmarsat data in constructing a drift model you are approaching the problem in the “purest” sense possible. Subsequently using the results of this pure model along with the Inmarsat data is the best approach. You are not allowing what you think you know to influence the model. This is what Geomar has done as well as others using reverse drift analytics.
We (you and I) have had this discussion before, and it also did not end well, and that was my reason for not choosing to discuss it any further with you. You simply are clueless relative to the benefits of using an unbiased problem statement. Having said that, I am very fond of Richard’s work relative to the debris finds. Doing reverse drift analytics is beyond the ability of anyone but very qualified specialists. Richard’s work is as good as it gets relative to forward modeling (my unqualified opinion).
@DennisW
I rather like to have an open discussion with you. I know we have clashed before but we’ve also found some mutual agreement and respect on some issues imo (at least I have).
AF447 is quite different. Debris was spotted in a few days after the crash. Then reverse drift analyzing has its value cause the distance is still short. It’s like VictorI has done with the debris mentioned in this topic.
Over thousends of miles and 1 1/2 year later this kind of drift modeling becomes useless. It doesn’t take a qualified specialist to understand this.
I totally agree forward drift analyzing requires an arbitrary selection of starting points. Indeed if the Inmarsat-data did not excist it would be impossible to do a forward drift analysis. But the Inmarsat-data do excist. It’s all we have, and the found debris.
That’s why I put in the remark; ‘if we except the Inmarsat-data and found debris (locations) as facts’.
The ‘pure’ reverse drift studies (like GEOMAR et al) have proven to be usefull only to show that they are not able by far to predict a specific crash-area. And that has been obvious from the start (to me at least) for they shift reality asif the plane crashed where debris was found. Turning increasing ‘chaos’ to the East instead of the West (where the plane came down).
No qualified specialist is needed to see this either imo.
If a qualified specialist can refute my arguments I’m glad to hear and I stand ready to correct anything I said on this issue.
I also highly respect the work of Richard Godfrey, Pattiarachi and others who did their work on forward drift studying (and those who did reverse drift studying too).
I’m a generalist taking all information into account within my capabilities and drawing (temporarily) conclusions from them.
That’s all. Good day to you.
correction/clarification; ‘ Turning increasing ‘chaos’ to the East instead of increasing ‘chaos’ to the West (while the plane came down in the East along the 7th arc).
@Ge Rijn
Over thousends of miles and 1 1/2 year later this kind of drift modeling becomes useless. It doesn’t take a qualified specialist to understand this.
I suppose the Geomar people did not know they were wasting their time.
The degrading effects of distance and time have the same influence regardless of the direction of the model. In physics this is known as T-symmetry. Very few physical processes (all at the sub-atomic level) are not T-symmetric.
Something else. The latest CSIRO/Griffin study did not use actual historical buoys that landed on WA shores over 20 years but virtual ‘buoys’. Those virtual buoys were never found on Australian shores till now.
This imo down-grades their predicting value regarding their latest forward drift study.
Bold statement again I realize but nothing has been found on Australian shores to date to support their study.
@DennisW
‘I suppose the Geomar people did not know they were wasting their time.’
I would like to hear their opinion now. They’ve been quiet since. I think they know by now they have wasted their time. It would be nice if one them had the nerve to admit it on this blog (or somewhere else). Or to defend their case. They never did after their published study.
I don’t expect any reply from them. Which would be telling imo.
So I ask; GEOMAR respond!
@DennisW
‘The degrading effects of distance and time have the same influence regardless of the direction of the model’
True, but the starting point of the direction in the model makes all the difference in which way the degrading effects will cumulate.
I think debris drift was discussed ad nauseam, as @ALSM likes to say. There is nothing or little new to add at this time. I am only surprised that a lot of contributors here only write their comments, but do not read previous comments of others.
Once again, GEOMAR study, which resulted in origins near Indonesia, was conducted without taking into consideration wind. MeteoFrance study included several scenarios: the windless scenario was consistent with GEOMAR, but not those, which included wind. For some reasons people do not bother to read details. ADRIFT does not consider actual conditions, and it is useful only for general understanding of flow patterns. I think it can be discarded. Other specific comments:
@Ge Rijn,
Re: “Not suprising I preferre Pattriarachi’s ~32S I think.”
Based on the arrival time of the ‘Roy’ fragment, yes, this is probably the best origin. My study is in a good agreement with Pattiaratchi’s in this regard. However, you are totally ignoring the aerial search and barnacle analysis. Both of them make north of 30S more likely. 28-30S is the most ‘balanced’ in my opinion, but 26-28S is also possible.
@Dennis,
Re “That is a ridiculous concept. Forward drift studies only show what is possible not what is probable.”
That is false. Forward drift studies address probabilities by screening various locations. On contrary, backward drift studies are subjected to bifurcations, so this factor may corrupt statistics, unless properly treated. We have discussed this. An additional drawback of a reverse drift study is unknown real arrival time, taken as start time in a reverse model.
@TimR,
Re “Two drift studies undertaken by professional organisations in independent countries free of confirmation bias gave a high probability MH370 was between 8°S and 23°S at the 7th arc.”
False. Did you read these studies? Did you follow multiple discussions at Victor’s and JW blogs? If no, then please read. If yes, do not spread false information.
@IG,
You guys sound like a broken record. Most of this stuff, including floating debris photographed by RNZAF Orion, was discussed in great details many times, in particular at JW blog.
@Ge Rijn
I think they know by now they have wasted their time. It would be nice if one them had the nerve to admit it on this blog (or somewhere else). Or to defend their case. They never did after their published study.
As far as I know nothing has been discovered that refutes the Geomar study. The ATSB search failure only reinforces it. What do you think has happened since the Geomar study was published that invalidates it? You simply keep digging a bigger hole for yourself.
What I would really like to see is an expanded study by Geomar that includes all the verified 9M-MRO debris. Their (Geomar’s) work only used the flaperon find. I am sure their study would undergo substantial refinement if all the debris were now included.
The greatest benefits to a resumed search by OI would be IMO:
1> An analysis by Metron
2> A new Geomar study using all the debris
As it stands, I don’t believe it is prudent to resume the search. Certainly if it were my money, I would not be putting AUV’s in the water at 35S.
@Oleksandr
That is false. Forward drift studies address probabilities by screening various locations.
Funny shit. Unless the entire ensemble of possible starting points is analyzed their is no way to assign a denominator. You might be able to say something about the relative probability of the starting points you selected, but that says nothing about the absolute probability of any particular starting point.
@DennisW @Oleksandr
If GEOMAR would include all found debris the ‘chaos’ would only increase more, eading in half of the Indian Ocean as a possible crash area. Then you’ll have ~25 different crash sites along Eastern Africa and its islands, reversed back towards the fast east Indian Ocean between ~8N and ~45S.
Completely useless outcomes.
A new GEOMAR study on the same premises is completly useless. You just fail or ignore to see this.
@Oleksandr
The above is the basic flaw that reverse drift studies like GEOMAR are ignoring imo.
ADRIFT considers about ~60% of their drifters to be undroged in general after some time. Therefore they state they make a reliable estimate between only water-current moved buoys and those that were also been influanced by wind.
The aerial search is very much doubted in its accuracy. And I agree on this for none of the objects photographed and published were ever recovered as far as we know.
The barnacles enigma is only focussed on the flaperon. This has many possible explanations imo. In this regard the Roy-piece is more telling then the flaperon. Much larger barnacles f.i.
I stay for the time being that ~28S till ~30S is too far north. Arrival times especially the flaperon are way too soon and arrival of debris on Madagaskar/Mozambique shores is way too soon also before discovery.
@Oleksandr
An additional drawback of a reverse drift study is unknown real arrival time, taken as start time in a reverse model.
Unknown arrival time also corrupts the conclusions of forward analysis unless the discovery time is earlier than the forward analysis predicts. Late arrival times are essentially meaningless.
correction again.. ‘predicted’ arrival times..
@Ge Rijn
Arrival times especially the flaperon are way too soon and arrival of debris on Madagaskar/Mozambique shores is way too soon also before discovery.
You are confusing arrival times with discovery times. The flaperon could have arrived in the vicinity of ReUnion much earlier than when it was found. In fact, there are reports of the flaperon sighting in May.
@DennisW
Much earlier? If this is true (I won’t argue your May suggestion) we are talking about two months earlier but not a year or more earlier as those drift studies predict starting from above 30S.
@Ge Rijn
cut-pastes from Richard’s paper below:
From a starting point on the 7th Arc at 30°S, the simulation shows debris arriving south west of Reunion after 487 days. This fits the timeframe of the Flaperon find after 508 days.
From a starting point on the 7th Arc at 32°S, the simulation shows debris arriving a long way south of Reunion after 487 days.
@Ge Rijn
If GEOMAR would include all found debris the ‘chaos’ would only increase more, eading in half of the Indian Ocean as a possible crash area.
The Web is full of opinions on why more data is not better. The reality is that all of these opinions are wrong. Dr. Kalman pretty much proved that any data is useful when properly weighted. More Geomar data can be intelligently combined to actually greatly narrow the probable terminus.
@Dennis,
Re: “You might be able to say something about the relative probability of the starting points you selected, but that says nothing about the absolute probability of any particular starting point.”
Correct. It’s all about relative probability. What do reverse drift models tell you? Let’s take MeteoFrance plots, for example. Most of the shown particles did not come from the 7th arc. Would you conclude that the origin at the 7th arc is
extremely unlikely? Or would you define a narrow band to estimate relative along-arc probability?
Re: “Late arrival times are essentially meaningless.”
Absolutely no. They are very helpful to eliminate some origins.
@Dennis,
Re: “In fact, there are reports of the flaperon sighting in May.”
My model shows elevated probability of the arrival of the flaperon, compliant with De Deckker’s analysis of temperatures, in May-July 2015 (origin 28S). Do you have any reference to the sighting in May?
@Ge Rijn,
Re: “Arrival times especially the flaperon are way too soon and arrival of debris on Madagaskar/Mozambique shores is way too soon also before discovery.”
and “…we are talking about two months earlier but not a year or more earlier as those drift studies predict starting from above 30S.”
1). You are always forgetting that you are looking at a statistical model, not probabilistic.
2). Peak arrival probability of the ‘Roy’ fragment falls on the period March-May 2016 (in accordance to my model), subject to the location of the origin. Isn’t it in a good agreement with the discovery time?
3). The most likely arrival of the flaperon-like object, track of which would be consistent with De Deccker’s SST, was in May-July 2015. Also not bad.
Yes, there should be a plenty of objects in Madagascar, which arrived before the flaperon. We still don’t know. The authorities seem to be reluctant to organise a proper search.
Speaking about MeteoFrance… It’s a miracle that @Rob did not read it carefully.
@Ge Rijn,
Re “The flaperon-like ‘blue panel’ image was taken at -32.476367/97.824467.”
Do you assume it did not drift during 3 weeks, when you use this fact to support an origin in the proximity to 32S?
@Oleksander
Do you have any reference to the sighting in May?
I could not find it with a quick search, but a woman and son reported it. It was also stated in the same article that it was used by a local fisherman as a table. All of this is hearsay, of course.
@Oleksander
Absolutely no. They are very helpful to eliminate some origins.
A late discovery cannot by cataloged relative to actual time of arrival. Because something is found in month N does not mean it was not in the area at month N-3. Discovery and arrival are very different things.
@Dennis,
I meant that if an object was discovered at a location, arrival to which was modelled significantly later, then respective origin can be eliminated.
@Oleksandr
Agreed.
Regarding the flaperon date: Blaine and I discussed this today. He confirmed that rumors about a possible earlier sighting were not true. It beached circa July 28, 2015. He went to Reunion and interviewed everyone concerned. I believe his report.
Regarding the RR debris, it was first found in December 2015, but not picked up at that time. It was “rediscovered” 3 months later (March 22, 2016) in the same area (Mosselbay SA).
@ALSM,
Re: “It beached circa July 28, 2015”
1. It seems you confuse reporting/sighting time with beaching time.
2. How do you derive that July 28 is the beaching date? How does it follow from your discussion with Blaine?
3. Why the flaperon was covered with sand?
I don’t think it is possible to establish exact arrival date of the flaperon. Long time ago I analysed storm events and tides at Reunion, but I did not find any obvious correlation. This is in contrast to Inhambane fragment, which likely beached a day earlier due to a favorable combination of high tide and storm.
@Oleksandr
Perhaps I am mistaken but seem to recall (sat photos?) of earlier arrival of flaperon.
@TBill
I am OK with the late July date on the flaperon. Whether it was a month or two earlier is not really relevant in the grand scheme of things.
@DennisW
On your cut-pastes from Richard’s paper.
I didn’t and can’t say he is wrong on the flaperon starting at ~30S.
I only can say CSIRO/Griffin came to same conclusions but then from ~35S.
Which makes it hard to tell which model is the most correct.
I think it could be a save bet to take the middle at 32.5S.
And I should have written ‘up to a year or more’ after 30S.
When I do simulations in ADRIFT from ~30S, debris starts to arrive at Reunion in ~12 months. At ~28S this has become in ~7 months.
At ~32S this is about ~15 months. I know it’s a rather crude method but to me this gives good indications based an actual drifters.
@Oleksandr
On your comment:
‘Re “The flaperon-like ‘blue panel’ image was taken at -32.476367/97.824467.”
Do you assume it did not drift during 3 weeks, when you use this fact to support an origin in the proximity to 32S?’
97.824E is about 120 miles from the 7th arc at that latitude.
Oleksandr:
Stop insulting people. Of course I know the difference between arrival and find dates. Did you read my post? It clearly stated the date beached and how we know it was the date beached and found a day later. If you don’t believe Blaine is capable of going there and interviewing all parties concerned to reach his conclusion, why don’t you go there yourself and ask arround?
@Ge Rijn
Hard to say where to start really. My guess is OI will follow CSIRO’s recommendation initially.
@ALSM,
Fortunately I did not insult as many people as you did. Stop your personal attacks. Stop spreading false statements. Stop fitting data into your narrative. Stop mispresenting your assumptions as given facts. Everyone knows that you are getting anxious when you can’t defend your posts.
@Ge Rijn,
“97.824E is about 120 miles from the 7th arc at that latitude.”
OK. I did not check the distance, but I trust you did.
@Ge Rijn, @DennisW: All indications are that OI will start with the 25,000 sq km area and then proceed north. Chances are that both the 32.5S and 30S latitudes will be covered. Only OI knows how far north they are willing to commit resources to extend the search, which will be influenced by weather. By progressively working north, they are scanning the areas with the most adverse weather during the most favorable part of the year.
I think we have to just patiently wait to see what is covered after the search starts.
@all
Just a heads up so you will be able to discuss the next phase of the MH370 search problem with me. Plus that, it will be good for you.
https://techyhealth.com/artificial-intelligence/free-training-courses-machine-learning-artificial-intelligence/
Following on from Victor’s excellent posts on “Possible MH370 Paths along Great Circles” and the “Possible MH370 Debris Seen in Aerial Search in March 2014” originating at 29.7°S:
Here is a link to a presentation with the results of over 700 reconstructed MH370 flight paths showing the most likely MH370 end point is around 30°S near the 7th Arc:
https://www.dropbox.com/s/atn7cnx2i27hrgb/Reconstructed%20Flight%20Paths%20-%20Richard%20Godfrey.pdf?dl=0
Here is a link to a presentation with the results of the drift analysis for all MH370 floating debris finds showing the most likely MH370 end point is around 30°S near the 7th Arc:
https://www.dropbox.com/s/clks6522vquagwb/Drift%20Model%20Results%20using%20an%20Independent%20Drift%20Model%20-%20Richard%20Godfrey.pdf?dl=0
The combined data from the Aerial Search, Reconstructed Flight Paths and Drift Analysis adds significant weight to the likelihood of a MH370 End Point around 30°S near the 7th Arc in my view.
@Richard: Thank you, Richard. That’s an excellent summary of a lot of hard work. If the plane is not found in the first 25,000 sq km, the descent/hold scenario becomes more likely, and the 30S latitude definitely justifies special attention.
@All: I’m curious to know what probability the contributors here would assign to finding the plane in the first 25,000 sq km of the subsea search.
@Victor
“I’m curious to know what probability the contributors here would assign to finding the plane in the first 25,000 sq km of the subsea search.”
I gave my number last thread, so I have to try to find. If we say 95% chance it is between 20-40S, and 90% chance it is +-30 nm of Arc7, the amount of Arc7 contained between 32.5-36 is about 10% of the reamining area (since some of that area has been searched). If we say the selected area has 3x more probability as a hot spot, you get maybe: 25% chance.
Not sure I am doing that correctly, but that is my approach.
@TBill: Do you discount the satellite images, which would put the impact at around 35.5S latitude?
@Richard
Awesome work. Much appreciated. Hopefully OI is paying attention, and their contract does not require them to search the CSIRO area first.
@VictorI
The major ‘weak’ points in the latest CSIRO/Griffin model are IMO:
– the area close to the 7th arc has already been searched till ~32.8S
– the CSIRO hot-spots based on their satelite images and small-band ‘drift-stream’ to the east are outside the already searched area till +/- ~20 miles from 7th arc. Imo this falls out of the range of high speed descent/dive impacts near the 7th arc. Not a real problem imo but:
– the satelite images are not detailed enough by far to serve as possible evidence. Then your images and the ‘blue panel’ make much better ‘evidence’ compared to those.
– their latest computer model uses ‘virtual debris’ which predicts many beachings on Australian shores above ~35S. This is not supported by any evidence and contradicts their model in this regard. Nothing has been found to date in Australia.
– their model predicts arrival time of the Roy-piece far too late after discovery time.
Strong point is; they included the leeway and windage of the flaperon based on their findings with an altered original flaperon.
But then I think we can still not be absolutely sure the damage on the trailing edge was caused during the crash. There still is a change it was caused by battering on the rocky shores of Reunion for some time.
IMO Richard’s model has it’s major ‘weak point’ in the delay of arrival times and discovery times at Reunion and Madagaskar. Which Griffin also estimated up to year too soon in an answer on a mail of yours (published on your blog).
And his model shows in general debris passes Reunion well north coming from ~30S.
Strong point IMO; it better explains the Roy-piece and the total lack of found debris in WA.
Based on this I assign only ~10% probability of finding the plane in the 25.000km2 zone and then only in the most northern part of it.
@Victor
“Do you discount the satellite images?”
I suppose I am? I thought it was drift studies that pointed to 35.5S, and also surface search per ATSB? I am assigning 3x “hot spot” probability to the CSIRO-proposed area…I would also give other areas (eg; McMurdo) maybe 2x hot spot rating, amd some areas like 20-25S less chance. Of course it has to material balance to 100% in the end.
to add another strong point imo: Richard’s model uses the Global Drifter Proram data based on actual historical drifters.
I think this is more reliable than virtual drifter models.
@DennisW
Many thanks for your kind comment. I also hope that OI is paying attention and that their contract does not require them to search the CSIRO area first. Time will tell.
@Victor, you ask: “I’m curious to know what probability the contributors here would assign to finding the plane in the first 25,000 sq km of the subsea search.”
If by the first 25,000 km2 you are referring to the search area defined by the ATSB First Principles Review, then apart from the search area you defined around 34.3°S 93.7°E, en route to the South Pole, which is borderline to the previously searched 120,000 km2, then my answer is “close to zero”.
@Oleksandr: “@IG, You guys sound like a broken record.”
My apologies that we are boring you with stuff that you have long since known.
@Richard
As you may have noticed I assign a lot more probability to the area between ~32.8S and 30S. I put my cards on around 32.5S but that’s only arbitrary.
I also hope OI is not restricted to search the CSIRO-area first.
Anyway they are still not restricted by any means for it seems no contract has been signed yet.
And it seems they are heading towards the IO, contract or not.
On a no cure/no pay base they can as well start the search completely independent and make their (financial) demands if they find the plane.
And to mention also; I highly regard your work.
to add.. for OI it could well serve as another trial to test their new equipment and capabilities.
If they succeed they make head-news, settle their name at once and probably make a lot of money on the spot.
If they fail to find the plane, they collect a lot of data and experience useable for the future.
In fact they will loose nothing either way. It would be part of their investment in the future like they’ve done with previous trials.
@Richard G
“…showing the most likely MH370 end point is around 30°S near the 7th Arc:”
All the paths on your page 9 except the last 4 or 5 have acceptable errors. What is special about the Constant Track 173°T case?
@Richard Cole
You ask: “What is special about the Constant Track 173°T case?”
On slide 5 of the first presentation “Reconstructed Flight Paths” you see that 6°N at the 2nd Arc is the only start latitude that fits both BTO and BFO or as I have termed it, is the cross over point. This translates to around 30°S given the other findings in the presentation: CTT or LNAV, LRC or Constant Mach 0.8M, > 33,000 feet etc.
In the second presentation the only end point that matches all the debris finds in both location and timing is around 30°S.
Finally Victor has shown above, that the RNZAF debris sightings fit an origin at 29.7°S.
These 3 independent analyses make a MH370 End Point at around 30°S special, in my view.
Time will tell, if these combined analyses are right.
What is your most likely MH370 End Point?
@Richard
Just have to react on this statement of yours;
“On slide 5 of the first presentation “Reconstructed Flight Paths” you see that 6°N at the 2nd Arc is the only start latitude that fits both BTO and BFO or as I have termed it, is the cross over point.”
I think it is a bold statement to pinpoint 6N. This latitude needs loiters or other hold-ups between 18:22 and 19:41 which make no sence imo. An off-set and (long) descent after 18:25 are quite possible imo but loiters and other complicated manouvres are unlogical and unlikely (imo).
Regarding the flying distances, time required, guessed altitudes and expected general speeds between 18:22 and 19:41 a crossing point latitude just north of the equator is probably more likely imo (~2 or 3N) I think.
Pinpointing 6N requires knowledges of all data (altitudes, headings, speeds etc. between 18:22 and 19:41) which is not possible imo.
@Ge Rijn
I have reconstructed over 700 flights, including all feasible starting points, altitudes, speeds, navigation modes with the respective air temperatures, winds speeds and directions and where appropriate magnetic variations.
My conclusion is that MH370 was at around 6°N on the 2nd Arc at 19:41 UTC.
I accept that this makes no sense to you. Unfortunately that does not mean it could not happen.
@Ge Rijn
cut-paste below from my post on 13 Novemeber 2017
I just weighted my eight computed residual values at 17:07, 18:25, 19:41, 20:40, 21:40, 22:40, 23:15, 00:11.
I agree with your comments regarding measurement errors, and more importantly AES oscillator drift. I was merely looking for another way to estimate 00:11 latitude.
For what it is worth, the weighted integration came to ~11 Hz.
The table of values associated with sampled endpoints on the 00:11 arc relative to IGARI is listed below.
18S – 8.04Hz
22S – 9.04Hz
26S – 10.0Hz
30S – 10.92Hz
34S – 11.79Hz
38S – 12.78Hz
4 degrees of variation per Hz!!
Hence, my conclusion that 30S was the closest fit to a wildly simplistic (and most likely meaningless) analysis. I told you I was not afraid of being embarrassed. To my credit I did not use this information in my list of reasons for picking the 45S 104E path. Likewise, I did not use any geometric algebra arguments (which I still don’t understand completely) which also seem to provide weak support for 30S (and a 19:41 latitude of 6N).
Obviously, I support Richard’s conclusion, which is based on far more understandable (and reliable) analytics.
@Victor. Some weeks back you mentioned that you were working on the “missing middle” – the path scenarios that join up the radar track with the best candidate great circle paths. Although you have described what is involved (fly further W, loiter, then south) I haven’t seen the details and I’m eager to understand what assumptions are required to accommodate observed BTO/BFO pre-1941Z.
@Richard
@Brian Anderson
In the land of the blind the one-eyed man is King (that is a compliment). So let’s go with 6N that says about 93.5E right?
That’s on a line from LAGOG to BEDAX if approached from the West, or also on a line from NOPEK to TOPIN if approached from the East. So we have some options to get there.
Can we combine Richard’s 6N 19:41 judgement with Brian’s speed assessment? Because Brain said the only thing the was missing was the 1941 point. It could imply a faster speed until out of Indonesia air space for a later slow down, I have been thinking is one possibility.
Correction:
From the West the line is APASI to BEDAX (not LAGOG to BEDAX)
@Paul Smithson: You have a good memory. Yes, I worked on the “missing” middle. Here is a summary slide. The sequence events is:
1. Pass MEKAR at FL340 and M0.84.
1. Near the last radar point at 18:22, turn towards SAMAK and reduce speed to 225 KIAS.
2. At 18:37, initiate V/S descent at -2600 fpm to FL200.
3. During 19:40-19:41 interval, VS=-2600 fpm, GS = 430 kn.
4. At 18:44, pass SAMAK at Malaysian-Indian boundary at FL200 and 225 KIAS.
5. At 19:06, enter holding pattern near VOCX.
This is one of many possibilities.
@Richard
Impressive work on the reconstructed flight paths.
You wrote:”On slide 5 of the first presentation “Reconstructed Flight Paths” you see that 6°N at the 2nd Arc is the only start latitude that fits both BTO and BFO or as I have termed it, is the cross over point. This translates to around 30°S given the other findings in the presentation: CTT or LNAV, LRC or Constant Mach 0.8M, > 33,000 feet etc.”
If I read slide 8 correctly there is a wider range of 00:19 latitudes that fit both BTO and BFO data (LNAV, LRC), so there 30S is not unique. Why do you favor CTT?
@TBill,
In my view the precise aircraft track between about 18:28 and 19:41 is virtually indeterminate. However, there are a variety of possibilities to use up the time/distance equation before 19:41. I remember creating some possibilities more than 3 years ago. None help to determine the ultimate location of the intercept with the 7th arc. Rather, bearing in mind the paucity of data available at the time, it was an exercise to try to establish possibilities for the location of the 2nd arc intercept.
If we accept that the 19:41 intercept is at 6N, and there is now a deal of additional information which supports this, then with the speed estimate [average speed to 20:41] it is possible to very quickly determine the southern-most 7th arc location. This point assumes a Great Circle path from 6N, satisfying the speed to 20:41, and terminates at approximately 31.3 degS. There is a particular nav mode required for this of course, but other nav modes will always terminate a little further north [NE] on the 7th arc.
Here are plots of NE to SE independent storm events at Reunion and Inhambane overlaid on top of the astronomical tide (I combined two plots I posted long ago into one):
https://www.dropbox.com/s/w9en2zp7oz12f7h/storms_reunion_inhambane.jpg?dl=0
Wind is sourced from Wunderground (measured); tide is sourced from Marine Jeppesen C-Map.
The situation with Inhambane fragment arrival requires no comments, especially bearing in mind that it was waterlogged on the moment of discovery. However, the flaperon’s arrival date is ‘fuzzy’ due to the multiple preceding storm events and relatively small tidal range at Reunion.
According to Johnny Begue, the flaperon was half buried in sand when it was found:
http://www.dailymail.co.uk/news/article-3179584/Malaysia-certain-MH370-Investigators-despatched-Indian-Ocean-island-prove-six-foot-wing-flap-belongs-Boeing-777.html
http://www.smh.com.au/world/mh370-reunion-beachcomber-johnny-begue-makes-the-find-of-his-life-20150730-giodlj.html
It could potentially be useful if Blaine asked Johny a few right questions, but unfortunately this does not appear to be possible due to a broken line.
@Oleksandr
Here is the reference you asked about earlier.
http://www.therakyatpost.com/world/2015/08/02/mh370-was-flaperon-actually-found-in-may-on-reunion-island/
@Richard,
Re: “My apologies that we are boring you with stuff that you have long since known.”
A major issue is that the IG throws away most of the ideas suggested by others, which do not fit IG’s current narrative (or consensus, if you wish). Then a year of two later you guys present these ideas/discussions as something new.
Re: “My conclusion is that MH370 was at around 6°N on the 2nd Arc at 19:41 UTC.”
This is a well-known thing if you get rid of a single FMT between 18:25 and 18:41 followed by a constant AP mode assumption.
Btw, I just happen to look through old Yap’s paper, I downloaded on June 26, 2014 (unfortunately I did not store the original link):
https://www.dropbox.com/s/tliakuxjg99kpzt/yap.pdf?dl=0
Interesting, isn’t it?
@Dennis,
Thanks for this reference.
That is pretty consistent with what models show.
@Richard Cole,
Re: “What is special about the Constant Track 173°T case?”
To me any (TT, TH, MT, MH) 173° is special. That is because it can be logically explained as a result of a wrong exit from some preceding maneuver (or inability to complete it), which comprised 120° or 60° turns.
@Victor asked:
“I’m curious to know what probability the contributors here would assign to finding the plane in the first 25,000 sq km of the subsea search”
From my drift models based on the flaperon analysis:
– Up to 13 percent assuming DGA’s 18° drift angle;
– Up to 1 percent assuming DGA’s 32° drift angle.
I guess ‘flip-flop’ mode would result in something in between. The probability would be lower if one takes into consideration absence of debris in Australia, but that factor is not quantifiable.
@Richard: You said you thought the probability of finding the plane in the 25,000 sq km was next to zero, except for the impact point along the 180T path. If you include this in the 25,000 sq km area, what probability would you assign?
Here is a small study I did regarding the constancy of BFO’s during various phases of flight. Backdrop – during the 18:40 phone call, the BFO values (51 total) over a period of about 1 minute had an rms value of 1.3 hz.
I used the entire MH371 log (both IOR and POR data), excluding the LGA nd R600 values, but including the pre-takeoff log from MH370. [I applied the channel-dependent BFO offsets determined from the MH370 logs; the MH371 logs include additional frequencies, but no additional offsets seem necessary.] There are many long runs of BFO values. I considered all runs of length 1 minute for which data points exists for at least 45 seconds, with > 10 values. I computed the rms BFO error within each 1 minute run. I divided the flight into 4 phases: ground, ascent, cruise, and descent. Here are the 1st quartile, median, and 3rd quartile points of the 1-minute rms values for each of those phases (units of hz)
Ground 0.7 0.9 1.7 (n=92)
Cruise 0.7 1.1 1.6 (n=144)
Ascent 1.1 1.5 2.0 (n=44)
Descent 3.8 5.8 20.4 (n=60)
It is seen that on the ground and during normal cruise, the BFO values can be quite stable. The ascent phase is also stable because the values included here occured during the climb out of Beijing, when the vertical motion of the aircraft was nearly perpendicular to the line-of-sight to the satellite. Descent, however, is another story – the BFO values bounce all over the place. Note that the descent rate was never as large as -2000 fpm. To get an rms of 1.3 hz, the geometric (not barometric) descent rate must be constant to at least +/- 100 fpm, and probably much better. We do not know the actual descent mode used by MH371. However, suppose it were FLCH (adjust pitch to maintain constant airspeed at idle thrust), one would not expect the BFO values to bounce around so much unless the atmosphere was far from ideal – small-scale variations in temperature, density, etc., all of which conspire to disrupt carefully crafted flight profiles.
Victor has long proposed that a descent in V/S mode (constant vertical speed) could mimic a South heading in level flight during the 1st SATCOM attempted phone call (and specifically, V/S = -2600 fpm.) However, what the manual says what a plane will do and what the plane actually does are two different things. Virtually all evidence that I have found (including crawling through FlightAware logs plus my own monitoring of ADS-B broadcasts from descending aircraft) say that a constant vertical descent with the accuracy required to maintain the BFO error within 1.3 hz almost never happens.
I am happy to be proved wrong, but the burden is on Victor (and any fans of constant V/S mode) to do so.
[Here’s a link to a flight that DID use V/S mode – it got caught in an overspeed. http://avherald.com/h?article=47a7f9c1&opt=0 ]
Richard,
In your document “Drift Results”, you say that you present results from “drogued buoys”. Is that correct? If so, why? Drogued bouys record currents at a depth of 15 meters. They will not provide a proper measure of the motion of debris at the surface, such as flaperons and “NO STEP” items. Or is there something I am missing?
@sk999
I computed the rms BFO error within each 1 minute run
What that does not address is the BFO error due to oscillator drift over extended flight times. Not sure what would be typical for that. Not even sure that “typical” is meaningful.
Do you have any explanation for the existence of channel dependent BFO offsets? Anyone?
DennisW,
Yes, there are oscillator drifts. There are also “drifts” in the form of normal variations due to the plane moving, the satellite moving, etc. I did consider them all, but decided that they were small enough over the course of ONE MINUTE that they could be ignored. Furthermore, even if some exceptional event were to intervene, as long as it were transient, it would be filtered out in the computation of quartiles and medians.
R.E. “channel depended BFO offsets”, I don’t even know what “channel” refers to. Something in the AES? Something specific to Inmarsat? It’s just an empirical result. Zero physical basis. Kind of like how “AI” or “deep learning” algorithms work. Do you have any better insight?
@sk999
I am working on the AI and deep learning algorithms. No insights there as yet.
You said that you applied the “channel dependent BFO offsets” from the MH370 logs so I thought you may have had some insight into what you were talking about.
@sk999
I did get distracted on the Google Tensorflow freebie, but I deny being “deepflakes”.
https://motherboard.vice.com/en_us/article/gydydm/gal-gadot-fake-ai-porn
DennisW,
R.E. “AI and deep learning algorithms”, I admit to having a bias. Ages ago I signed up for a course on AI from Marvin Minsky. Very interesting, but I didn’t have the time it needed, so I dropped it. Fast forward. “Deep learning algorithms” are a subset of AI today. Very impressive accomplishments. However, there is a problem. What are the “errors”? In other words, say your algorithm identifies a “hot spot”. It can never pinpoint the precise location – there must be some search box. How big is the box? Not clear if those fancy algorithms can tell us.
R.E. “I thought you may have had some insight into what you were talking about.” Nope – just doing that AI/Deep Learning thing.
@sk999
I know you have a bias. You probably think Gibbs has all the answers you need (which he did for R3).
I am bias free and will listen to just about anyone. Currently I am most interested in the work being done by Gell-Mann and Peters relative to expected value theory since I am retired and have a portfolio to manage. Financial advisers are basically clueless despite an army of MBA’s in their firms. They often ask me to do discounted cash flow analytics for them. Maybe I should come out of retirement as a quant.
@Victor: You asked, if I include an impact point along the 180°T path, in the 25,000 sq km area, what probability would you assign?
I assign a probability of 14.3% to the ATSB residual search area between 36.0°S and 32.5°S.
By comparison the various Hot Spots range between 6.1% and 10.3%.
@sk999: You ask why I use data from drogued buoys.
This is a good question and much discussed by the experts like Griffin and Pattiaratchi. As you know, a drogue is a form of sea anchor and there have been a number of designs.
NOAA organise the Global Drifter Program (GDP) and state that the drifters in their Surface Velocity Program do not perfectly follow the water column averaged over the drogue depth. For example, water can downwell (sink to great depths from the surface), while the drifter is forced to stay at the sea surface. Also, the drifter can “slip” through the water. The resulting speed of the drifter is thus a combination of the large-scale currents at 15 meters depth, plus the upper-ocean wind-driven flow, plus the slip. I wanted to get as close to the actual surface velocity as possible.
Fortunately, the GDP data includes the date-time when a drifter loses its drogue. In my software, I have a switch to select drogued, undrogued or both drifter data sets. I ran various comparisons for transoceanic tracks lasting between 500 days and 1000 days. I get the best fit with data from drifters with drogues. I will continue to investigate the data from drifters without drogues.
@Oleksandr
Many thanks for reminding me of Yap’s excellent paper entitled “MH370 Path Optimization Analysis” dated 24th June 2014. Yap also wrote another paper entitled “Autopilot Flight Path BFO Error Analysis” dated 24th March 2015. Yap is one of the giants of the IG, who has contributed much to our current understanding of the MH370 tragedy.
I could not have built my MH370 Flight Models without help from Yap and many other IG members, Duncan, Victor, Mike, Don, Henrik, Barry, Geoff, Brian and Sid.
@sk999, @DennisW
“channel depended BFO offsets”
The AES, more specifically the SDU, employs a single modem for all packet data traffic while (R and T channels) while it employs five separate voice modems for the C channels. The reference clock and the IF-RF upconvert chain is common. Precompensation occurs in the upconvert chain.
Within the GES, the Channel Units employ multiple modems where each modem is fixed to a specific receive channel. The reference clock and RF-IF downconversion is external to the Channel Units.
My assumption has been that any channel dependency is attributable to the GES demodulation process.
Each SATVOICE call attempt defined the same ‘called terminal ID’, ie the same SDU modem & onward connection.
Interesting approach to categorise data into flight phases. Transmissions occurring immediately after ToD would be the reference to compare with the proposed 18:39 descent. Later in descent, while in approach control likely to be more complex.
@sk999: Thank you for your thoughts about the BFO and a V/S descent.
Let me start by saying that I have no “burden” to prove anything. I am proposing a possible scenario (a descent/hold) that might explain terminations along the 7th at more northern locations. If evidence is presented that makes this scenario unlikely, then I accept this.
Really, there are two questions:
a) Was MH370 flown with a V/S descent at 18:40?
b) For a V/S descent, do the BFO values have lower fluctuations than for common descent modes?
I believe there are several circumstances in which a pilot may choose to use V/S descent rather than the more common FLCH or VNAV:
1) For small changes in altitude where the “all or nothing” thrust schedules are not appropriate. The vertical speeds using the V/S mode would be lower than other modes.
2) To get back on the target flight path for a VNAV descent if lower than the target path. The vertical speed can be reduced while still maintaining some minimum descent rate until the path is captured and a VNAV descent can continue.
3) When there is a desire to cross a particular point at a particular altitude that is not in the flight plan. In this case, V/S can be used with the “banana” in the ND to estimate where the target altitude is reached.
4) Under conditions of high turbulence. Using a speed controlled by pitch (such as FLCH) will cause swings in pitch due to turbulence. On the other hand, controlling V/S using pitch will result in a smoother flight, as the small variations in airspeed would not be corrected. In fact, the FCOM advises in regards to severe turbulence: During climb and descent, use of VNAV of flight level change may result in excessive pitch changes as the AFDS attempts to fly speed with the elevators. Therefore vertical speed mode (speed on autothrottles) is recommended for climb and descent in severe turbulence.
I think that for MH370, (3) is possible as a reason for the V/S descent, and (4) lends some weight to the assumption that the BFOs would be more stable during a V/S descent. We don’t know why V/S might have been chosen, but we do know that MH370 was not a typical flight.
Another interesting feature of the V/S descent at high descent rates is that the indicated airspeed does not remain constant (unless the pilot uses speedbrakes to introduce drag). For instance, if 225 KIAS is set as the MCP speed at FL340 and a descent to FL200 is programmed at V/S=-2600 fpm, the true airspeed increases from 389 kn and then holds a nearly steady value of about 430 kn, corresponding to a descent angle of about 3.4° (L/D=17), and the indicated airspeed increases as altitude falls. Once the final altitude is captured, the airspeed again reduces to 225 KIAS, corresponding to a true airspeed of 308 kn.
I have some slides showing the TAS, IAS, altitude, V/S, BFO, and other aspects for a descent from FL340 to FL200 that I can share if there is interest.
@Brian Anderson
“This point assumes a Great Circle path from 6N, satisfying the speed to 20:41, and terminates at approximately 31.3 degS”
Thank you for the estimate. What about if the mode is CTH? and of course I am open to pilot input after Arc5. Could that hypothetically “move the chains” a bit further West?
@Victor
I also like V/S mode after Arc5. On the FS9 sim, I don’t use FLTCH much but I am not a real pilot either. But the MH370 pilot would have no need to achieve company policy descent mode out over the Andamans.
@sk999
“Backdrop – during the 18:40 phone call, the BFO values (51 total) over a period of about 1 minute had an rms value of 1.3 hz.”
What about the 23:14 call, I think is more scatter…so what does that say if anything? Also what can you say about the calibration of the 18:40 call from that channel? Do we add “+4” to bring the BFO in line with the normal pings, or do we just don’t know, amd assign +-5 (or more?) that channel.
@Richard G
In response to your post and question.
“.. 6°N at the 2nd Arc is the only start latitude that fits both BTO and BFO”
From the DSTG report the BTO standard deviation is 29microsec = 8.7km (I see you have used 32microsec). The worst BTO error (I assume rms) in your table is 10Hz, which is not statistically excluded by the rather few data points available, so I don’t think BTO error can be used as a discriminator between the cases. As we agree, the BFO errors are getting large for end-points north of 30S.
It’s 500km from the north of the First Principles Area to 29S, so another 25000sq.km. to search even a 50km strip. Statistically, the track statistical significance is quite flat along that 500km line, as your analysis supports. If an extended OI search started with the area around 30S, on the basis of the RNZAF debris sighting, that would seem a reasonable trade-off to me. However, my guess is that OI will not do much outside the First Principles Area.
@Don
My assumption has been that any channel dependency is attributable to the GES demodulation process.
Interesting. That would never have occurred to me.
@Richard Cole said: However, my guess is that OI will not do much outside the First Principles Area.
Based on conversations I’ve had, I believe that OI understands that there is a significant probability that the wreckage will not be found in the first 25,000 sq km area, and is prepared to extend the search north rather than accept defeat. How far north is unknown, but considering that there are debris sightings corresponding to an impact at 30S, I would be surprised if the extended area does not extend north to at least this latitude.
As you say, for LNAV (great circle) paths starting at 19:41, it becomes difficult to use BTO/BFO statistics to discriminate between paths across a range of latitudes. I tried to show this in my last post. I don’t think the significance of this is fully recognized. Unless there is added weight put on additional evidence or assumptions, such as favoring a path towards a specific waypoint, or considering debris sightings, the search should be a continuous process of elimination rather than search by specific hot spots.
@Victor
the search should be a continuous process of elimination rather than search by specific hot spots.
I agree if for no other reason than to be able to accurately catalog the area searched. Jumping from place to place just seems like a bad idea relative to archiving a search history.
I do, however, think OI would be better off starting farther North than the First Principles Area. Perhaps your earlier suggestion of getting the Southern area out of the way as early as possible due to weather considerations is a factor.
@TBill,
The Great Circle path terminates at approx 31.3degS. The nav mode is of course LNAV. As I said, any other modes will terminate further north east along the 7th arc as a result of the strong westerly winds as the track progresses south, or as a result of magnetic deviation, or a combination of both.
@VictorI
“the search should be a continuous process of elimination rather than search by specific hot spots.”
I would think searching/eliminating the CSIRO hot spot areas first could be a good strategy. This is a relatively small portion of the 25.000km2 area. Roughly about 6000km2 between ~34.3S and ~36.3S.
Then move on to your 29.7S area and Richard’s 30S taking a +/- 20 miles of the 7th arc (between 29.5S and 30.5S f.i.) which takes about ~4000km2.
Then there will be ~15.000km2 and more time left to decide where to search next if necessary and less time is wasted (when searching in a continuous process of elimination) on areas where potential probability is regarded low by now.
@DennisW: Yes, I think weather is a consideration for starting the search in the south and working towards the north. There is also the consideration of the Pleiades satellite images, which suggest large, possibly man-made objects were detected near 35S latitude. The ATSB and CSIRO are optimistic that these objects are from MH370. Many of us assign higher uncertainties, but it would be wrong to completely dismiss the significance of these images. For these reasons, I support starting with the 25,000 sq km and working towards the north. That also would capture the 180°T path.
to add; If OI has 90 days of searching time window they have to find the most time-efficient strategy. I think searching the mentioned hot spot areas first would be more time-efficient then moving along the 7th arc to north in a continuous search from the start.
@Ge Rijn: I don’t think the 90-day constraint that was reported in the press is correct. OI is privately owned, and they can continue to search until they decide to stop.
Regarding the channel dependent BFO bias:
I am fairly sure the “channel dependent bias” is an artifact of how the Square Peg Channel Units (CUs) are designed, not the SDU. Here is a quick summary of the channel dependent bias: https://goo.gl/KcAhp7
@VictorI
Yes, that’s a good thing. But if they want to succeed this winter (SIO summer) starting in early January they’ll have the constraint of a ~90 days weather-window to consider anyway I think.
@ALSM
Thanks, nice summary. That (a CU basis) feels good to me, as did Don’s earlier similar comment. The original write-up in Duncan’s blog did not have a causality postulate (or I missed it).
@Ge Rijn: Weather is nicer to the north. After all, 23.4S is the latitude of the tropics.
@VictorI
Yes, I’ve crossed that latitude in WA 6 years ago with Christmas near Coral Bay, 49 degrees C in the shade..
Hope their weather time-window will be till May at least..
They just need some luck too ofcourse.
Dennis:
Back in 2014, I was reluctant to attribute the BFO bias differences to the CUs because in all my experience designing and manufacturing earth station demods, we always used the station reference as the demod frequency standard. That was almost universally a Rubidium or Cesium based standard. If the CUs did use a Rubidium or Cesium based standard, it is hard to imaging how they could end up with BFO measurements differing by as much as 5 Hz from CU to CU. However, over time, I have developed more confidence in the assumption that the SDU Doppler compensation algorithm is the same for all channels, and the CUs are the source of the BFO bias differences. (It is a shame that Square Cube was slapped with a gag order when the official Annex 13 Investigation team discovered SP was discussing the issues with me back in 2014. I was about to get an answer to this question when the discussions came to a halt.)
A close read of the CU data sheet indicates that the CUs can use an external 10 MHz reference, but they can also use an internal reference. The data sheet states “Multiple CUs can be synchronized to an internally generated or externally provided timing pulse.” I don’t know what the internal standard is, but it is likely an OCXO, not a TCXO. If that is correct, then the observed CU to CU bias is consistent with the use of the internal standard. If the internal standard was a TCXO, then the differences would be much larger. The internal standard may have been chosen for the Perth installation for the simple reason that it was more than adequate for the communications purpose, and easier to configure.
https://squarepeg.ca/index.php?page=mcp-channel-unit
@ALSM
I think that pretty much bags it. I also could not imagine the SDU as the source of bias differences. As I mentioned to Don, I never even considered the GES since I assumed, like you say, that the earth station would have an atomic reference of some sort for the whole station.
@Victor
I hear what you’re saying about the use of VS mode in situations where the pilot might require a reduced rate of descent and I certainly wouldn’t rule it out. However, a few comments:
‘For small changes in altitude where the “all or nothing” thrust schedules are not appropriate. The vertical speeds using the V/S mode would be lower than other modes.’
The thrust schedules used by other modes are not necessarily ‘all or nothing’. FLCH mode attempts to reach the MCP selected altitude in two minutes. The speed is controlled by pitch and the rate of descent is controlled by thrust. For a small altitude change of about 1-3,000 ft, the thrust reduces part way to achieve the altitude change in two minutes. The thrust only reduces to idle if the required altitude change is greater than about 4,000 ft and the two minute target then becomes irrelevant because it can’t be achieved. VNAV SPD mode also uses pitch to control the speed and thrust to control the rate of descent. If VNAV SPD is engaged for a cruise descent, the thrust reduces part way to provide a rate of descent of about 1,250 ft/min. The rate of descent only increases if the aircraft subsequently intercepts the idle descent path to the destination, in which case the thrust reduces to idle.
‘To get back on the target flight path for a VNAV descent if lower than the target path. The vertical speed can be reduced while still maintaining some minimum descent rate until the path is captured and a VNAV descent can continue.’
It is certainly possible to use VS mode in this situation, but it is by no means necessary. As above, VNAV SPD mode should provide a vertical speed of about 1,250 ft/min to intercept the idle descent path.
‘Another interesting feature of the V/S descent at high descent rates is that the indicated airspeed does not remain constant (unless the pilot uses speedbrakes to introduce drag). For instance, if 225 KIAS is set as the MCP speed at FL340 and a descent to FL200 is programmed at V/S=-2600 fpm, the … indicated airspeed increases as altitude falls.’
Therein lies a reason why VS mode should NOT be used with a high rate of descent. In VS mode, the rate of descent is controlled by pitch and the speed is controlled by thrust. If a high rate of descent is selected, the thrust reduces to idle in an effort to control the speed. Thereafter, VS mode cannot control the speed and the pilot must increase drag by using the speed brakes, as you noted.
@Andrew: Thank you for your comments. It makes sense that FLCH uses intermediate levels of thrust for small altitude changes.
If a quick descent is desired but without using speedbrakes, it would seem a V/S descent would fit the bill, especially if the starting speed is slow, such as the 225 KIAS considered here. Here is a slide that shows how the speed variables change during a descent from FL340 to FL200 at V/S=-2600 fpm. Even without applying speedbrakes, the maximum airspeed only reaches about 314 KIAS. If the descent were to continue, speedbrakes could have been used to limit the airspeed while maintaining the vertical speed.
But fundamentally I agree that a V/S climb can result in speeds that are too low, and a V/S descent can result in speeds that are too high, so extra care must be taken.
@Victor
Yes, a descent could be flown in that manner, but it would be very unusual for a pilot to select a high rate of descent and then let the speed run away, even if it maxed out below Vmo. In my opinion, most pilots would use FLCH in the scenario you described. For what it’s worth…
@Victor,
I know you do not subscribe to my via Medan theory, but what would your descent graphs look like for my MERIM (at 18:24) to UPROB track on L774, using the same time slot from 18:37 to 18:45, assuming the descent started at 18:37/18:38, either (case a) maintaining track to Uprob, or (case b) commencing the first slight left turn as depicted below ?
https://2.bp.blogspot.com/-hM-ohl4NTBQ/WDwj9bmkP5I/AAAAAAAAC5g/xN0ojWmSzrQL9H6LU1rg_d_uNm8R8d_pgCLcB/s1600/Escape.gif
@Victor
We will get some idea of rate of progress after a couple of weeks of watching the OI ship track, assuming their scheme is structured in a way that can be monitored, or they are more open with coverage information than Fugro/ATSB. We might then be able to estimate how far they could get in the baseline 90 days.
@Andrew: I understand. I think the cautions that you and @sk999 raise are valid.
@ventus45: If the track was as you depict between 18:28 and 18:44, there is no need for a steep descent to match the BFO at 18:40.
@Richard Cole: With 8 AUVs per host vessel, and only the vessel providing Satellite-AIS positions, I think it will be hard to estimate what area was scanned, except in a general sense. As for coverage rates, I’ve heard some numbers that would indicate that the first 25,000 sq km could be scanned within 30 days.
@Victor
“..it will be hard to estimate what area was scanned”
If the approach is structured, e.g. block by block, it may be easier to follow. If the ship movements don’t give the impression of such a scheme being followed it will be much harder, I agree.
@All
In line with this topic I like to put some more attention again on the ‘blue panel’ image.
I’ve corrected color, contrast, light and sharpness and endhanced pixel resolution.
Tell me what it looks like floating upside down and pay attention to the lines on top and the dimensions/angles of the (trapezium shaped) rectangular and colors.
To me it shows a shorter ‘south-end’ side (compared with the north-end) with a clear curved shape with color and sections like a B777 flaperon:
https://www.dropbox.com/s/stc5oqydhyysl2o/Xflaperon1.jpg?dl=0
It was reported to be about 4m2 which ~fits the size.
If this could be a MH370 flaperon it would be the left wing flaperon floating upside down.
@Ge Rijn: Can you determine exactly where the impact on the 7th arc would be? Likely, this area will be scanned after the first 25,000 sq km is completed.
@VictorI
I thought about it.
I had CSIRO/Griffin’s small east bound current in mind crossing the 35.5S area which then sharply turns west-bound crossing the 7th arc between ~33S/~31S.
Considering 97.82E is about ~120 miles from the 7th arc to the west and the image was taken on 28 march this would be 20 days of drifting at an average speed of ~10km a day if originated on- or near the 7th arc.
It seems a bit slow to me but that’s just a guess.
In the latest CSIRO/Griffin’s drift model, concerning their hot spots, the turned around current to the west (after first heading east) seems rather straight forward.
If on- or very near the 7th arc I would estimate ~32.5S/96E but I think the impact point would be more to the east starting between ~32.2/33S and ~96.4E.
excusse me.. I switched east(-bound) and west(-bound) again.. so, they ather way around..
just a bit too tired.. cann’t even type correctly anymore.. I’ll take a break..
@Ge Rijn: Well, that latitude will likely be searched, as it is just beyond the 25,000 sq km area.
@Richard,
In your paper you wrote that “Majority of MH370 floating debris were flat panels with little or no windage”
Firstly, where did you get this from?
In my paper I’ve derived a theoretical value of 3.3 percent for a thin flat-floating object, which is in a good agreement with the DGA’s measurements of the leeway factor of the flaperon.
Secondly, MH370 debris include not only 9M-MRO fragments, but also remaining of what it carried…
@Olex: I’m guessing that Richard’s “little to no windage” might extend to your 3%.
And certainly, it can be said that the majority [nearly all] of the RECOVERED, IDENTIFIABLE debris consists of shards of flat panels.
“What it carried..” Not identifiable, though authorities likely dismissed the Orion images for apparent lack of seat cushions and other cabin contents.
Which brings up a question I haven’t seen addressed: What floating debris would be expected at 2 or 3 weeks [when these images were acquired] for either high-speed or low-speed impacts. And, for any drift analysis, how would you model a decay curve as particles break up or lose buoyancy and sink? It will be interesting to see if Pattiaratchi and Gibson find any new beachings in Australia this season [implying residence in the SIO gyre]. My bet is that they won’t — that even the composite fragments will have long since gone to Davy Jones.
@Andrew: Here is another scenario to consider.
Let’s suppose that the waypoints are constrained in this way: MEKAR, FL340/225KIAS and SAMAK, FL200/225KIAS, but the MCP altitude is left at 34000 ft after passing MEKAR. So, there is a delayed VNAV descent to FL200 at SAMAK. At some point, there is a desire to expedite the descent to hit SAMAK at FL200. The VNAV descent screen says V/S=-2600 fpm (or something similar) is required to hit the altitude constraint at SAMAK. The pilot initiates a V/S descent at this descent rate AND deploys the speedbrakes. Then, the plane will hit FL200 near SAMAK and there will be no risk of overspeed. It’s similar to a FLCH, speedbrake/Vmo descent, but the descent will not be as steep and it can be timed to hit SAMAK at FL200. (I suppose a similar effect can be achieved by with a FLCH descent using speedbrakes with a speed less than Vmo, but the vertical speed would be less steady.)
@Ge Rijn
Interessant!
I tend to see lines running across the panel in the length direction. Difficult to associate that with a flaperon.Have you looked at the “Pemba” piece (outboard flap), it has roughly such a pattern of lines. It is definitely worth analyzing the “blue panel” images more in detail!
@Niels
The found flaperon shows fastener lines at the leading edge and the trailing edge at about the same positions.
In the image they are faint but sure make the same impression.
And yes I’ve looked at the Pemba piece in close detail back then.
It’s basically devided in same fastener-line sections as the flaperon.
@Ge Rijn,
I think all these photographs were discussed in great detail when they just appeared in the media. See, for example this excellent BBC summary (I am not sure why Victor did not include it in his post):
http://www.bbc.com/news/world-asia-26705073
Frankly, I would be more interested in knowing why these aerial images were not given any further consideration.
Ge Rijn, It seems to me that the 777 flaperon is an irregular polygon, with the longer ‘chord length’ on the inboard side and the leading edge longer than the trailing edge.
In any case, if one had good plan view drawing or photo of the flaperon it may be possible to compare the ratio of the inboard/outboard chord lengths as well as respective leading/trailing edge lengths between the debris photo and a drawing or photo. (It is not necessary to make absolute measurements.) It would be possible to apply a reasonable correction for the slanted view on the photo.
@Oleksandr said: I think all these photographs were discussed in great detail when they just appeared in the media. See, for example this excellent BBC summary (I am not sure why Victor did not include it in his post)
The specific photographs from March 29 were not discussed anywhere that I know of. The blue panel from March 28 was discussed extensively. In my last post, I provided a link to Bernard Lagan’s article, and even included a link to a video taken while he was aboard the craft. I have no idea what more you expect me to do. I certainly did not ignore the press reports on the blue panel that @Ge Rijn is analyzing.
My recommendation is that you go away, write your own summary, talk about whatever you want in whatever depth you want, and provide a link here when you complete your task.
@Victor,
“My recommendation is that you go away, write your own summary, talk about whatever you want in whatever depth you want, and provide a link here when you complete your task.”
Yes, that is certainly a good recommendation.
@all,
The often referred to “blue panel” photographed from a RNZAF Orion P3K was located immediately to the south of the Broken Ridge escarpment. Its shape when corrected for aspect and parallax matched that of an intact flaperon, but as the altitude of the Orion was known along with the image meta-data, it was deduced the object was actually smaller than a flaperon. Likewise, with spectra manipulation the object appeared to have two spars aligned lengthwise with about 33% spacing across cross its width. None of these (spars?) matched rivet rows etc.. expected to be found on a genuine flaperon. It was likely an item of fishing gear, and possibly in use.
An animation showing the apparent transformation of the “assumed” overall dimensions of a Flaperon into that presented by the “blue panel”. This animation is dated 16 September 2015 and was published in the JW blog some time ago – 2016?. Don’t forget to check the Rectangle as a comparison.
@Barry Carlson: I’ve heard it referred to an “Esky Lid”. For those not Australian, Esky is a brand of cooler boxes.
@All
An “EXKY” was originally one particular manufacturere’s product name, for their range of portable cooler boxes, originally marketed to the public as suitable transport / cool storage for perishables on family picknics in our VERY hot summers.
(And as I type this in Sydney, it is 11am, the sun is beating down, the temperature inside is already 27.7 C, the temperature outside is already 35.5 C, and the forecast is for a max of 40.1 C, and it is not even our “hot month” yet – which is February !!
Since then, the term “an ESKY” has entered Australian Slag to refer to ANY portable cooler box.
@ventus45
I’ve never heard of an EXKY. My understanding is that when the Sydney-based refrigeration company, Malley’s, first made a their portable insulated cooler in the early 50s it was called the Esky Auto Box. Over time that was eventually contracted to just Esky.
”For those not Australian, Esky is a brand of cooler boxes.”
Also known as a ‘chilly bin’ by our Kiwi friends, pronounced as ‘chully bun’. Go figure.
@Mick Gilbert.
My Typing error. (X and S are next to each other).
It was not e”X”ky but e”S”ky.
My bad.
We are probably confusing our overseas readers with this, so I found a “history link” which should resolve the confusion.
https://www.australianfoodtimeline.com.au/1952-the-esky-invented/
@Victor
Re: ”Let’s suppose that the waypoints are constrained in this way: MEKAR, FL340/225KIAS and SAMAK, FL200/225KIAS, but the MCP altitude is left at 34000 ft after passing MEKAR. So, there is a delayed VNAV descent to FL200 at SAMAK. At some point, there is a desire to expedite the descent to hit SAMAK at FL200. The VNAV descent screen says V/S=-2600 fpm (or something similar) is required to hit the altitude constraint at SAMAK. The pilot initiates a V/S descent at this descent rate AND deploys the speedbrakes. Then, the plane will hit FL200 near SAMAK and there will be no risk of overspeed. It’s similar to a FLCH, speedbrake/Vmo descent, but the descent will not be as steep and it can be timed to hit SAMAK at FL200. (I suppose a similar effect can be achieved by with a FLCH descent using speedbrakes with a speed less than Vmo, but the vertical speed would be less steady.)”
It is certainly possible to fly a descent in the manner you described, but it would be very unusual. VS mode is not normally used to expedite a descent when the aircraft is above the required descent path, for the simple reason that a high rate of descent in VS mode may cause an overspeed, as previously discussed. Speedbrake can be used to keep the speed under control, but the pilot then needs to carefully monitor the speed, thrust and speedbrake position to avoid the SPEEDBRAKE EXTENDED EICAS cautions, Master Cautions and beepers that will be triggered if the thrust increases while the speedbrakes are extended. That extra monitoring increases the pilot’s workload, something pilots try to avoid wherever possible – we like to keep things simple! FLCH is normally a much better mode to use in the scenario you described even though the rate of descent will fluctuate with atmospheric changes during the descent. If there is a requirement to cross a waypoint at a particular altitude, most pilots would probably use FLCH in the first instance and then finesse the last few thousand feet at a reduced rate of descent using VS mode or possibly VNAV.
re: esky
Aussie Jingle Bells
Dashing through the bush, in a rusty Holden ute,
Kicking up the dust, esky* [*thermal cooler-box] in the boot [trunk],
Kelpie [an Australian sheep herding dog breed] by my side, singing Christmas songs,
It’s Summer time and I am in my singlet [undershirt], shorts and thongs [sandals]
@TBill
Thongs = sandals?? Try ‘flip-flops’. Sandals are something an Englishman (or lady) might wear in the Australian heat (with socks).
@Andrew,
Esky = an Aussie contraction of Eskimo. Yes, we call them “chilly bin”, which an Aussie would pronounce as “cheely been”.
“Flip flops” are Jandals. Every NZer knows that.
Chilly bins are for keeping your beer in when you go to the beach. Aussies usually have something branded XXXX in their Eskys. The generally accepted view is that the company chose to brand their product XXXX because most Aussies couldn’t spell BEER.
Anyhow, the “blue panel” looks to be much larger than an Esky lid, and appears not to be rectangular.
@Shadynuk @Barry Carlson @VictorI
I guess @Barry Carlson gave you the answer on your suggestion with his animation?:
http://countjustonce.com/mh370/svg-flaperon-trans.html
@Barry Carlson
I’m aware that I cann’t proof the features I mentioned (lines, curves etc.) actually are real. But it’s what this picture shows/suggests.
Add your animation and you have further indication this could have been a LH flaperon.
It was never proven it was fishing gear either. The two spars you mention can be optical illusion too (just like the features I mentioned) and was interpreted from the other image (near the window edge).
But we’ll never know since the piece was not recovered. Just like the other (up to) 13 pieces that were observed near the blue panel.
@VictorI’s article by Bernard Lagan states:
” “It is my impression that we have entered a debris field,” the aircraft’s tactical co-ordinator announced over the radio.
On a clear day with little cloud and with sea conditions calmer than in recent days, the crew was able to pick out up to 14 objects. The largest was a light blue panel about four metres square that was clearly picked up by the three cameras on the belly of the aircraft and relayed to screens inside.”
I see it probably never can be proven this was a LH flaperon or any airplane debris but since none of the named debris was retrieved the possibility cann’t be rejected either imo.
The article also mentions the piece was clearly picked up by the three camera’s on the belly of the aircraft.
Suggesting there must be at least three images of the piece instead of the two published. If this is true I like to see those other images too.
@Brian Anderson
”Chilly bins are for keeping your beer in when you go to the beach. Aussies usually have something branded XXXX in their Eskys. The generally accepted view is that the company chose to brand their product XXXX because most Aussies couldn’t spell BEER.”
XXXX comes from that curious place called ‘Queensland’. Most non-Queenslanders wouldn’t be caught dead drinking the stuff.
@Oleksandr said: “In your paper you wrote that “Majority of MH370 floating debris were flat panels with little or no windage. Firstly, where did you get this from?”
My analysis is based on images and data contained in a large library kindly assembled by Tom Kenyon of the IG, which includes information sourced directly from the various finders as well as the official and unofficial analyses from various experts from a number of countries.
@Ge Rijn,
“The article also mentions the piece was clearly picked up by the three camera’s on the belly of the aircraft.
Suggesting there must be at least three images of the piece instead of the two published. If this is true I like to see those other images too.”
The images were obtained under an Official Information Act request to the NZ Defence Force by @Brian Anderson. The highest resolution image(s) available was/were probably not supplied. I have reason to believe that the aircraft was equipped with some very high resolution cameras, and images they took on the High Res camera at the altitude that was being flown (600 feet) would have revealed quite clearly what we have been discussing.
IMO it certainly wasn’t an Esky lid, as there appears to be some underwater attachment partially visible on the bottom left-hand corner.
However, the lack of positive physical identification of most (if not all) of the items photographed, doesn’t rule out that some of those seen were 9M-MRO related.
@Andrew: Yes, an expedited descent using V/S would be odd. So would flying a plane into the SIO. If there was an ongoing descent at 18:40, the only A/P mode I think would produce the measured BFO variation would be V/S. It’s also possible there was no descent at 18:40. There are possible paths in which the plane was flying south at 18:40, later flew north, and had a delayed final turn to the south. There is little certainty about what occurred between 18:22 and 19:41, and many of us have wasted a lot of time guessing. That’s why there is a lot of sense in starting the path reconstructions at 19:41.
@Oleksandr
I know these ‘blue panel’ images were dicussed in detail before, a long time ago. I was one of those who dived in to it too back then.
They were kind of dismissed/put aside back then partly also because the area was far outside the ATSB priority search area I think.
These days the area north of 36.6S has become the new priority area.
And @ VictorI shedded new light on the aerial finds on 28 and 29 March 2014 with this topic. So I thought it was the right time to put some extra attention on the ‘blue panel’ again also.
Your BBC-link is nice but it’s about debris spotted before 23 march (date of the article) far more south (see the graphic in the article).
I agree with you it would be very interesting to know exactly why the aerial images of 28/29-3 were not given further consideration.
Few things can be said conclusively about the satellite and aircraft debris photos along the 7th arc. One thing is for sure. The distance between the satellite debris photos (~S35) and the NZ P3 photos (~S30) mean they can’t both be from MH370. If they can’t both be from MH370, then at least one set is not from MH370. Thus, the only fact we can take away is that there was some non-MH370 debris close to the 7th arc ITVO S35 or S30, or both, and probably other places.
@Brian Anderson
Best cooler review I have come across. You don’t want to take chances with your beach beer. I emailed these guys to try to get them intereseted in MH370.
https://www.outdoorgearlab.com/topics/camping-and-hiking/best-cooler
BTW, Amazon (and their vendors) ship to New Zealand. I would appreciate it if you used them. I have a large position in AMZN.
ALSM
Yes, sharp reduction. One of those sets must be wrong, considering the distance between them. Or both, considering the lack of direct evidence.
Have to mention the ‘blue panel’ and the observed/reported debris field of 14 pieces around ~32.4S/97.8E on 28 march sets this ‘debris field’ apart right in the middle of the 35S and 30S ‘debris fields’.
In fact we have three ‘debris fields’ all around the same time.
The coincidence that all those three debris fields had nothing to do with MH370 at that time and in that very remote locations is rather slim I think.
But which one makes the best probability?
I hope more detailed information becomes available.
@Barry Carlson
You wrote re the “blue panel”: “but as the altitude of the Orion was known along with the image meta-data, it was deduced the object was actually smaller than a flaperon.”
I’m really curious how that calculation was done. Has it been documented somehow?
@Neils
The calculation is actually pretty trivial if the range to the object (aircraft altitude and angle to the image), and the focal length used for the imaging lens is known from the meta-data.
Crime scene photos ofen include a reference object such as a scale in the image. My pet peave was when my marketing people would create and ad for a board level product and include a picture without a reference object such as a coin (or a beer can for Aussies and Kiwis).
Bernard Lagan (quoted above): “a light blue panel about four metres square that was clearly picked up by the three cameras on the belly of the aircraft” Perhaps I underestimate Australians, but US beer coolers are rather smaller than 4 meters on a side!
I zoomed in till 600ft altitude above a B777 in Google Earth parked at Schiphol Airport at -3m altitude and made the comparison with the original ‘blue panel’ image/object and a flaperon on that B777.
The ‘blue panel’ image/object fits its dimensions rather nicely from that altitude.
Not a ‘scientific approach’ ofcourse but I think this comparison is better than an ‘Esky-Ltd’ beer cooler..
I’ll share the Schiphol B777 Google Earth picture I made at 180m altitude:
https://www.dropbox.com/s/f5j26wlw1x9yn4m/schiphol%20flaperon%20B777.jpg?dl=0
I found some meta-data on the ‘blue panel’ image:
-camera was a Nikon D2xs 12Mpixel with 24/16mm censor
-focal lenght 125mm at which photo was taken
-35mm crop-factor 1.5 which makes an effictive focal lenght of 187mm
-altitude of photo taken 184m (600ft)
This probably shows the panel slightly bigger in the image than it was viewed with the naked eye from this altitude.
Data on the GPS position are given;
-lat 32.28S long 97.49E
This shows different than the data provided by Richard Godfrey but I suspect this has to do with dergrees or minutes notation.
to add: the 35mm(sensor) focal lenght value of 187 on the 24/16 censor only affects the angle of view not the ‘zoom factor’. So the object dimensions stay the same on the sensor as with 125mm.
Something elso to consider is the object is viewed in an angle from 600ft altitude. Which increases the viewing distance to maybe 700 to 900ft.
@Ge Rijn, DennisW
Indeed, with the right data on distance, camera (image/CCD size) and lens used in principle it would be possible to estimate the size of the object, see for example:
https://nl.wikipedia.org/wiki/Beeldhoek
I hoped to be able to check such a calculation using for example image 5836, assuming it shows a SLDMB, however these images/files don’t contain the GPS info…
@Niels
Why not check on image 5809 which is the ‘blue panel’ image?
All meta-data is under ‘bestand/file’, ‘eigenschappen/features’ including GPS.
I made a rough calculation estimating the viewing distance at ~800ft with the focal lenght of 125mm (which closely resembles view with the naked eye but slightly larger).
I conclude dimensions are just about right compared to the GE Schiphol B777 flaperon picture I linked from straight above at 600ft.
But I sure would like to see a more precise calculation by you or @DennisW (or someone else).
@Barry C, others
I’d be interested to know what other EO imaging systems might be feasible as equippage on the Orions.
This presentation from RNZAF 5 Sqdn confirms the view I have held for some time. That is, for the spectrum of targets likely in the surface search for MH370 ‘Mk1 Mod 1 Eyeball’ is the most effective sensor. Most of the maritime patrol aircraft engaged in the search (Orions from Aus, NZ, RoK, JMSDF, and JCG G-V) were equipped with ELTA EL2022/M radar, the P-3K2s have Wescam MX-02 Electro-optical turrets similar to the RAAF AP-3Cs equipment but the field of view of these devices is not condusive to wide area search. When we previously put focus on these images from NZ and Aus we researched the Wescam and Safire EO system capabilities: each used COTS DSLRs as the sensor integrated into a turret.
Recent military deployments for Orions, providing ISR over ME and SW Asia theatres, are well suited to the publicly acknowledged radar and EO capabilities of the aircraft, particularly when the general area of the ‘target’ is known.
Higgins in The Australian today. ATSB etc have briefed OI on where to search, apparently in London.
The leaders of the failed Australian government-led search for Malaysia Airlines Flight MH370 have briefed staff of the private underwater survey company Ocean Infinity, which has made an audacious bid to launch a new hunt for the aircraft.
The move comes as one of the world’s top air crash investigators, US-based John Cox, has joined other international investigators and senior airline captains in saying the central assumption behind the Australian Transport Safety Bureau strategy for the first search was probably wrong.
Rather than the ATSB’s assumption of a “ghost flight” of incapacitated pilots and the aircraft crashing down rapidly after fuel exhaustion, Captain Cox said evidence from recovered wing flaps suggests the more likely scenario is that a pilot flew the aircraft to the end and ditched it.
At a two-day meeting in London early this week, officers of the CSIRO, the ATSB and other bodies involved in the original search briefed Ocean Infinity staff.
The move indicates Ocean Infinity is confident its in-principle contract with the Malaysian government to search for MH370 on a “no find, no fee” basis will go ahead, enabling the search to start in a matter of weeks. “We only talked about where they would search,” David Griffin, a CSIRO drift modelling expert, told The Weekend Australian.
ATSB spokesman Paul Sadler referred questions to the federal government’s Joint Agency Co-ordination Centre, which in turn referred inquiries to the Malaysian government.
Dr Griffin said “key people from the Australian investigation were all represented”.
MH370 disappeared on a scheduled flight from Kuala Lumpur to Beijing on March 8, 2014, with 239 passengers and crew on board. While the Boeing 777’s secondary radar transponder was turned off and radio contact terminated about 40 minutes into the flight, primary radar and automatic satellite tracking data suggest it doubled back over Malaysia before turning on a long track to the southern Indian Ocean.
The two year ATSB-led search of 120,000sq km ended in January with no trace of the plane, at a cost of $200 million to Australian, Malaysian and Chinese taxpayers.
Ocean Infinity, which is headquartered in Houston, Texas, but has offices in other parts of the world, a few months ago put an offer to Malaysia that it would search for MH370 but receive an agreed payment of up to $90m only if it found the wreckage.
It will send a vessel with advanced sonar scanning equipment to survey a 25,000sq km area to the north of the last search zone, identified by the ATSB and its fellow agencies as particularly promising based on new analysis of the satellite data and drift modelling of several parts of the aircraft found washed up on and off the coast of Africa.
Several aviation experts including Canadian air crash investigator Larry Vance have said the ATSB’s “ghost flight” and “death dive” theories are wrong, because the pattern of damage to the trailing edge of the mostly intact flap and flaperon indicates they were lowered by a pilot to slow the plane for a ditching.
Captain Cox has reviewed Mr Vance’s work, and said: “Based on that analysis I think it is likely, possibly highly likely, that there was at attempt to ditch the airplane.”
Neither Mr Sadler nor the JACC would say whether their officials still supported their original theory of a pilotless crash.
Second try – Victor, sorry, please delete previous attempt.
“Rather than the ATSB’s assumption of a ‘ghost flight’ …”
Pernicious myth. One of the little-know aspects of the DSTG “hot spot” is that is assumes, indeed may require, an actively piloted aircraft after the FMT. Fig 10.6 is revealing.
@sk999: I agree. In fact, now that the paths that terminate in the 120,000 sq km have been eliminated, the remaining paths that fall in the next 25,000 sq km area are for the most part curved, e.g., CMH paths, and require changes in Mach number after 19:41. On the other hand, a more complicated path before 19:41, e.g., a hold or excursion, allows paths after 19:41 requiring no pilot input.
When Larry Vance and John Cox talk about the ATSB fallacy of a “ghost flight”, really they are referring to whether or not the plane was in a controlled glide after fuel exhaustion. Terms like “ghost flight” and “death dive” are inexact terms that just cause confusion, and I have encouraged people here to not use them.
Re-visiting the DSTO Annual Review 2013-14.
Page 27: JORN
Page 40: MH370
@David
ATSB spokesman Paul Sadler referred questions to the federal government’s Joint Agency Co-ordination Centre, which in turn referred inquiries to the Malaysian government.
Dr Griffin said “key people from the Australian investigation were all represented”.
Well, that is all very clear. This effort is nothing more than a comedy by my standards. The original search was ill-advised, and the renewed search is ill-advised (in the new priority area). I have seen this movie before, and it did not have a happy ending.
@Don Thompson
It seems to me the WESCAM MX20-EO must have been able to lock-in on the ‘Blue panel’. The Bernard Lagan article states:
“The New Zealand crew had actually been assigned to search an area further to the west but dropped a tracker beacon over the debris field before moving on.”
This suggest to me they made some passes over it before moving on. Time enough probably to direct the WESCAM MX20 to the ‘blue panel’ after visual ‘MK1 mod eyeball’ sighting. This MX20-EO can read a car licence-plate at 750ft.
And also if you look at following video it must have been capable of very detailed images of the ‘blue panel’ and other spotted debris:
https://www.youtube.com/watch?v=jtYaejwd1Y0
I think it would be quite odd if they did not use the WESCAM at such an extraordinairy occasion.
So I assume it could be quite possible also images of the ‘blue panel’ exsist made by the WESCAM MX20.
Is there a way to find out?
@Don Thompson
“I’d be interested to know what other EO imaging systems might be feasible as equippage on the Orions.”
Anyway imo it would be ridiculous when a RNZAF P-3 Orion on a mission to find MH370 debris would only have used a by that time already far outdated Nikon D2 handheld camera to take pictures.
Any amateur could have done better than that with a more modern DSLR-camera.
@Ge Rijn,
“suggest to me they made some passes over it before moving on”
You could be more sure by inspecting the time of image acquisition metadata and using that data to recreate the Orion’s track (as Victor has described in the above post). The timing of image sequences for single objects do not indicate that the aircraft made mulitple passes over objects.
Putting the EO system on a target, and maintaining a ‘lock’ on that target, isn’t a trivial operation when flying a linear track at 200KIAS at low altitude.
@Don Thompson
That was a good advice.
Time elapsed between image 5809/5810 and 5811 is ~42 minutes.
Distance covered between ~97.8E (image 5809/5810) and ~96.6 (image 5811) is about 80 miles/130km.
This would make a ground speed of ~180km/h for the plane over that distance. Which seems quite slow to me.
Suggesting some loiters were made above the debris field.
Another indication is they dropped a beacon at the site. You have to prepare for this which takes time too.
I’m convinced a professional team on a P-3 Orion mission would have used the WESCAM MX20 in such an occasion or other specialized equipment.
Not only an outdated handheld Nikon D2.
My question would be; where are those images?
to add.. to drop a beacon they at least had to turn back to the spotted debris location. This would have given them time to prepare the MX20 which has highly advanced automated capabilities to identify and track possible ‘targets’ in a known area/field. Which would have been the case in this occasion.
@David
I find the article very encouraging to think there is active planning to resume the search soon.
As far as bringing Cox and Vance opinions into the article, well that was an intentional diversion. I do think they have a valid point that very little (publicly disclosed) official thinking has explored the hypothesis of an active pilot to the end.
The good thing about the ATSB orig search, is that is rules out 18:40 FMT straight path to 38S. Now, as Victor says further above, we are back to using 19:41 ARC2 as the starting point (ATSB orig approach). I personally would not assume inactive pilot after 19:41 though.
@TBill
I think the paradox of the flaperon and outboard-flap damage (and other found pieces) and a high speed nose down impact is still not solved and therefore has to be considered by OI.
An attempted ditch after a steep descent is still a real possibility.
An impact point farther from the 7th arc has to be considered IMO by OI.
And indeed the initial straight paths more probably indicate pilot-induced or controlled flights after FMT, no ‘ghost-flight’ after FMT.
Like @VictorI suggests it’s more about a pilot-controlled flight after FMT and a pilot-controlled impact at the end.
Which could have been a pilot-controlled high speed impact or a steep descent with a recovery and a ditch attempt.
Not about ‘ghost-flights’ or ‘death-dives’.
I agree with Cox and Vance the damage to the flaperon and outboard-flap (and other pieces and their positions on the plane) strongly indicate the latter (ditch-attempt).
The final BFO’s on the other hand suggest a high speed dive impact scenario.
Quite a dilemma..
@Ge Rijn
I agree with Cox and Vance the damage to the flaperon and outboard-flap (and other pieces and their positions on the plane) strongly indicate the latter (ditch-attempt).
The final BFO’s on the other hand suggest a high speed dive impact scenario.
Quite a dilemma..
To my knowledge Cox and Vance have never actually examined the debris, but are basing their conclusions on photos like the rest of us. I find it hard to believe that laboratory forensics have not provided more definitive information relative to the debris condition, and how it relates to the impact.
@Victor,
It is very odd, that OI seems prepared to spend millions on an independent search, when Malaysia is quite obviously not interested in entering into ANY agreement at all. Did OI put a deadline on contract negotiations ? If so, and if that deadline has expired, will OI decide to “go it alone” anyway ?
It is also very odd that the ATSB and associates (CSIRO, GeoSciences, DSTO/DSTG et al) are also apparently so willing to assist OI, but again, without any (apparent) willingness to pay for OI’s efforts.
Consequently, it is logical to assume, that OI must have “a reason” for searching independently of the Australian and Malaysian analysis and contracts.
It is also odd, that OI apparently intends operating out of Durban, instead of Perth / Fremantle. Given the relative distances to the “sanctioned” search areas, that is interesting. There must be a significant reason for that.
Therefore, one must wonder, is there a hidden party behind all these games, and if so, whom ?
Has OI done it’s own analysis (or has some other agency done it for them) and are they therefore going to search some other “not oficially sanctioned area(s)” ?
If so, what areas ?
Either way, if they do find it, in international waters, and they collect direct evidence of a “criminal act” (remembering that their ship is fully equiped with all necessary recovery gear, including a 250 ton crane with 3,000 metres of cable reach) is it likely those recovered items end up in OI’s home country, ie, the USA ?
@ventus45: I think you are making a lot false assumptions, including the port that OI will use for servicing.
@VictorI
I did some processing on the ‘panel with wires’ image.
Enhancing one color (in this case red) lights up all features in the image that have similar reflexion.
You can see the ‘wires’ are actually wave-structures like showing all over the image:
https://www.dropbox.com/s/ypdrsxzvoq6w9lr/Panel-and-Wires2.jpg?dl=0
..then adding/enhancing only yellow lights up those wave-structures better;
https://www.dropbox.com/s/pjtd22tjec3veo1/Panel-and-Wires3.jpg?dl=0
Ventus45 wrote “Consequently, it is logical to assume…”
Are you sure ‘logical‘ is the appropriate adjective to use in the context of your entire comment? Perhaps consider ‘wildly speculative’.
@Ge Rijn: Interesting. That object could be more fishing gear.
I’ve always thought that the two debris fields are the most interesting of all the images (including the “blue panel”).
@Ge Rijn: Let’s assume for the sake of discussion that there was an attempt to ditch the plane. Do you believe the damage to the flap and flaperon are consistent with extended or retracted flaps? ATSB/NTSB/Boeing believe the flaps were retracted because of the alignment of seal pan stiffener damage with the support track, and the alignment of damage to the flap and flaperon near their respective rear spars. This was all detailed in the ATSB report released on Nov 2, 2016. If you do not agree, why not?
@Ventus
It is also very odd that the ATSB and associates (CSIRO, GeoSciences, DSTO/DSTG et al) are also apparently so willing to assist OI, but again, without any (apparent) willingness to pay for OI’s efforts.
None of the above people end up paying OI. It is the citizens of the country that foot the bill, be it Malaysia, Australia, China, or some combination of the three. Like any business deal, it is important to understand where the authority to make a purchasing decision lies, and what factors effect that decision. At this point the tri-partite group is essentially “free and clear” of MH370. They have made a beyond good faith attempt to find the aircraft.
Another factor to consider is the objective value associated with finding the aircraft. My own guess is that the CVR will be blank, and that the FDR will show the aircraft was flown to where it was found. 150M+ USD has already been spent in the search effort. IMO, that is far too much money considering the potential benefit (closure to family members is a soft issue that I will not touch since sensitivities vary so greatly relative to that). From my perspective the only value of recovering the FDR lies in finally putting the flight path between 18:25 and 19:41 to rest for whatever that is worth. Does anyone except a few die hard failure advocates believe the FDR will reveal some complex sequence of events?
Consequently, it is logical to assume, that OI must have “a reason” for searching independently of the Australian and Malaysian analysis and contracts.
It would be hugely beneficial to OI’s future business if they are succcessful in finding 9M-MRO.
@VictorI
The ‘panel with(out) wires’ shows very irregular with broken edges all around and also an irregular/broken surface.
I don’t see how this can be related/identified to any kind of fishing gear.
This also counts for the ‘blue panel’ imo.
On your question, for the sake of discussion, about the damage to the flaperon and outboard flap section and retracted or not:
First I like to say the trailing-edge damage and leading-edge lack of damage and the separation of both would be easier explained when both flaps were deployed on impact. I also see this in the context of most other found debris. Especially the right wing aileron piece, the left wing outboard flap trailing-edge piece, the left and right wing flaperon closing panels, the left and right wing flap fairing pieces, the three engine cowling pieces, the horizontal and vertical stabilizer pieces.
And the nose landing gear door piece.
But I think it could have probably been caused all also with flaps retracted when the pitch angle was positive enough on impact.
I don’t believe the ATSB explanation the rear spar of the flaperon collided with the outboard flap causing the dent/damage in the side pan of the outboard flap. The rear spar of the flaperon is ~20cm away from the outboard flap in that (aligned) position.
The flaperon has nothing to do with the damage seen on the outboard flap side pan imo.
The outboard flap side pan (and top and bottom) clearly show compression induced damage on- and towards the top and tension induced damage from the bottom towards the top. Particularry visable in the large crack through the side pan and the dent in the upper side pan and top surface (which ATSB contributes to impact of the flaperon end-spar).
Combined with the broken away trailing edges and almost intact leading edges of both flaps and in the context of more then 90% of the found debris this clearly points IMO to a ditch-like impact.
And also important is that all debris analysid by Malaysia showed mostly- or only tension damage. None or very little compression damage.
So a nose-dive impact could not have generated this pattern of damage and this specific set of found debris.
This is in short my motivation. I’ve been though this in much more detail long time ago on JW’s blog
@Ge Rijn: There is no reason to rehash your opinions. On the surface, it seems unlikely that ATSB/NTSB/Boeing could have been so wrong about the debris, but certainly it is possible.
Regarding damage to the trailing edges of control surfaces when retracted, look at the damage to the aileron of US1549. As you know, ailerons don’t extend. Yet, the right aileron was ripped from the wing during the ditching.
@DennisW
On your comment;
“I find it hard to believe that laboratory forensics have not provided more definitive information relative to the debris condition, and how it relates to the impact.”
I also find this hard to believe. Cox and Vance in case of having the debris under their hands would be able to make a decisive conclusion I suppose.
The French, the ATSB, Boeing, the Malaysians, even the Dutch 😉 must be able to do the same.
Why are the detailed forensic reports kept away from us comes up easily.
@VictoI
I also mentioned the aileron of US1549 back then. And all other similar features and damage on flaps and engine cowlings (even relative recently with ALSM).
To be more precise; the aileron was not completely ripped away on US1549 but shunks of its complete ‘trailing edge’ were ripped away during the ditch. The tip of the wing stayed intact.
A similar right wing aileron shunk was found in SA from MH370.
When I first learned about OI’s offer I was intrigued but at the same time never believed Malaysia would go for it. There are many conspiracy theories about Malaysia but my underlying logic is politics. Sometime in the first half of 2018 Malaysia is going to have national elections. The longtime ruling party has alienated the Indian and Chinese minorities and also many Malays — so the opposition has a chance, maybe even could win. Announcing you are contracted to pay up to $90 million if the plane is found seems like a black eye no politician would willingly occur. Imagine if the prime minister in Australia agreed to this, there would be a huge back lash. What Malaysians really care that much about finding the wreckage, the families and maybe a few more. But many could object to more fiscal irresponsibility, which is part of the ruling party’s electoral weakness.
Anyway we should know in a few weeks but if negotiations conveniently break down I hope OI is committed to go ahead anyway, hopefully they have a friendly billionaire who hasn’t spent all his money on sports teams.
Back on Oct. 9, I noted that, based on information given in the JIT report (Appendix B of the ATSB Final Report) for March 31, 2014, the BFO calculation for a position on the 6th arc near the Ocean Shield position seemed to be in error by ~12 hz.
Appendix C of the final report gives Inmarsat’s version of events around then. On April 8, it repeated the calculation (start from 2nd arc; adjust speed and starting latitude) using a newer version of the AFC+Sat model, and found that the final position that matched the BFO shifted from -23 to -28 degrees. Wow, that’s a lot! Yet there was sill likely an error of 6 hz in the BFO calculation. What was going on?
Inmarsat generated three main sets of BFO models: the original “OAMS” models, the “Eclipse” models, and the “Unified” models. The main differences seem to be how the AFC+Sat and uplink/downlink Doppler calculations were done. These differences have never been described quantitatively by Inmarsat but they were obviously substantial.
Fortunately, the impact of errors in the calculation can be quantified fairly simply. The BFO at 00:11 on the 6th arc depends only weakly on the speed and heading of the aircraft (perhaps 1 hz or less for early v. late FMT) but strongly on the final latitude, since the bulk of the Doppler is due to motion of the satellite. Further, the dependence is nearly linear: to a good approximation,
BFO(6th-arc) = 221.1 – 0.884*(latitude-on-7th-arc).
Going in the reverse direction,
(latitude-on-7th-arc) = -1.13*[BFO(6th-arc) – 221.1]
All search areas up to April 8 that incorporated BFO data (including the Ocean Shield search area) were defined using the OAMS model. The initial underwater search area in the ATSB report from June, 2014, was defined using the Eclipse model. Note, however, that Appendix G described the Unified model, which had just been released. The search area defined in the October, 2014 report and all subsequent search areas used the Unified model. An error of 12 hz at 00:11 translates to an error in latitude on the 7th arc of 13.6 degrees. That accounts almost entirely for the change in high priority search zone from the Ocean Shield zone to the Fugro zone between April and October of 2014.
For the record, the measured 00:11 BFO is 252 hz, and my equation gives a corresponding latitude of -34.9 degrees. I will assign a 1-sigma error of 2.6 degrees. Victor, that is the area I recommend OI search.
Addendum: I no longer use Table 4 from the June, 2014 report, but rather use a combination of the OAMS figures from JON (Fig 11) and the Final Report (Appendix C, Fig 2). It leads to a simpler BFO calculation and eliminates some cruft that Inmarsat left in its original formulation of the BFO calculation.
@sk999
Yeah, we batted this around before with agreement.
Another possibility (my personal preference) is that the aircraft was already descending slightly at 00:11. Take your pick.
DennisW,
Yes, you provided a good sanity check back then.
I did not consider driftdown but should have, and can do so now. I get about -17 hz change in BFO for 1000 fpm of descent. It is not a Gaussian error – instead, always in the direction of mimicing a more Northerly latitude. So if there were some driftdown, it would mean the true latitude is further South, but that is in the direction of the area that has already been searched.
@sk999: I think your latitude range of 34.9 +/- 2.6 deg is for the 6th arc. What latitude range along the 7th arc would you propose?
Victor,
No, that’s for the 7th arc.
@sk999: OK. Thanks. Your latitude is consistent with straight paths that minimize the BFO error, which is not surprising. My BEDAX-SouthPole path is around 34.3S latitude, and Inmarsat’s proposed path ends at 34.7S latitude.
The first 25,000 sq km to be searched extends north to 32.6S latitude along the arc. You are suggesting that the area be extended to 32.3S latitude. I would be surprised if this was not searched.
@sk999
So if there were some driftdown, it would mean the true latitude is further South, but that is in the direction of the area that has already been searched.
You are correct. The ravages of time and age strike again. Now that I think about it I needed the descent to make the original search area work. 🙂
@sk999
Can you be more specific on the limits in speed and track between which the linear relation “BFO(6th-arc) = 221.1 – 0.884*(latitude-on-7th-arc)” is a reasonable approximation?
The reason I ask is that there are for example strongly curved paths consistent with BFO/BTO ending in the lower (S) 20 degrees latitude. There must be some (hidden) assumption in your approximation.
Niels,
There really are no constraints as long as the speed is roughly constant. Consider the original Ocean Shield route – it started near LAGOG at 19:12, flew at 400 knots, and crossed the 6th arc at -23.4 latitude on a heading of 157 deg; the BFO I calculated was 240 hz. Now consider a route with an FMT at 18:29, just after the first arc, and reaches the 6th arc also at -23.4. Now the ground speed is 340 knots, the final heading is 145 deg, yet the BFO is 242 hz, a change of only 2 hz. Both of these routes are curved.
The most Northerly endpoint I computed was for -15 deg latitude (speed of 300 knots).
I should say that I ran routes for both an early FMT (18:29) and a late FMT (18:39 at IGOGU). The linear fit is just to the IGOGU routes. My best-fit endpoint of -34.9 is virtually the same as Inmarsat’s sample route, which ends at -34.7, not surprising because we are both make simlar assumptions.
@sk999
Interesting. I think in principle one constraint in addition to the BFO/BTO data would be enough to indicate a unique location (ignoring the BTO and BFO errors). I also found -34.9, by minimizing the path curvature. Possibly, minimizing speed variation works as well; I have to think a bit more about your procedure to fully understand it.
Practically, the problem is of course in the (partly unknown) BFO errors.
Victor, in this post, you wrote “ATSB/NTSB/Boeing believe the flaps were retracted because of the alignment of seal pan stiffener damage with the support track, and the alignment of damage to the flap and flaperon near their respective rear spars. This was all detailed in the ATSB report released on Nov 2, 2016. If you do not agree, why not ?”
“If you do not agree, why not ?”
I don’t know what Ge Rijn thinks about the “provenece” of these debris items, but my uneasy conclusion is, that the debris of the flaperon and the inner section of the outboard flap, are very possibly, even highly probably, the flaperon (now known to be the original production line build fitted item from the internal component serial numbers) and flap (original history unknown, but also presumed, at this stage, to be the original production line build fitted item) that were still fitted to 9M-MRO, when it’s starboard wing was severely damaged in the ground collision in Shanghai.
Consider this scenario.
That collision occured “on the ground” sitting on it’s undercarriage, when both flaperons and flaps were in the “up” or “faired” position, as is quite normal when the aircraft is on the ground.
The wing was “hit hard” in that collision, in the starboard wing-tip area.
The whole aircraft moved significantly in the collision, and the impact “flexed” the entire starboard wing “rearwards”, thus “compressing” the trailing edge components, such that the flap implacted the flaperon laterally. Thus, the ATSB/NTSB/Boeing contention that the items “damaged each other” when in the “up” or “faired” position, is entireley logical, and reasonable.
Those damaged items were removed and replaced with other new or recycled repair pool components in that overall wing repair.
The damaged components were subsequently returned to MAS.
My uneasy conclusion is this. These two items of “recovered debris” (the Reunion Island Flaperon and the Pemba Island Flap) are most probably these ground accident damaged items, which were carefully “sanitised’ by removing all identification plates, and have been deliberately “planted”.
It is probable that the French BEA have also deduced this, and it is very likely “why” the French Judicial Authorities have retained “The Reunion Island Flaperon” and “steadfastly refuse to surrender it to Malaysia”.
What should have happened, quite obviously, (and in any “normal accident investigation” would have – without question) was that the Pemba Island Flap, should have been taken to France, so that both it, and the Reunion Island Flaperon, could have been forensically examined, together, side by side.
That did not happen.
Not only that, but neither the Malaysians, nor even the French, made any attempt whatsoever to secure the Pemba Island Flap at all. Why ?
The fact that the Malaysians have so blatantly not been interested in recovering “ANY” debris, from anywhere, ever, is well documented.
But why didn’t the French, who already had the flaperon, (and quite obviously, many suspicions of their own), not make a bee-line for Pemba, to recover the flap ? That still needs an explanation.
In any case, it ultimately fell to the ATSB, (of all people), to arrange for the Pemba Island Flap to be brought to Australia.
The fact that “side-by-side” examination of the two components did not happen, and the fact that the ATSB has now (apparently) returned “The Pemba Island Flap” to Malaysia, speakes “volumes” for the overall “transparency”, and “integrity”, of this “investigation”.
If the aircraft went in at high speed from a near vertical spriral dive, there will only be fragments, scattered around 6 identifiable major dense components, ie, the two engines, the APU, and the three gear sets.
But if the aircraft went in at low speed in a “ditch like” manner, there will also be substantially intact, significantly sized components, of the wings, tail, tailplane, and fuselage.
In this case, “the condition of the trailing edge of the starboard wing, and it’s components, will be most telling”.
Should OI (or anyone else) ever find the wreckage of 9M-MRO on the sea floor, their first priority must be, to find the starboard wing, ALL of it, and scan it, and document it, in the highest resolution possible, and immediately broadcast it to the world.
@Victor
The reality is we do not have a clue where the aircraft is. 35S?? 30S?? Distance from the 7th arc?? Is a new search really a smart idea? Not if I was funding it. My guess is that if the OI contract gets a go ahead that it will be the end of OI (or a windfall). Take your pick. Not something (a no find no fee contract) a prudent man would do for sure.
@ventus45: At this point, there is no reason to think the debris was planted. There are far simpler explanations for all the evidence.
@DennisW: OI hasn’t asked any of us whether or not they should search for MH370. Rather, they have sought advice from the ATSB and others as to where to search. From what I can tell, they have an understanding of the risks involved. I am encouraged that they are willing to search, and I hope they succeed. Whether you or I or others here would be willing to undertake the search is irrelevant.
@VictorI
On the right wing US1549 aileron I like to post an illustrating photo including the intact wingtip:
https://www.dropbox.com/s/xlv69gmiebgfp3n/hudson3.jpg?dl=0
And worth pointing out is that also in this case the first section of the outboard flap broke off.
@Ge Rijn,
Concerning your image of the USAirways Airbus. It was taken as the airframe was being recovered at the river bank. Prior to recovery, the airframe had settled to float right wing down deep into the river and it was evident the the wing tip had scooped up river mud in the time before its recovery. That image is of no use for making a determination that the aileron departed in the water landing.
Don Thompson
And ofcourse this (your) argument comes up again too:
The aileron’s complete ‘trailing edge’ broke away when the right wing settled in the rivermud and when the plane was recovered the weight of the mud did further damage.
But miraculously the wingtip and winglet survived this all undamaged…
It’s just a nonsens explanation imo. The most obvious and simple explanation is this aileron’s ‘trailing edge’ broke away during the ditch.
This picture shows it clear:
https://www.dropbox.com/s/6cm1n3prs3ymmwg/US1549aileron-winglet.jpeg?dl=0
The wingtip is hardly damaged. Even the wingtip lights are in place and the winglet stayed mostly attached. Also the slats stayed attached undamaged.
No way only this much stiffer aileron ‘trailing edge’ could have been broken away by ‘mud’.
It shows all signs of chipping/breaking off during the ditch.
Same kind of damage pattern as the found flaperon and outboard flap section.
@GeRijn
Add to the observation that the remainder of N106US is substantially intact. The lower fuselage was breached but the interior, the empennage, the leading & trailing edge panels all remained intact. The N106US engine cowls did suffer damage but not to the extent evident from the remnants of -MRO’s inlet and fan cowls.
It’s disingenuous to describe N106US as having simply ‘settled in the mud’. I did not use those words above. The airframe drifted some way down the river and was temporarily tied up on the Manhattan side against a stone/concrete river wall. It may have collided with submerged objects during its drift or even a boat during the rescue of passengers. I remain highly sceptical that the aileron and wingtip damage was incurred during the water landing.
I’ve suggested before that you set aside your ‘opinion’ for the condition of 9M-MRO’s flap and flaperon, to consider what the condition of the other recovered items might suggest. That really would be a worthwhile exercise.
@GeRijn
Thank you for providing the image depicting the N106US starboard wingtip.
For reference, here is an image of the port wingtip. Note the water landing was completed with wings level, the port winglet is intact and its paintwork quite undamaged, and the port aileron is intact.
Do try that exercise I suggested.
@GeRijn
If the damage to the right aileron of N106US was caused by the ditching, would you not expect the adjacent flap section to exhibit similar damage? The following photo shows the flap section was relatively undamaged:
https://static01.nyt.com/images/2009/01/19/nyregion/19plane_span.jpg
@Andrew: Good point.
Don Thompson @Andrew
Yes that are seemingly conflicting observations.
Same as why did the left wing engine separate and the other stayed attached?
Why did only the first section of the outboard flap broke off?
Why did the left wing slat at the wing tip broke away and the right wing slat stayed undamaged?
Maybe when loosing the LH engine on impact the weight of the RH engine caused the RH wing to drag deeper and longer through the water?
You can actually see the plane is drifting with its left wing tip above the surface and the right wing tip slightly under.
I know I cann’t prove how all this damage occured. As far as I know there has been no specific investigation into the damage and separation causes by the NTSB.
But to me the pattern is clear; most of it, if not all, points to damage caused by the ditching.
And the similarity of the parts affected, their positions on the plane and their kind of damage have striking resemblance with the fast majority of pieces found from MH370.
In this sence US1549 has significance to MH370 and should not be ignored or called useless imo.
@sk999
That’s about (34S) where I am too on location. It fits the BFO/BTO data best and I am less model-centric and more a data-centric. Not to say there are not other solutions, I am just too slow to identify them. I imagine if OI hopefully finds the crash site, we will still have many path proposals for how it got there.
@Victor
Whether you or I or others here would be willing to undertake the search is irrelevant.
🙂 For sure. I fall into the trap all the time of imagining what I would do in a given situation, and freely dispensing advice on that basis. It just pisses everyone off – my kids, my SO, my friends,…
So, I will do it again. OI would be wise, IMO, to get Metron involved. Perhaps they could negotiate a greatly reduced upfront fee based on a share of the finder’s reward. Like you, I truly want them to succeed.
@GeRijn wrote “You can actually see the plane is drifting with its left wing tip above the surface and the right wing tip slightly under”
This video clip is comprehensive with views from both shores of the landing, drift and rescue. The aircraft slewed to port as it slowed on the water, both wingtips are visible above the river surface. It continued to drift with the starboard wing facing downriver. The larger ferry boats approached close from downstream, so as to manoeuvre against the river flow. The outboard end cap of the outboard flap is also damaged, indicating post landing collision, either with a boat or underwater obstruction as it continued to drift and while it was restrained at Battery Park Esplanade/River Terrace rather than damage sustained in the landing.
We are perpetually at loggerheads with these interpretation of observations so without anything fresh and persuasive let’s just set this aside.
@DennisW: Out of curiosity, what capabilities do you think that Metron brings to the table that don’t already exist within a team comprised of OI, ATSB, DSTG, CSIRO, NTSB, Boeing, Inmarsat, and Thales?
@Victor
what capabilities do you think that Metron brings to the table that don’t already exist within a team comprised of OI, ATSB, DSTG, CSIRO, NTSB, Boeing, Inmarsat, and Thales?
Pedigree. The “team” you refer to above is not in the business of finding lost airplanes. Just not their bag. Conversely, I would not engage Metron to build a 777 or an AES.
Don Thompson
I agree to let this aside unless something new comes up.
There’s no point in repeating all pro’s and con’s all over again.
I’ve made my overall point of view clear I hope.
@DennisW,
Metron’s analysis for the impact point/debris location of AF447 was not the sole contribution that guided the final, successful, search phase by Woods Hole Institute. WHOI, UMASS staff and a NZ contributor submitted work that was also accurate in describing the impact point. A third contribution to BEA was also accurate. So ultimately, three contributions ‘marked the spot’ for WHOI’s successful operation.
I find that comparisons between the search for AF447 and MH370 require much qualification, the extent of the total area in scope to search for AF447 was only 17,600km^2, tightly bounded by maximum 10mins flight from a solid LKP deemed to define radius of 75km. The BEA incremental approach resulted in only 88 days expended while working the deep ocean search.
@Don
I am not focussing solely on AF447. That effort is certainly a positive, but we also have the USS Scorpion, the SS Central America, and the role Metron has played in the design of the US Coast Guard SAROPS system. As a team they have far more relevant experience than the team Victor refers to above (I would also include the IG and many people here) who have never found anything or even participated in an effort of this type. I would not even describe them as a team, but rather as loosely knit group of non-colocated collaborators.
@DennisW: Whether or not the addition of Metron as a team member would help find the plane, I assure you that the team OI has assembled has deep experience finding aircraft and ship wreckage on the seabed, including but not limited to AF447.
@Victor
There is no doubt in my mind that adding Metron would be helpful. If I had a financial interest in OI, I would be screaming at them to engage Metron. If I was a director, I would insist on it. Why the ATSB did not engage Metron early on remains a mystery to me if for no other reason than as a cover your ass strategy.
@DennisW,
I understand that AMSA also use the SAROPS software. David Mearns finds lost ships too, his technique might be described as little more ‘old school’: thorough research of all possible sources, and traditional nautical charts. He credits David Griffin’s oceanography in the searches for HMAS Sydney and Centaur.
Earlier you responded to my comment about OI’s apparent trials work across the Atlantic Ocean since July. After its most recent port call at Trinidad, while enroute for Durban, Constructor took two days out to search a ‘box’ in the vicinity of a site where a WW2 wreck is located (disclaimer: I assume that wreck was the target, open source information from AIS tracking and wreck site data). An ability to survey >600km^2 a day is ‘disruptive’ to the subsea industry’s status quo. It just occurred to me that perhaps Metron’s raison d’être has also been ‘disrupted’!
@VictorI
You obviously know more than probably anyone else here about the team OI has gathered. And to me this sounds reassuring.
In any case I support any company who has the guts to step in and continue the search on a no-cure no-pay basis.
OI has the opportunity to test their new technology in a real and very important search. I think it would be well worth the investment whatever the outcome for them.
And they’ll have at least 3 months of world wide news covering advertisement too.
If they succeed also in finding the plane their name will be settled in this industry and in history.
And I think they make a good chance in finding the plane. Odds are much better than the ATSB ever had.
At least if they don’t forget to consider a possible glide and ditch too.. (yes sorry, I won’t stop bringing this forward..)
@Don
It just occurred to me that perhaps Metron’s raison d’être has also been ‘disrupted’!
Could be. I imagine that MH370 would have high appeal to Stone, but I could be completely off base with that opinion.
600km^2 per day is impressive. I had not heard that before.
@DennisW,
600km^2 day: David & I had exchanged views on it a while back, based on 100km^2 day for Fugro’s work x 6 Hugins operating concurrently.
@Don
Ok. So that is an estimate.
I imagine some time would be lost retrieving the AUV’s and replacing batteries, refueling surface vehicles, etc. I also have the impression that OI will be scanning at lower resolution looking for srtong reflections rather than actually mapping the seafloor. So their area coverage per sensor might be greater than the Fugro case.
If OI actually gets started it will be fascinating to see how their approach works in difficult conditions and over extended time periods.
Re: Search rate of the OI AUV vehicles
I have done some analysis on the recent OI search (December 13-17th) around the reported position of the wreck of the German blockade-runner Weserland, sunk by the USN in 1944.
The work seemed to be in three phases:
1. A Search phase of 50hr, starting at arrival
2. A Recovery phase of 21hr
3. A Stationary phase of 6hr, ending at departure for Durban
The ‘recovery’ phase has five points where the ship is stopped for two hours on each occasion. These points are symmetric with the historic wreck location and I hypothesize that five AUVs are being recovered at these times. On that basis the area has been divided into five roughly equal areas, each of 300sq.km.
The search phase of 50hr corresponds to the endurance of the Hugin AUV. If I assume that the total contiguous area was searched in that period then the achieved rate is ~140sq.km/day/per AUV, for the time in water. The additional time taken to recover and refurbish the AUV is not included in this rate.
The points of deployment of the AUVs are not identified in the S-AIS data, perhaps because the ship is only stopped for a short time, while the recovery takes much longer (2hr).
In the ‘stationary’ third phase the ship’s position is stable to a few metres, which may indicate that the ROV is deployed. This behaviour was seen in the Fugro ROV deployments.
The two diagrams indicate i) Phase 1 and ii) Phases 2 and 3. In the second diagram I have sketched an indicative AUV track which would take 43hrs to execute at the expected AUV speed and pitch.
https://www.dropbox.com/s/c0dqpjscm31bwwq/Weserland_activities_-_Dec_2017.jpg?dl=0
The recorded detailed movements of the ship in the centre of the search area are not fully explained. The ship moves over the same area twice, whether tracking one or two AUVs, or doing something else, is not clear. A sixth AUV may have been deployed and recovered before the ‘recovery’ phase identified above but is not clear in the data.
@Richard Cole: That’s very, very impressive. First, the location of the sinking of Weserland was reported as 14°55’S, 21°39’W, which as you say puts it in the center of the scanning area. Your calculated value of 140km2/day/AUV is also very close to what I have been led to believe is possible, which would mean scanned areas exceeding 1,000 km2/day for 8 AUVs. This is a game changer.
@Victor, Don, Richard Cole. Page 4, https://www.sut.org/wp-content/uploads/2016/07/Jan-Ingulfsen-distribution-approved-PPT-at-v2-Ocean-Infinity-Perth-distribution.pdf
“up towards 900 line km/day”. At an effective 1600 m search swathe that is 1440 sq km/day.
I have mentioned this before, noting that perhaps this would be for one day in ideal conditions and not include replenishments like battery change or data downloading (physical or electronic), or any transits or turns. Still, it is a guide and does suggest that 600 is conservative.
@David,
Thanks for reminding of the OI presentation at SUT, Perth. I’d nearly forgotten sharing that back at the start of November!
The geological feature represented on page 33 is in the vicinity of N43º57′ W27º6′(courtesy of historical data at my new favourite vessel tracking service). The image appears to show an AUV working in ‘un-aided operations’ over some challenging bathymetry feature.
@DennisW – might be useful to clarify ‘mapping’ vs ‘surveying’, if there is a formal distinction or other terms. I’ve tended to regard the output of MBES as mapping data, to build a digital elevation model, while SSS is surveying for detailed features but that may not be accurate.
@DennisW
When OI gets a signed contract, they will have a seat at the high stakes poker table. If successful, they could benefit a lot from the contract and the publicity, but they could also loose a bundle.
Whatever areas they choose to search, everyone hopes they will be successful. But if their search fails, this could be the last organized search by anyone for a very long time.
DennisW notes: “I imagine that MH370 would have high appeal to Stone, but I could be completely off base with that opinion.”
Perhaps, but here are some slides from a talk given by Stone in April of this year at the Naval Postgraduate School. Nearly half the slides are about MH370:
http://www.nps.edu/documents/103424533/106018074/Bayes+Search+for+Missing+Aircraft+NPS+20+Apr+2017.pdf/051a76bc-18cc-47a7-b8b8-52d92d618dfe
(there is also a video here, which I have not yet viewed)
http://faculty.nps.edu/gsois/OR%20Video/Stone_4-20-17.mp4
@Don
Mapping and surveying have broad definitions. What I am leaning toward in the case of the OI search is more like signature extraction or target detection. Analogous to people using a metal detector at the beach and digging when something is sensed. From earlier comments made by Victor I don’t get the feeling that OI will be generating what look like the Fugro pictures of the seafloor.
@sk999
Thx for the Metron stuff. I don’t recall seeing the whole PPT before, but I do seem to recall seeing the pre and post flaperon liklihood distributions on page 43. While it seems the influence of the flaperon was slightly greater than in the DSTG versions, it (the flaperon) still seems to be given a rather small weight.
In the conclusion section of the AF447 work the drift contributions seem to be portrayed negatively. Perhaps that is the reason for the weighting in the case of MH370.
Regarding the 18:25:27 BFO value:
In a recent conversation with several “inside people” familiar with the Inmarsat data, they offered an alternative explanation for the first 18:25 BFO value. They believe that BFO value (142 Hz at 18:25:27) was corrupted by a collision on the R600 random access channel. This occurs occasionally (by design) on a random access channel. The evidence for this theory is quite strong. I should have noticed this last summer when the unredacted file came out.
Note that the Rx Pwr at 18:25:27 was -52.34 dBm while the next 4 packets averaged -54.58 dBm, a difference of 2.23 dB. This indicates that there probably was a second signal in the time slot, probably overlapping for nearly all of the packet time (960 msec), but the second carrier was ~1.72 dB lower in Rx PWR, allowing the 9M-MRO signal to dominate. The effect of the second carrier was to raise the noise floor in the channel, and drop 9M-MRO AES C/N0 12.4 dB to 30.28 dB-Hz, borderline for demodulation of a 600 bps R1/2 coded channel (S/N=+2.5 dB).
Time CRC Correct Rx Power (dBm) C/No Frequency Offset (Hz) Estimated BER
7/03/2014 18:25:27.421 Yes -52.34 30.28 142 5
7/03/2014 18:25:34.461 Yes -54.59 42.65 273 0
7/03/2014 18:27:03.905 Yes -54.15 42.61 176 0
7/03/2014 18:27:04.405 Yes -54.7 42.88 175 0
7/03/2014 18:27:08.404 Yes -54.86 42.75 172 0
This means that the 18:25:27 BFO value should be ignored, but all the other 18:25-18:40 values are fine. With the 18:25:27 value ignored, all the other BFO values fit the same cold start transient observed by Holland in 6 cases, and the one on the ramp at Beijing at 12:50. https://goo.gl/arwsKH With the 18:25:27 anomalous value removed, the rest of the values can be corrected with reasonable accuracy as Bobby Ulich and I previously suggested.
Note that by the same type of analysis on the 00:19:29 R600 logon packet, it appears that there may have been a collision at that time too, but the second signal was much weaker, and thus did not cause any bit errors, or corrupt the BFO measurement.
@ALSM
That makes sense. Thx.
@DennisW said: From earlier comments made by Victor I don’t get the feeling that OI will be generating what look like the Fugro pictures of the seafloor.
I’ve inquired about this, and I now think the imaging resolution will be equivalent. However, a mapping mission requires high precision of position, as the location of all seabed features must be accurate. A search mission does not require this accuracy. Therefore, a higher level of gyro drift is permissible from the inertial navigation system in the AUV, which means longer periods are permissible before re-calibration with a surface vehicle is required.
@Victor
I was thinking the side scan debris field “imagery” would resemble figure 25 in the Stone PPT presentation posted by sk999 above.
@Victor
I am happy we are apparently finally getting around to searching for an aircraft versus accurately mapping a seafloor.
@DennisW: Perhaps Don Thompson can weigh in, but I don’t think the side scan sonar images will be any different than what Fugro was able to achieve with the towfish.
Dennis: The sidescan hardware used with the Fugro towfish may be similar to what the OI AUVs are using. I have not looked into that question. But what I do know is that the agility (maneuverability) of the AUVs is far superior. That should enable superior results.
@Victor, DennisW
The side scan sonar and MBES, for nadir coverage, that are configured on the Hugin AUVs are both Edgetech products and will deliver the same format of data product collected by Fugro’s towed and autonomous vehicles. OI’s SUT presentation lists some of the post-processing & workflow management tools that are exploited onboard.
@ALSM. Thanks for your earlier post on the singular 182527 BFO. That puts one puzzling anomaly to bed. Do you have a view (independent from Dr B’s) on the magnitude of the thermal transient distortion and therefore the likely underlying (“calibrated”) true BFO between 1825 and 1828?
I am persuaded that at 1828ff we are looking at a “true” BFO of 142 or 143 Hz. But what about the intervening ones? In your view, does the BFO tell us that the aircraft maneuvering or could it all just “transient”?
@Don
My curiosity is not sensor related, but rather how the sensors will be used. No matter. Hopefully we will get to see some OI output.
I guess it’s all up to OI now. I think all truly interested and involded have done their best in their own ways to provide them with as much as possible information.
I still hope and expect new important debris will show up in coming months.
And hopefully further new other data and insights.
And I hope in such a case they will have the flexibility to act on this.
Something the ATSB did not do in time after the flaperon was found imo.
I wish them the best of luck and wisdom.
Paul Smithson:
Here is my best estimate of the BFO corrections needed at 1825-1828. The 18:25:27 value cannot be corrected.
Time Correction
7/03/2014 18:25:34.461 -128
7/03/2014 18:27:03.905 -31
7/03/2014 18:27:04.405 -30
7/03/2014 18:27:08.404 -27
7/03/2014 18:28:05.904 1
7/03/2014 18:28:10.260 -3
7/03/2014 18:28:10.398 -3
7/03/2014 18:28:10.559 -2
7/03/2014 18:28:10.718 -2
7/03/2014 18:28:10.879 -2
7/03/2014 18:28:11.054 -2
7/03/2014 18:28:14.904 2
Here is a graphic showing the correlation between the 1250 and 1825 cold boots.
https://goo.gl/6w4LWQ
Paul Smithson asked:
“In your view, does the BFO tell us that the aircraft [was] maneuvering or could it all just “transient”?”
The affect of (horizontal) maneuvering on the BFO between 18:25 and 18:28 would be small compared to the thermal transient caused error. Therefore, it is impossible to separate the two terms. Thus, we can’t infer much about maneuvering from the BFO values during this time. The BTO values do infer some maneuvering. That said, this analysis clearly shows that there was a COLD power on transient, not a warm boot. So we can be sure there was an extended period of Left Bus power off. That is important. It also shows that values after 18:28 are essentially free of transient error.
Paul: Referencing the graphic I posted above, note that what I assumed for “steady state” was the cluster of values around 02:40. Assuming that the transient was identical at 1250 and 1825, then the 18:25:34 value is about 6 Hz high. Of course, I could have made the difference at 18:25:34 zero, in which case the values at 18:28 would be 6Hz low. Another explanation is that there was no change in BFO due to motion during the 1825-1828 period and the transient scale factor was slightly different. The ambiguity can’t be resolved to make the data useful during this period, except to confirm we have good data after 18:28 and a COLD Boot at 18:25.
@DennisW
Your comment concerning slide 25 of ‘Bayesian Search for Missing Aircraft’: yes, that’s a typical image of what is rendered by processing the side scan sonar data of a debris field scattered on a sediment formed sea floor.
In that image, the dark strips depict the nadir gap under the vehicle, the overlap between swathes is also apparent as brighter strips. The debris field spans the overlap.
If the seafloor geology is more challenging (harder, igneous boulder field, for example) then discrimination is more difficult. The response to that might be to increase the sonar frequency & narrow the track spacing.
@Don
Thx. I have zero relevant experience on what to expect.
And we’re back to Left system failures again.
Left Main AC Bus off – Satcom
Left Transponder failed – Mode S from Left AIMS
Left Systems ARINC 629 Bus – no radios
And the common thread – they’re all next to the oxygen bottle in the electronics bay, which was topped up prior to departure by Malaysia.
The bottle probably failed during the turn at Igari when the g-loading increased and at a point where the pressure differential on the cylinder was the greatest.
@ALSM: Speaking of manoeuvers at around the time of the log-on at 18:25, I am becoming less persuaded that MH370 left airway N571 before or during the log-on. If the timestamp for the last radar capture is delayed by about 2 minutes (1m 56s as best as I can calculate), then starting with the last radar position at 18:22:12 (which would be adjusted to 18:24:08), the agreement to the BTOs and BFOs is excellent by continuing at a constant ground speed of around 498 kn along N571.
Victor: If one assumes all the BFO values could be in error by +/- 3 Hz or so, the logon transient at 18:25 matches the 12:50 static logon perfectly without any maneuvers.
@ALSM: I don’t think you understand my point. If we accept the timing and position of the last radar capture, then a manoeuver is required to match the BTO and BFO of the log-on sequence, even if we only accept the final BFO. That was the genesis of the SLOP and later the turn-deceleration. However, if we accept the position but assume the timestamp was off by about 2 minutes, no manoeuver is required to match the BTO and BFO sequence, and the plane could have remained on airway N571.
@ALSM – thanks for your response. If I’m getting you right I think you are saying that we can’t be sure enough about the magnitude of the transient to reject the null hypothesis (no change of heading). But nor can we say that BFO strongly indicates a manoeuver.
@Victor – agreed with your above. As I said before, if you delay the plane by about 100s prior to 182527 then you will get a perfectly acceptable BTO fit across all of that cluster of BTOs (whether flying along N571 or indeed pretty much any heading 295 +/- 10 degrees). It is difficult to see how this could be done without a turn if we accept 182215 as correct in position and time. But if (per DSTG) everything beyond 1802 is questionable/inferred, then it only takes a slow-down to be able to match those BTOs without a SLOP. If we think [and I think this is far from proven] that the BFOs are indicative of no change of heading between 182530 and 1828 then I think it makes a slow-down a better candidate path for fitting the first BTO cluster. But at the expense of casting doubt on “LIDO” radar path positions/timestamps, particularly after the lacuna.
@Victor
the agreement to the BTOs and BFOs is excellent by continuing at a constant ground speed of around 498 kn along N571.
I am a huge fan of the N571 path although I was not able to make it work, and esssentially just gave up. ALSM’s recent eureka experience with the 18:25:27 BFO value does not alter my failure since I was assuming the 142Hz was correct. Pretty bizarre to have a value corrupted by noise come out to exactly what the value should have been under “normal” circumstances. I am buying into it, but remain puzzled by why the left bus was depowered. The FDR recovery will be an awesome sanity check for all of us.
Victor: I did understand your point. I think this new BFO transient analysis is consistent with your hypothesis. If the timestamp or position of the last radar point was incorrect, it all fits better. Of course, what happened after 18:28 up to 19:41 is still a head scratcher.
@Paul Onions
The bottle probably failed during the turn at Igari when the g-loading increased and at a point where the pressure differential on the cylinder was the greatest.
And Z, being the conscientious FO that he was, modeled that failure scenario on his simulator.
If MH370 was over Palau Perak at 18:02:59, then at a True Airspeed of 498 knots (ground speed 503 knots), it would be at 10 nautical miles past Mekar (last primary radar position) at 18:22:35, not 18:22:12. This would then make the 18:25:27 log on occurring 0.4 nautical miles prior to Nilam, which would be around the time that the Right High Gain Antenna is exposed to the satellite if the aircraft was rolling left at Nilam for a left turn to Sanob then Banda Aceh.
A slower True Airspeed of 491 knots from Palau Perak would result in the 18:25:27 log on occurring at 3.2 nautical miles prior to Nilam.
Thus another left system failed – the Left High Gain Antenna mounted on the left side of the fuselage.
And there was no flight ID as well, which is also supplied by a Left system – the Left AIMS Cabinet.
Now to confirm the BFOs at startup, all we need is data for a cold start of Satcom, inflight, with a damaged left side electronics bay and a depressurised freezing cabin – but that doesn’t exist.
Or they could release the Indonesian Military Radar.
Or the Primary radar shown at the Lido Hotel.
Or the First Officer’s phone connection log with Banda Aceh.
Or they could just search the accident spot – just beyond 40 nautical miles past Bayesian hotspot.
@Paul
Not FO. Captain.
Z was not the First Officer, he was the Captain.
The flight simulator data included flights from March 15, 2014 – which is a week AFTER the disappearance.
@sk999
Thank you very much for the links to the Stone presentation video and slides file.
A talk given by Lawrence D. Stone, on 20th April 2017, at the Naval Postgraduate School, titled:-
“Bayesian Search for Missing Aircraft”.
Video:
http://faculty.nps.edu/gsois/OR%20Video/Stone_4-20-17.mp4
Accompanying Slides:
http://www.nps.edu/documents/103424533/106018074/Bayes+Search+for+Missing+Aircraft+NPS+20+Apr+2017.pdf/051a76bc-18cc-47a7-b8b8-52d92d618dfe
This video is a “must view” for all of us here, even though it is a 667 Megabyte File. The length of the video is just under an hour (58 minutes 56 seconds). The accompanying slide file contains 52 slides, and is only 3.61 Mb.
I have extracted the audio from the video into a smaller audio only file. (107 Mb mp3).
https://www.dropbox.com/s/vdtjxjrmym1wrd9/Stone_4-20-2017.mp3?dl=0
@Ventus
Thx. In limited BW boonie land the video and audio are not an option for me. I will be most interested in the impressions posted here.
@Dennis: I thought I had decent streaming here, but it showed 90 minutes to download the Stone video. Not sure why his lecture needs higher rez than the average YouTube Russian carcrash video. In any case, better things to do!
@DennisW: Larry Stone explained the Bayesian approach to conducting a search in fairly simple terms. It is hard to find fault with a methodology uses all the available data by weighting it by its reliability, even though that weighting can be subjective for some kinds of information.
It was painful for me to watch Dr. Stone stumble through the MH370 portion of the presentation. It appeared that the modeling of the flight dynamics and satellite data leading to the path reconstructions are not his forte. He also expressed much admiration for the work of the DSTG, although he was mystified that not much more was said about the failed search other than the plane was not where they searched.
I came away from the video with a stronger feeling that the search for AF447 was child’s play compared to MH370. For AF447, they never considered searching outside of a 40-NM radius from the last known point, which is a total of only 17,240 sq km. Metron’s main contribution appears to be that they considered the possibility that the ULBs had failed, and produced one version of the heat map that dismissed the evidence associated with the failure to detect the pings.
@ALSM, @Paul Smithson, @DennisW: I think we’re all on the same page. The heads-up that there might have been a collision at 18:25:27, and Mike’s explanation of how that manifested itself in the power, BER, and BFO data, ties it all together.
@Victor
Thx. The learning curve on the sat data is indeed steep even for people closely alligned with the major concepts. For sure, AF447 was a much simpler problem. Good summary. I have confidence in your impresssions.
@DennisW
Stone makes some very candid remarks, including an admission of a mistake he made in the initial AF447 work, (at the time 26:10 mark – slides 28 – Critique of Metron AF447 Work), and beyond for a few minutes, which centers on computing Liklihood Function(s)) and lessons learned from it.
He then applied that to the work he did on MH370, which he sent to the ATSB (discussed 26:50 to 27:28) – and he said they (ie, ATSB) did not take his advice. (I was previously unaware of that !).
He discusses the creation of his “MH370 Debris Likelihood Function” from (40:30 to 42:15).
He talks about the use of oceanographic models to construct the debris liklihood functions.
Then comes the most interesting bit (from 42:15 onwards to 42:55).
He displays Slide 43. (https://www.dropbox.com/s/11t4vtrekg9xwky/Slide-43.PNG?dl=0) (485 Kb).
(42:49 – “We sent that to them (ATSB) in August of 2015 – they ignored it – fair enough – 42:46)
That was quite funny actually, you have to see his body language in the video.
Slide 43 speaks for itself, he suggested, “go further north”.
Mike Chillit will no doubt take some comfort from this.
With regard to Slide 46 (SUMMARY) – (the relevant section in the video is from 45:45 to 6:35). https://www.dropbox.com/s/erz93s31dbf0d3e/Slide-46.PNG?dl=0
He was slightly critical of the DSTG Baysean Study Report (as published) – “not the people who did the work” – in one particular respect.
He was not happy with the “they (DSTG Book) only had one paragraph on the unsuccessful search”.
He clearly thought there was something not quite rite about that.
@Ventus
Thx.
The weakness of the Bayesian approach, my opinion only, is the assignment of weigths to various inputs which are not statistically self-weighting. In many cases (the drift analytics, for example) it is a judgement call.
@Victor
If MH370 is 2-min late then UAE343 is only 4-min behind, assuming we know where UAE343 was in time. From the recent report, it sounds like the JIT/ATSB knew UAE343 path and timing and was able to pick out MH370 as a second aircraft from the primary radar. I am thinking we have two versions of UAE343 data: Flight Aware (not right on N571), and corrected (to N571).
Don Thompson kindly obtained the AMSA SLDMB data. Position and time data for 35 SLDMBs between 20th March 2014 and 14th April 2014 covering latitudes between 18°S and 60°S (yes 60°S!) and longitudes between 89°E and 104°E were provided. The SLDMBs lasted between 1 day and 36 days (average 16 days) and provided between 114 and 1,801 data points each. The tracks followed various patterns including circles, gyres (circling back over the previous track), lines, snakes or simply got stuck. 4 areas were covered from Mid Ocean around 20°S, 7th Arc north of Broken Ridge, the area around Broken Ridge and Mid Ocean below 40°S. Remarkably the 7th Arc south of Broken Ridge was not covered.
In the following link, tracks marked in white are very slow, increasing in speed through yellow and orange to red, which are very fast:
https://www.dropbox.com/s/s5n2br45xi3fs3b/AMSA%20SLDMB%20Search%20Data%20Comparison%20Map%20-%20Richard%20Godfrey.pdf?dl=0
4 SLDMBs in the Mid Ocean area around 20°S fit the average historic data from GDP almost perfectly with an average -3.5% error in speed and +1.4° error in bearing, but with a large variance. The sample is too small to draw any conclusion. 15 SLDMBs close to the 7th Arc north of Broken Ridge showed a large error in speed of around -50% and a large error in bearing of around +30°. 5 SLDMBs close to Broken Ridge showed a large error in speed of around -30% and a large error in bearing of around +80°.
The table of results is found here:
https://www.dropbox.com/s/dugvmvu4qf7ef6m/AMSA%20SLDMB%20Search%20Data%20Comparison%20Table%20-%20Richard%20Godfrey.png?dl=0
My conclusions are:
1. The drift simulator I developed appears good enough for broad brush analysis of transoceanic paths.
2. Detailed short trajectory analysis is prone to all the random processes of the ocean and surface winds.
3. There are huge variances between average historic data and actual data at a given time, except for the main ocean currents.
In summary, where the main ocean currents act, the drift is relatively fast and the speed and direction very predictable. Outside these areas, the drift is relatively slow and unpredictable, subject to local weather conditions, gyres, jets, eddies, upwelling and downwelling.
@Victor, Dr B. While I do not pretend to know much about BFOs to me it seems possible it could be shown that the 7th arc was at low altitude. The below expands on that.
@Dr B. I add some comments for you on your paper on your paper revised on 12th March and Dr Hollands of 7th February.
https://www.dropbox.com/s/hh3bcy420ghu5l0/Low%207th%20arc%20and%20Decompression%2020th%20December%20rev.docx?dl=0
@Richard
On your remark: “Remarkably the 7th Arc south of Broken Ridge was not covered.”
What about the 177 historical (20 years) GPD drifters that crossed the previous priority search area between ~32 and ~39S?
@MPat and Griffin included those in their drift-analyses back then.
It was established none of those drifters reached WA crossing the area above 36S. Only 7(@MPat) to 13(Griffin) reached WA shores crossing below 36S.
31 reached east African shores and islands after ~22 months. One reached Sumatra and the rest was stuck in the gyres of the Indian Ocean.
This studies reflect the reality to date imo. With 30 pieces of debris found on EA shores and none on WA shores:
http://jeffwise.net/2016/07/07/guest-post-where-mh370-search-area-debris-has-historically-gone/
How do you value those data?
@David
Interesting. I await @DrB and/or @Victor’s reply but cann’t wait to ask you if you assume or state a normal IFE log-on would follow within 6 seconds after the previous SDU log-ons. Not after 89 seconds.
Indeed this could have important consequences regarding the end-of-flight scenario.
What is a normal complete SDU log-on time including a IFE log-on?
That’s what I wonder about now after your post.
@David: Thank you for organizing your thoughts in the paper.
If the altitude was low at 00:19, this has two implications:
1) The 7th arc moves inward towards the subsatellite position. For instance, the 7th arc at 0′ is positioned inward about 5 NM from the 7th arc at 35,000′.
2) The possible distance traveled after the log-on becomes small.
The ATSB recommends searching at a distance of +/-25 NM from the 7th arc. The reference altitude in the past was 40,000′. The new reference altitude should perhaps be some intermediate value, such as 20,000′. If the aircraft was at low altitude at 00:19, the point of impact should fall well within the search zone.
@Ge Rijn: Others like @Don Thompson have looked at this carefully, but if you look at the sequence for the power-up and log-on around 12:50, you see that there was a log-on request at 12:50:19.735. At 3m45s later, there was a request for a T-channel assignment at 12:54:04.240, which led to the IFE log-on starting at 12:54:09.735. Presumably, the delay in the request for the T-channel assignment is due to the time for the IFE head to boot. A delay of 2m38s is seen at 18:25, but this delay was only 1m24s for the log-on at 16:00, presumably because the IFE head was already booted.
@VictorI
With 3m45seconds delay after 12:50:19 for the T-channel request and IFE log-on within 6 seconds after that, I wonder if the IFE-switch was switched to ON at that time.
Is it normal practice to switch off the IFE after- or just before landing?
And switch it on again just before- or just after take-off?
Probably @Andrew knows. Or I’m assuming nonsence.
@Ge Rijn: “How do you value those data?”
As previously commented:
https://www.dropbox.com/s/clks6522vquagwb/Drift%20Model%20Results%20using%20an%20Independent%20Drift%20Model%20-%20Richard%20Godfrey.pdf?dl=0
@Ge Rijn: “What about the 177 historical (20 years) GDP drifters that crossed the previous priority search area between ~32 and ~39S?”
That is exactly the data I use in my Debris GDP Drift Simulator!
@Richard
That’s why I wondered about your remark; “Remarkably the 7th Arc south of Broken Ridge was not covered.”
I must have interpreted this wrong then.. But also I see none of those 177 drifters passing through the area between 32s and 39S south of Broken Ridge obviously reflected in your work. It’s all centered above Broken Ridge.
How come this discrepancy? I still don’t understand this.
Please explain.
@David
I have been thinking low altitude Arc7 definition for a while now. From FS9 studies I envision the pilot if active would may want to descend into the cloud layer to hide contrails and reflectivity off the the rising Sun. Also possibly to pick a crash landing spot free of vessels. Seems to have been solid cloud cover south of about 22S, if that’s where MH370 went.
I had no prob with your Word Docx
@VictorI
I recall short before landing the IFE seat-screen goes black suggesting the IFE has been switch off. But I’m not quite sure if this happened all the time at flights.
@Ge Rijn,
Please describe, in factual terms, the consequence of the IFE/Pass switch being cycled in/out and out/in.
NO opinion, no assumptions: facts, with references.
Thank you.
@Ge Rijn
The AMSA SLDMBs have nothing to do with the GDP drifter buoys.
They are actually a different design of drifter buoy.
The SLDMBs are short lived buoys dropped by search aircraft during the aerial search and designed to last 21 days maximum.
The GDP drifter buoys are part of the NOAA program for mapping the global ocean drifts and can last well over one year.
My purpose was to use the SLDMBs to calibrate my drift simulator using data from the GDP program.
@Andrew
If I not mistaken, in the last report, ATSB inferred that they could determine MH370 radar position to MEKAR by comparing to other commercial flights. Presumably that was EK343, and presumably the JIT had a lot of interest in UAE343 several weeks after MH370 was lost. However on FlightAware all we (public) get is estimated position when the flight gets as far as MEKAR.
Q: is the exact historical position of UAE343 something that an airline would know and keep in the records? How long?
@GeRijn
RE: ”Is it normal practice to switch off the IFE after- or just before landing?
And switch it on again just before- or just after take-off?”
“I recall short before landing the IFE seat-screen goes black suggesting the IFE has been switch off. But I’m not quite sure if this happened all the time at flights.”
No, it is not normal practice for the flight crew to select the IFE switch OFF for landing then ON before/after take-off. The switch is provided to allow the flight crew to remove power from the IFE during an in-flight emergency such as a fire involving the IFE. In normal operations the switch is always selected ON.
Some airlines require their cabin crew to turn off the video side of the IFE at the purser’s station for take-off and landing, but nowadays it’s common for airlines to leave it running for the entire flight.
@TBill
RE: ”Q: is the exact historical position of UAE343 something that an airline would know and keep in the records? How long?
At the time when MH370 disappeared, there was no requirement for airlines to track their aircraft during flight. The only record an airline might have of an aircraft’s position would lie in the ACARS reports sent from the aircraft, however most airlines did not use that data for tracking purposes. I don’t know how long that data might be kept.
@Tbill
As regards to your earlier comment.
Arrival time at Vampi for EK343 on the date in question is predicted as approx 02:16:51 if that is EK343 continued along N571 at the same speed from its last known location at 02:07.
From Vampi, the predicted time it would take is approx 5 minutes 21 seconds for EK343 to be at its possible location (6 23 29.26N 96 52 6.31E) along the N571 at 02:2212.
The estimated approximate distance between the predicted EK343 and the predicted the MH370 locations at 02:22:12 is 60.18KM.
That means it would take EK343 at it’s still estimated speed along the N571 approximately 3 minutes 53 seconds to be at MH370’s last known predicted radar position.
@Victor, Don Thompson. “Others like @Don Thompson have looked at this carefully…”
Fair enough. This may be a probability issue, not either/or, in which case it might worth seeing if there is a way of establishing one solution as more probable than another.
The lack of delays in the 6 ground logs-on is striking and perhaps explained by a SDU/IFE reboot sequence but that explanation is not immediately apparent and may depend on unknowns, but probability estimates might be possible.
What also is striking is the ATSB (etc) basing their initial IFE connection expectations at the 00:19 log-in on what just happened to be the 89 secs observed at 18:25, as distinct from describing that there might be a range of possibilities. This gives me the impression still that variation from 89 secs was not something that had occurred to them, or others for that matter.
For example, lack of an IFE connection by 89 secs does not mean that the aircraft would have crashed by then: the IFE might still be stewing/brewing.
@Don. I missed previous discussion of this. Do the lack of IFE delays in the ground logs-on have a bearing? Do you see the explanation I have offered for the 89 seconds as possible and this as a probability issue?
David,
The on-ground Log Ons depicted in Holland’s paper would be expected to comprise the GES Log On (incl 2x R-ch bursts), an ACARS Link Test msg (2x R-ch bursts), then the IFE connections (one of 2x R-ch bursts, one of 1x R-ch burst + 1x T-ch burst).
The log record for an R-ch burst is recorded with a single BFO, a T-ch burst requires multiple records: one per SU including unique BFO as each corresponding interleaver block is decoded.
IIRC, Holland’s paper doesn’t attempt to identify the high level message operations on the datalink.
It seems that your narrative doesn’t consider the ACARS Link Test msg, it’s this that occurs approx 6 seconds after the completion of the GES Log On when the aircraft powers up on the ground.
At 18:25, and 00:19, no ACARS Link Test msg was initiated by the aircraft.
@Andrew
OK thanks…I was thinking about ACARS or any other data logging of the type that might be on the flight data recorder. I was not thinking of real-time tracking, but I was wondering if there is a flight data record held that could be used to figure out path/waypoints used etc. Even flight plan would do, on the assumption the flight probably stuck to the plan, as presumably no exceptions were requested by ATC.
@Joseph Coleman
But who is doing those calculations of arrival time at VAMPI? How do we even know UAE343 was heading to VAMPI? (whereas the FlightAware estimated path seems offset from N571).
If I understand, Victor is postulating above that perhaps MH370 was approx. 2 minutes later to the radar point than reported, so that would presumably take your MH370/UAE343 separation time estimate down to maybe 1 minute 58 secs behind MH370?
I am really just pointing someone (eg; ATSB) probably has an official (secret) version of this data.
@Don.Thanks for that. The 6 sec ground ACARS Link message was followed by another 2 or 3, within the space of 20 secs. Were these IFE related?
The absence of the ACARS Link test message at 18:25 and 00:19 sounds like another topic. I presume the IFE messages at !8:25 are independent them?
@Ge Rijn. “What is a normal complete SDU log-on time including a IFE log-on?”
My brief answer is don’t know.
IFE boot time apparently can have wide limits it seems. What would influence that to extend 89 secs beyond the 18:25 log-on is beyond my pay grade – and in particular whether that could read across to 00:19 as the ATSB has done and others accept. That is the question.
The SDU reboot is a minute the ATSB says but it seems likely the manufacturer’s testing they base that on did not include cold ambients, just OCXO cooling to lab ambient temperature, in other words it assumed normal cabin temperatures.
Maybe that is valid but maybe not.
Then I expect it can be delayed if it is short of input from other de-powered systems, more likely on the ground but how likely even there I do not know.
This all seems to offer scope and the incentive for a further look, though I am wary that it has been done to death earlier.
@Tbill. Thanks for your comments.
@TBill: Here is an image that supposedly shows the FR24 data for EK343 up until 18:22, along with the radar captures from the Malaysian military. This is from a comment posted on the FR24 site just after the military radar data was shown to the NOK at the Lido Hotel in Beijing.
@Victor. Your response to Ge Rijn was about IFE connection generally, “you see that there was a log-on request at 12:50:19.735. At 3m45s later, there was a request for a T-channel assignment at 12:54:04.240, which led to the IFE log-on starting at 12:54:09.735. Presumably, the delay in the request for the T-channel assignment is due to the time for the IFE head to boot. A delay of 2m38s is seen at 18:25.”
This is for the T-channel which @Don Thomson associates with the second IFE connection, for BITE.
About the 89 seconds it took from log-on at 18:25 for the R-channel initial IFE connection, for SMS/email, the 1559 log-on to the IOR satellite from the POR took 21 secs to the initial IFE transmission, though doubtless with the IFE already warm.
Don Thompson
I asked a question on the IFE-switch regarding the seat-screens going dark before landing.
@Andrew politely cleared this up for me.
If I had all the facts I would have no reason to ask any questions won’t I.
@Richard
On your comment:
“The AMSA SLDMBs have nothing to do with the GDP drifter buoys.
My purpose was to use the SLDMBs to calibrate my drift simulator using data from the GDP program.”
Oke, that’s cleared up too then. I hope you understand my initial confusion on my question and your answer:
“What about the 177 historical (20 years) GDP drifters that crossed the previous priority search area between ~32 and ~39S?”
You replied:
“That is exactly the data I use in my Debris GDP Drift Simulator!”
@ David
It seems to me you made a good point by questioning the 89 sec IFE log-on delay after 18:25 being the same after 00:19.
It might be beaten to death already in the past. But you’ll never know if you don’t ask.
@Tbill
These predictions on EK343 I posted are my own at a GS speed of 929kph as seen on from last known location and speed at 02:07 as shown on flight aware, also presuming that EK343 stayed at the same flight level too.
Perhaps the JIT may have just looked at FR and done a similar estimate. Or perhaps they may have had better info from surrounding radar or ACARS if a periodic report was due around that time. You’d have to ask them.
@Andrew,
I presume, that since virtually all airlines have QAR’s, (in adddition to CVR’s & FDR’s) which record just about everything, and are routinely downloaded after every flight for various maintenance,and crew action monitoring purposes, in aid of the flight safety departments concerned, eg hard landings, speed excursions etc , that there must be a huge database of every flight in every airline. Indeed, the QAR’s are often “pulled” by ATSB etc after incidents, since they apparently contain more parameters and at higher sampling rates than the FDR’s themselves.
Can you confirm that is basically correct, and if so, is it not reasonable to assume that the records of all aircraft in the area that night would have been routinely “pulled” by the JIT, to check their time/position reports to clarify the primary radar returns picture ?
If that is correct, why don’t they say so ?
@TBill
RE: “I was thinking about ACARS or any other data logging of the type that might be on the flight data recorder. I was not thinking of real-time tracking, but I was wondering if there is a flight data record held that could be used to figure out path/waypoints used etc. Even flight plan would do, on the assumption the flight probably stuck to the plan, as presumably no exceptions were requested by ATC.”
Aircraft latitude & longitude are two of the parameters recorded by the DFDR; however, the DFDR data is not normally downloaded unless there is an accident or incident involving the aircraft, or there is a maintenance/quality assurance checking requirement. Data that is not downloaded is overwritten within the next 25 hours of recorder operation.
Airlines also use quick access recorders (QAR) that record thousands of parameters related to the operation of the aircraft, including aircraft position. The QAR data is routinely downloaded and used in flight data analysis programs (FDAP), also known as FOQA – flight operations quality assurance. Airlines normally have strict protocols that govern the release of QAR data and the purposes for which it may be used. Where I work, the data is retained on a server for 42 months then transferred to tape for longer term storage.
@ventus45
Sorry, I didn’t see your post before I posted my reply to @TBill above. Yes, what you said is basically correct; the QAR typically records far more data than the DFDR and it is routinely used in incident investigations. However, as I understand it, that data would only be used during the investigation of an incident related to the aircraft from which the data was recovered. I doubt the Malaysian ICAO Annex 13 Safety Investigation Team for MH370 would have ‘pulled’ the QARs of other aircraft in the area that night as a matter of routine. Indeed, they may not have had the authority to compel airlines in other jurisdictions to release that data.
@ventus45, @TBill
“is it not reasonable to assume that the records of all aircraft in the area that night would have been routinely “pulled” by the JIT.
It would have been a lot more reasonable for the Royal Malaysian Air Force to ensure its records, acquired from its Air Defence Surveillance assets, were appropriately secured for later review.
During early and mid-March, the JIT had not yet been contrived. Malaysia’s Air Accident Investigation Bureau did not exist. DCA & politicos were running the operation.
David wrote “About the 89 seconds it took from log-on at 18:25 for the R-channel initial IFE connection, for SMS/email, the 1559 log-on to the IOR satellite from the POR took 21 secs to the initial IFE transmission, though doubtless with the IFE already warm.
That statement is incorrect: the 15:59 IOR Log On completed at 16:00:13, the IFE connections were initiated at 16:01:16, a delay of approximately 60 seconds after GES Log On complete. This delay of approximately 60 seconds is consistent throughout the Log during Log On renewals and region handover Log Ons. I have to say ‘approximately’ because the IFE correspondence is regarded as a passenger amenity service and attributed the lowest priority on the datalink, any other coincident packet data correspondence will be processed before IFE.
I advise using the LOA-LOA exchange, indicating GES Log On complete, as the marker for delays to subsequent events.
Holland’s depiction of BFO associated with received bursts in Table III, figs 8 & 9 really cannot be used to infer the messaging events taking place. It certainly would be useful to have visibility of the GES Logs, from the AES’ LOR through to both IFE connections completing, that correspond to events used in Holland’s analysis. If he was to update this work, including such information would be important to aid a more complete understanding.
@Ge Rijn. “It seems to me you made a good point by questioning the 89 sec IFE log-on delay after 18:25 being the same after 00:19.”
Thanks. Still looking into it.
@Victor. About the 12:50 boot, apparently cold, the delay from LOA to 1st IFE transmission was 5.5 secs, which is near enough the 6 secs from those second to third transmissions I quoted for the the DSTG ground transmissions. It appears to be an apparent example where the 6 secs can apply to IFE from a cold boot.
So the equivalent to the 18:25 89 secs during that log on was 5.5 secs.
There was a long delay then as you say to the second IFE transmission; 3mins 32 secs between the two so maybe a cold IFE boot delays just the second and is something to do with BITE.
In the 1559 LOR, with IFE warm one presumes, that delay from the 1st IFE to 2nd was 6 secs.
At 18:25 it was 62 secs.
On the face of just this one example I would speculate that 18:25 was a parallel to 12:50, if so removing a boot as cold as at 12:50 as a reason for the 89 secs delay.
@David wrote “About the 12:50 boot, apparently cold, the delay from LOA to 1st IFE transmission was 5.5 secs, which is near enough the 6 secs”
Incorrect: it is consistent throughout the full log for 7th March that no IFE correspondence occurs before 60 seconds has elapsed since the LOA exchange marking completion of GES Log On.
The Log Ons set out in Holland’s paper, numbered 1-6, were recorded after 9M-MRO was idle, on the ground, at KLIA. Holland was analysing ‘cold’ starts, one assumes he has correctly identified the Log Ons that correlate to ‘cold’ starts, ie aircraft powering up from total shutdown.
The event that consistently occurs 6 seconds after the GES LOA is the ACARS Link Test message, not an IFE related transmission.
Again, you write “In the 1559 LOR, with IFE warm one presumes, that delay from the 1st IFE to 2nd was 6 secs.” Incorrect. Following the 15:59 Log On, the IFE system opened its first connection after 1m03s.
If one wishes to assume anything: the minimum delay from LOA to IFE opening a connection over the SATCOM link is 60 seconds.
Figure 9 in Holland’s paper translates the ‘curves’ in the ‘Y’ axis for BFO measured at the LOA burst. To better represent the elapsed time after Log On, translating the curves so that the LOA is shown as ‘X=0’ would have been more useful.
@Richard
I’m sure you’ve seen this drift-study before:
http://www.tandfonline.com/doi/full/10.1080/1755876X.2016.1248149
What I found remarkable is the only one historical (undrogued) drifter that actually beeched on Reunion (figure 7 the green trajectory) passing the search area between 35S and 40S, crossed this area at it’s most northern part just under 35S heading north-west then turning east and passing north of 35S throught the current priority search area again.
@Ge Rijn
Yes thanks, I had seen the drift study before.
The paper is referring to the undrogued GDP Buoy 70854, which followed the 7th Arc from 40°S all the way up to 28.873°S before heading off to Reunion.
The full track is shown in the link below and took 802 days:
https://www.dropbox.com/s/kmr89njymz1ybha/GDP%20Buoy%2070854.png?dl=0
@Victor @Joseph Coleman @all
Joseph said we would have to ask, and that is basically what I am suggesting. ATSB in its recent report, if I recall correctly, said MH370 location at MEKAR was based off other commercial flights on N571.
So the first question to ATSB would be: was that UAE343 that was used to gauge MH370 location? Then the next questions are: was UAE343 on N571 the whole time (as Victor points out FR24 seems to clearly show UAE343 started there, whereas FlightAware is fuzzier estimated path), amd what was the separation time at MEKAR?
Also another data point I’d like to have is the ATC transcripts for UAE343, whereas the FI gives us mainly MH370. if there were commands given to UAE343 around 18:20-18:45 that could be interesting information.
@Victor question from above: Are you postulating one hypothesis that the Inmarsat 18:25 data suggests the radar data of MH370 was off about 2 minutes? if so I am thinking that scenario puts MH370 and UAE343 closer together.
As a further minor point about flight data logging, part of my interest is end-of-flight, there was another UAE flight EK425 out of Perth that in my estimates crossed Arc7 at about 00:19 at 20S (on airway L894). I would give lower priority to 20S on Arc7, but if that unexpectedly turns out to be the crash location, then the UAE425 whereabouts will be a hot question.
@Richard
Thanks. The full track time was not mentioned. Therefore I thought the full track time was near 17 months/~500days (flaperon discovery).
In fact this buoy crossed the 7th arc three times;
– just under 35S heading north-west
– then turning due east crossing the arc around ~34/33S
– then later turning west again crossing the arc around ~30S
Is it possible for you to find the ~500 days point/latitude in its trajectory? I don’t have/find access to the data of specific buoys.
I’m just curious. If it’s overdue or considered irrelevant, don’t waste your time.
@TBill asked: Are you postulating one hypothesis that the Inmarsat 18:25 data suggests the radar data of MH370 was off about 2 minutes? if so I am thinking that scenario puts MH370 and UAE343 closer together.
Yes, I think that’s possible. Unfortunately, I have little hope that we will ever learn more about the Lido Hotel radar data. The Malaysians refuse to acknowledge that the captures represent MH370, despite the timestamps and positions that are coincident with MH370. The path of EK343 does not match the radar captures near Pulau Perak, so we can be sure that at least some of the captures are not EK343. If the captures are MH370, I’m not sure we can be certain the timestamps are accurate to more than minutes. And if the timestamps are off by two minutes, no manoeuver from N571 is required to satisfy the satellite data.
@Richard @others
I like to add; this buoy in fact underlines the CSIRO/Griffin drift-pattern they based their latest priority search areas/hot-spots on.
First the small band current to the west around 35S, then a short turn to the north-west, then a sharp turn due east around 33S.
For what it’s worth. Pure coincidence probabbly but remarkable imo.
@Victor
P.S. – if OI cannot find MH370, one lesser hot spot is up on the flight paths around L894 20S-22S. One hypothesis could be MH370 was sticking to flight paths at the end of flight.
@Victor
I know many of the questions are rhetorical (secret info we will not be given) but it is perhaps instructive to itemize what we are not being told. In my mind the one of the biggest secrets is what info Singapore radar plane found out.
@TBill
I think the biggest info vacuum is the detailed forensics relative to the flaperon condition. Adding data points to the 18:25 to 19:41 path is important, no doubt, but not as valuable to the search strategy IMO.
@Ge Rijn
Here is a link to the GDP Buoy 70854 track with the start point marked in Green, the end point marked in Red and 100 day markers in Grey along the drift path.
The buoy took 435 days from when it last left the 7th Arc at 28.873°S on 6th September 2008 to arrive in Reunion on 15th November 2009.
https://www.dropbox.com/s/w6cf7cf20gatvzz/GDP%20Buoy%2070854%20100%20day%20steps.pdf?dl=0
@David,
Thanks for your many probing questions. I have been working since to provide solid answers. I will post in aday or two.
@ALSM,
I have modeled all the transient BFO data after eliminating the 18:25:27 BFO. The results are interesting. On thing I can say for sure is that the 18:25-18:28 BFOs do not match the other 6 Log-Ons, as you suggested aligning it and assuming there was no maneuver affecting BFO then. There are several alternative possibilities to secure consistency. It is too complicated to explain in a short post. I will post a link to a short paper on this subject as soon as I finish it.
DrB: The 1250 and 1825 transients match very well. What were the times of the 6 others?
@Don Thompson. Thanks for the information that ground logs on 1-6 were cold starts.
I had not seen your 3:40 am until just now so missed it in posting my 4:16. I fear that might have suggested some fixedness on my part.
The interpretation of what were IFE connections I have taken from the timings at page 9 of the Definition of Underwater Search Areas p9 table for both IFE connection completions at the 18:25 log-on.
I have taken those exact timings, BFOs and BTOs to the data communications log update. They correspond to “R-Channel receiver “Eleven Octet User Data”, 176 BFO for the SMS/email connection and “R-Channel receiver RX Ox22- Access Request (R/T Channel) 144 BFO for BITE set up.
I have found these markers in the same sequence in other logs on and noting the description of the second I have assumed that I could transfer the 18:25 meanings to those, though the analogy I drew apparently is oversimplified and flawed somewhere.
I will look into your 3:40 and 5:54 emails further thanks.
@Dr B. Noted thanks.
I see Fugro Equator has been put to good use — discovery announced for WWI Australian sub off Papua-New Guinea.
Still an 800-ton chunk of metal at 1000 feet is shooting fish in a barrel in comparison to MH370.
@ALSM,
You said: “DrB: The 1250 and 1825 transients match very well. What were the times of the 6 others?”
Actually they don’t match the rest of the data (#1-#5) at all. You are creating a transient error curve from the 12:50 and the 18:25 Log-ons only and ignoring all the rest. Your transient curve does not match #1-#5.
I am using the same seven events used by Holland. His #6 is the 12:50 Log-on, and his #7 is the 18:25 Log-on.
My approach is this: Fit a transient curve to events #1-6. This matches all 6 events very well. Then apply that curve to the 18:25 (#7) event (which may contain a maneuver). It can match, but not at the time alignment you suggested, unless there is a compensating step from an ongoing maneuver at 18:27.
If you assume no maneuvers (a constant baseline) for #7, it matches only if the BFO is rising rapidly at 18:25:34 to a much larger peak which decays back down by 18:27. You’ll see what I mean when I post the graphs.
The bottom line is that there is one good solution for 18:25 event per assumption regarding an ongoing maneuver. Different maneuver assumptions yield different alignments. An interesting case is when the BFO has a step caused by the first turn of a SLOP. If that occurs at 18:27, it looks like your proposed alignment (of the 273 Hz at the peak of the transient overshoot) can work. With no maneuvers, that alignment can’t match the other 5 events. In that case the 273 Hz is on the rising edge of a larger overshoot peak.
@Richard
Thanks for your detailed graphic on GDP buoy 70854.
I see you also put dots on the 7th arc to mark the degrees latitude from 40S on which is helpfull too.
Remarkable to see how long that buoy got stuck in that eddy around ~27S and how it speeds up when leaving it, crossing the IO from that point in ~300 days till beeching on Reunion.
I understand this is only one example out of many possible varieties.
But as a ‘real-live’ example it proves the flaperon could have taken a trajectory like that in the constraining time frame.
And that it’s more likely then it originated from between ~35S and ~29S than from latitudes south of 36S (considering the time frame gets too long).
@Don Thompson. So as I understand it the “Eleven Octet User Data” of the ATSB’s IFE SMS/email connection at 18:27:03 would apply to an ACARS link message in normal logs-on and its use at 18:27 was incidental. Also there are no IFE connections short of 60 secs after LOA, time variation in them being from low priority on the data link.
That leaves open what intervening transmissions there should have been aside from ACARS in those 89 secs and why there was none, bearing in mind that the 1-6 ground logs-on transmitted 3 – 4 times within 21-25 secs (5 of them) 35 secs (the other) of the 18:25 LOA. Any ideas please?
Other questions I mutter about, in case you or others care to comment, are:
• Why the ATSB would expect IFE SMS first ground connection specifically by 89 secs from the final LOA? I assume that to be a continuing error. As in my discussion paper, such non-connection has been advanced, including by them, as an indicator that the aircraft had crashed by the 89 sec mark.
• Following from any answer you have to the 18:25 intervening transmissions what transmissions should have been expected after the 00:19 LOA, within 89 secs, or even beyond?
• Another stray thought comes from your comment that IFE correspondence being a low priority: Inmarsat might know whether that could explain the 89 secs of this instance, noting it the data link was not busy at the AES end.
• What further investigation is warranted?
When you say you are unable to infer messaging events from Holland’s Table 3 and figs 8 & 9, you do infer from that all the same that the third transmission of his ground logs-on is an ACARS link test message?
To me Holland’s fig 9 has the LOAs at X=0 now. If you mean Y=0 what would could he use for 7’s? For my part I would like to see them clearer and/or the data tabulated. Magnifying has helped little.
@Richard
I saw later the dots on the 7th arc start from 41S and the 33S dot is right on Broken Ridge. So my interpretation shifts one degree but in essence stays the same.
@Ge Rijn
As you say, the GDP Buoy 70854 is just one example. More importantly, Reunion is just one of the destinations.
My previously published drift analysis is an attempt to use the average historic data from all the transoceanic GDP Buoys in the Southern Indian Ocean to fit all the locations where MH370 floating debris has been found:
https://www.dropbox.com/s/clks6522vquagwb/Drift%20Model%20Results%20using%20an%20Independent%20Drift%20Model%20-%20Richard%20Godfrey.pdf?dl=0
The more recent analysis of the AMSA SLDMBs is to see how close actual SLDMB data fits the average historic GDP data.
https://www.dropbox.com/s/s5n2br45xi3fs3b/AMSA%20SLDMB%20Search%20Data%20Comparison%20Map%20-%20Richard%20Godfrey.pdf?dl=0
Unfortunately there appears to be a good fit between the actual data and the average historic data, only in the areas where the main ocean currents influence the drift. However, once an item of MH370 floating debris reaches the South Equatorial Current, then it will likely remain under the influence of a highly predictable ocean current. If MH370 floating debris reaches Madagascar or beyond to the coast of Africa, it remains under the influence of highly predictable coastal currents. This explains how MH370 floating debris has been found in all 25 locations from Rodrigues to Madagascar and Tanzania to South Africa and an oceanographer like Charitha Pattiaratchi can tell Blaine Gibson exactly where to look.
As you have noted, in the example of GDP Buoy 70854, it can get caught up in gyres. It took 141 days to travel from 27.537°S 96.954°E to 27.815°S 94.074°E, just to the west of the 7th Arc, a distance of 154 NM at an average speed of 0.446 knots (max 1.603 knots) but going around in circles. In a straight line, 154 NM would be covered in 14 days. 127 days were wasted in the gyres.
Because of the uncertainties surrounding the currents, eddies, gyres, jets, upwelling, downwelling and surface winds, I do not believe that it is possible to be precise about the MH370 End Point from the drift analysis. In my view, 30°S +/- 1° is the closest the transoceanic drift analysis can get, as the timeframe is between 508 and 1056 days for the various items of MH370 debris found.
Victor’s analysis above of the Aerial Search, using David Griffin’s data, after an elapsed time of 21 days is much more precise and concludes a possible MH370 End Point of 29.7°S. In this case, we do not yet know whether any of the debris are from MH370.
@Richard: Are there any drifters that cross the 7th arc that land close to Mossel Bay, South Africa?
@Richard
Yes thanks, clear extra info.
It still makes me question the ~30S latitude as origin of the flaperon and other found debris though. I’ll explain why:
First buoy 78504 shows a rather similar trajectory as the average historical drift track in your drift model (figure 1) if you leave the gyre out (the 127 days).
Starting from 7th arc at ~29S it would have taken this buoy then 435-127=308 days to beech on Reunion. This would have been 200 days too early.
Including the gyre-time this is 508-435=73 days too early.
If you would place the start location of buoy 78504 farther south between 32S and 34S, I think this would make a better fit.
Also considering your average historical drifter track starting at 30S.
This track clearly shows it ends way north of Reunion.
Placing the starting point of this track at ~33S also makes a better fit with ending up around Reunion within the timeframe constraints imo.
I probably simplify the situation to extend but hope you get my point and give your opinion.
@Richard
Now @Victor mentions Mossel Bay an earlier thought about timeframe constraints comes up in my mind.
Only the flaperon and the Roy-piece were found with an abundance of fresh barnacles. Proving their arrival times must have been very close to discovery time.
Considering the flaperon was the first piece found and one of the closest to the 7th arc and the Roy-piece was the second piece found which still is the most far piece, I think the timeframe constraints in any case are also defined by those two pieces (~508 days and ~655 days).
I think most (if not all) of the other found debris arrived within this timeframe (between july 2015 and december 2015) but was found (much) later. Maybe only the Rodrigues-piece arrived earlier than the flaperon but considering the distance between Rodrigues and Reunion this would make not a lot of difference in the timeframe.
@Richard @VictorI
I’m in to this now so I go on for a while..
On @Richard’s comment:
“Victor’s analysis above of the Aerial Search, using David Griffin’s data, after an elapsed time of 21 days is much more precise and concludes a possible MH370 End Point of 29.7°S. In this case, we do not yet know whether any of the debris are from MH370.”
Yes, this is clear. We also don’t know if the flaperon-like ‘blue panel’ and the other 13 pieces of debris the RNZAF spotted when entering a ‘debris field’ around 32.4S/97.9E were from MH370.
The likelyhood of this debris field including the ‘blue panel’ has imo at least a same degree of ‘possibility-value’ as the pieces north-west of 30S and the CSIRO/Griffin satelite images.
If not more. The ‘blue panel’ is the only piece that can be identified as maybe a piece from MH370. It has stikingly similar features to a left wing flaperon. It has not been identified yet and therefore can not be discarded as ‘fishing gear’ or something else.
@Ge Rijn
I think we have to be careful regarding the blue panel (size): I made a quick estimate just now (I’m sorry for the delay, I’m too busy these days) and I estimate the short side to be 30 – 40 cm (based on 125 mm focal length, 23.7 mm image width and 200 m distance). I hope those who already did these calculations before can (finally) share the numbers/results so we can check..
I expected the opening angle to be much larger, but apparently it would only be 11 degrees (horizontal).
As said before I would have liked to test this “procedure” with an object of known size.
I like to stress this further.
If the ‘blue panel’ is actually a MH370 piece and the spotted debris field was also, at ~32,4S/97.9E this would fit all constraints.
If coming directly from the 7th arc it would have drifted ~130 miles from around ~33S in 20 days. This is within probability.
Regarding the average historical drift track @Richard provided concerning a flaperon track starting at 30S, ~32.2S till ~33S makes a much better fit imo regarding timeframe constraints and ending near Reunion.
And it still can explain the Roy-piece at Mossel Bay.
The location/latitude is within Inmarsat flight-path constraints.
The location/latitude is within fuel constraints.
@Niels
I did some research too. The censor cropp-factor of 1.5 only affects the angle of view not the zoom-factor.
125mm focus lenght increases the size of the object by about 20% compared to seen with the naked eye (I tested it with a variable zoom lens Mamiya 80-200)
The angle the image 5809 was taken from through the window of the plane is unknown but I estimated about 40/50 degrees.
Given the altitude of the plane was 600ft I estimate the distance to the object was about 700/800ft.
I think this would compensate for the zoom effect of the 125mm.
Now I take a break..
@Victor asked: “Are there any drifters that cross the 7th arc that land close to Mossel Bay, South Africa?”
The simple answer is no.
There are plenty of drifters that start further north on the African coast around Madagascar or Mozambique and follow the coast down to the Mossel Bay area.
There were 2 drifters that started near the 7th Arc and ended up off the coast of South Africa:
1. GDP Buoy 22412 started around 21°S just west of the 7th Arc and tracked across the Southern Indian Ocean passing Reunion and the southern tip of Madagascar to South Africa.
https://www.dropbox.com/s/obi36eqhxt1la11/GDP%20Buoy%2022412.png?dl=0
2. GDP Buoy 35927 started around 31°S at Broken Ridge just west of the 7th Arc and tracked across the Southern Indian Ocean passing south of Reunion and the southern tip of Madagascar to South Africa.
https://www.dropbox.com/s/8acwf94kobfaa1z/GDP%20Buoy%2035927.png?dl=0
David wrote “So as I understand it the “Eleven Octet User Data” of the ATSB’s IFE SMS/email connection at 18:27:03 would apply to an ACARS link message in normal logs-on and its use at 18:27 was incidental. ”
I really do not understand what you’re attempting to infer, above.
There is no similarity between the message content of an ACARS msg and ISO8208-SSN msg. Short messages of either ACARS or ISO8208-SSN correspondence will be exchanged using R-ch bursts.
An R-ch burst is merely a container.
How did the search end up centered around latitude -30 degrees in the first place? It is a difficult question to answer. Here is what I have culled out of the Final Report and prior reports.
Initially (Mar 18, 2014), the search area was defined by Boeing’s performance limit calculations. These calculations assumed that the plane flew at 473 knots (ground speed) to the final RADAR point, then immediately turned south at some particular altitude and speed until it ran out of gas. No winds. No deviations from ISA. No PDA. The only constraint was that the plane hit each arc at the right time, which meant heading changed at each arc crossing, and that it reach the 6th arc. The longest routes reached beyond -40 degrees. The search was initially centered near the 6th arc and -40, then diverged East chasing phantom satellite sightings of debris. (Both satellite and aerial searches were performed.)
On March 28, the search area suddenly shifted to the North, by over 10 degrees of latitude. What changed? The explanation given in Table 13 (and in Appendix 2) is that:
1. It was clear that the plane flew at 500 knots (not 473) over Malaysia while under primary radar coverage, increasing the fuel burn rate;
2. The BTO and BFO at the 1st arc were used to infer that the plane made a turn right after the last radar point; it was then assumed that it turned South right after the 1st arc. Thus, the FMT was now later and further North than was previously assumed;
3. The plane ran out of gas at the 7th arc.
One can find multiple figures of the new fuel endurance curve relative to the 7th arc, but Fig. 14 of the Final Report is good enough. The fuel endurance curve now ends at around -32.5, not around -42.5 as it did previously. However, items 1 and 2 are not enough to shift the curve as much as 10 degrees North. Further, the fuel endurance curve ends when it is still outside the 7th arc. Surely there must be viable routes that reach further South. Finally, if you check out the fuel endurance curve in the June 2014 report (Fig. 20), the curve has essential the same shape and location as the one from March 28, but now it continues to -37.5 latitude. Why did the March 28 curve end 5 degrees further North?
I can only offer a conjecture. It is possible that Boeing ran a coarse grid of models in altitude and speed, and when it applied the 7th arc cutoff criterion, it excised enough routes and its grid was coarse enough that the longest route ended at -32.5. Later, it repeated the exercise with a finer grid and recovered the missing longer-range routes.
Thus, it appears that the search around -30 was more a matter of happenstance than any data analyis that was firmly grounded in reality.
It should be noted that Boeing’s attention was soon diverted by the first (flawed) Inmarsat BFO analysis, which tried to provide an independent method for determining the location of the aircraft along the 7th arc. As is now known, Inmarsat’s analysis was not ready for prime time, and it contributed to (but was not the sole cause of) about 2 months of wasted effort in the search around the Ocean Shield position, which was yet another 10 degrees of latitude to the North.
The stumbling return of the search zone back South was entire due to Inmarsat’s improved (and ultimately successful) understand of how its own ground station receiver signal chain worked.
@sk999
I agree, and came to the realization fairly early on that BFO based conclusions were suspect. Likewise flight path assumptions were simply that, assumptions. The reality is that an underwater search should never have been started. The SSWG knew or should have known the frailty associated with their analytics. The people making spending decisions were simply not capable of understanding what the techies were talking about, and the techies were less than forthright in characterizing the strength of their recommendations. So here we are three plus years later in essentially the same situation. At least we have some debris finds that help.
I really believe that the biggest shortcoming of the effort to date is the lack of a world class scientist to anchor the “official” analytics. Feynman and the o-ring disaster is a case in point. As far as the SSWG is concerned we don’t even know who these people are, simply references to the organizations represented. Who is/was the spokesperson for the SSWG? Damned if I know. If you are a principal investigating a disaster why not put your name to it? We all do.
@Don Thompson. You said you did not understand what I was trying to infer from my, “So as I understand it the “Eleven Octet User Data” of the ATSB’s IFE SMS/email connection at 18:27:03 would apply to an ACARS link message in normal logs-on and its use at 18:27 was incidental.”
In a post the previous day, 21st December at 6:15 PM I briefly explained how I had arrived at my likelihood that the 1st BFO after the Holland ground logs-on LORs and LOAs signified IFE SMS/email connection completion. The above quote followed that and your response to it.
I will now spell it out more fully. The ATSB’s 18:27 IFE SMS/email connection confirmation timing matches the “eleven octet user data” “SU type” in the data comms log update. That was the first BFO after the 18:25 LOA and 89 secs behind it. It was followed 62 secs later by another transmission, that completing IFE BITE connection.
The 7th March cold 12:50 and warm 15:59 logs-on were also directly followed by a transmission of SU type “Eleven….” etc and thence another of the same SU type as for the post 18:25 log-on’s IFE BITE connection.
Since you told me that the third transmission in the ground logs-on was for ACARS and that in all logs-on, including those at 12:50 and 15:59 on 7th March, IFE connections would be deferred to at least 60 secs after log-on it was apparent that what I deduced as IFE SMS/email messages, like the first after the 18:25 log-on, were evidently ACARS. I really did not try to infer anything, only explain.
I should add that the above 18:25 nomenclature is that applying also at what you indicate was the actual 15:59 log-on IFE connection at 16:01:16, in place of my 16:00:17, both being followed by an R-Channel Ox-22 Access Data Request (R/T channel), which at the 18:25 log-on signified IFE BITE connection.
I have no idea what an “ISO8208-SSN” message is or how it bears on the above.
Returning to the main purpose of my last post in case that too was unclear, leaving aside that first transmission after the ground LOAs as being ACARS link related, the Holland ground logs-on, all cold, had several transmissions which went forward within 35 secs, none of which would have been IFE related. That takes at least 60 secs from LOA, following your advice as above.
From the time of the 15:59 warm log-on to when you said was the first IFE activity there were 12, all within a minute. After the 12:50 cold log-on there were 4 within the minute.
I do not know their purpose and since you say you cannot infer that from the Holland data, neither evidently do you. Unless they are all ACARS related one presumes they would apply still after in-air logs-on, except for that missing ACARS first transmission. Hence my question as to whether you can say why other than the first they were missing in the 18:25 log-on.
I would much appreciate a response to that and any comment on the supplementary points and questions.
Correction to my previous post about your preference that Hollands fig 9 had the LOAs at X=O I questioned whether you meant Y=0. I still do but whereas I raised the difficulty about the 7th log-on but did not make clear I meant its LOR, that now being seen as an unknown outlier.
@DennisW
Where’s Feynman when we need him? It is fun to think of what he might have said about MH370. Presumably the cause would not be a mystery to him and he would say so in the most clear language people would understand. Why can’t Hawking stand in for Feynman on this matter?
@TBill
It is frustrating at least for me. A lack of transparency has characterized this tragedy from the get-go. I blame it on Malaysia who is the “official” source of information. Australia, France, others,… have to respect Malaysia’s role since international relations dwarf a missing aircraft.
I find myself wondering if the OI “negotiations” are merely an excuse for Malaysia to delay a final report which is due soon.
@Niels
I did some further testing with the 125mm focal lenght on same objects viewed at ~600 and ~800ft away.
I find again that at ~600ft the object is enlarged by ~20% but at ~800ft (~ estimated distance image 5809 was taken) this enlargement decreases to the level as seen with the naked eye at ~600ft.
I still think the size of the ‘blue panel’ fits the overall dimensions of a flaperon.
I also think that this Orion crew would not have made an estimate of 4m2 on the size of the object when it was actually less than 1m2 as your numbers on the ‘short side’ suggest.
I would appreciate your thoughts on this.
@Ge Rijn
“I did some further testing with the 125mm focal lenght on same objects viewed at ~600 and ~800ft away.
I find again that at ~600ft the object is enlarged by ~20% but at ~800ft (~ estimated distance image 5809 was taken) this enlargement decreases to the level as seen with the naked eye at ~600ft”
It is difficult for me to understand what you did exactly, perhaps you can explain more in detail. Regarding the calculation: I asked a colleague who is good at optics (I’m not) and he came to similar estimates as I did. He used a slightly different procedure (through pixel size and magnification factor). Of course there could be an error in the parameters that we extract from the image meta data, so it would have been good to do a check with an object of known size.
@VictorI, other path modellers
It is quite common to take 19:41 as starting time for path reconstruction. I’m wondering what for example your fig. 2 from the previous post (“Match to satellite data after 19:41 for great circle paths”) would look like if you would start calculations at 20:41. The reason I ask is that I typically see path curvature and lower than average GS in roughly the first hour after 19:41 when generating paths with my tool.
Did you ever try/consider later starting times?
@Niels: Not really, considering that starting at 19:41, there are great circle paths with autothrottle engaged that match the satellite data over a large range of ending latitudes. However, I do think there is merit to what @sk999 has done–ignore all but the BFO at the 6th arc to determine the range of latitudes on the 7th arc. If we understood the uncertainty in the BFO measurement at 00:11, we would be able to calculate the uncertainty of the final position along the 7th arc.
@VictorI
I agree that the approach introduced by @sk999 is really interesting. It triggered me to try to derive two of these linear relations (BFOerr vs. Lat), one by path curvature minimization and one by forward acceleration minimization. The intersection of these two lines could then in principle give a unique value for both BFOerr and Lat. It turns out to be rather tricky in practice: the two linear relations I find are almost identical, so the result probably not very accurate.
@David,
Here let me address several key points you raised previously.
You said: “Bear in mind that it is unrelated to however cold the SDU was at 18:25 because log-on request (LOR) and acknowledgement (LOA) had been transmitted already so its OCXO was warm enough; and it was not because an operating band limiter was at work, for there are examples of transmissions of higher BFOs.”
You also said: “Bear in mind that as above a colder OCXO in the air would not cause it.” In this case by “it” you seem to be referring to a longer delay for IFE transmissions.
You also said: “However an ALSM post of 18th December has confirmed the DSTG view that the 18:25 LOR is an outlier and so that steep Ulich transient no longer has a basis.”
My most recent work over the last few days has demonstrated that two solutions are possible when the 18:25:27 BFO is excluded. If you assume that the “baseline” of the 18:25 event is flat, i.e., assuming that there were no ongoing maneuvers then, and using a transient curve which is fitted only to Holland’s ground Log-ons #1-6 and applying that same shape to the 18:25 in-flight Log-on #7, then the only fit possible still has a steeply rising error at 18:25:34. On the other hand, if you assume that the BFO error at 18:24:34 is either flat or declining, then you can only conclude that the baseline is not flat. The simplest (second) solution in this case is that the BFOs at 18:27 are higher than the baseline at 18:25/18:28 by about 24 Hz. This appears to also match a SLOP with the course being 296 degrees true (parallel to N571) at 18:25:30 and 18:28, while at 18:27 the course was close to due north (in the middle of the SLOP). I am in the process of checking this fit using the new, lower turn rates (0.6 deg/s) from a 15-degree bank angle limit. It looks like it can match the BTO and BFO, but the timing of the SLOP must fit in a brief time window for this to work with a low turn rate.
First, I think there may be some misunderstanding of how the SDU inhibits transmissions during BITE, which can occur due to several causes, one being when the OCXO temperature is not sufficiently close to the operating temperature. I am not aware of a tight “operating band limiter” in frequency space which would affect the MH370 log-ons, although there may be some coarse limiting in the SDU software based on the calculated frequency just to keep it within some receivable range. My understanding, and DonT may be able to correct me if I have got this wrong, is that there is a test of the OCXO temperature error. If the magnitude of the error is too large, the BITE fails and the transmissions are delayed. This probably occurs both when the OCXO is too cold and also when it is too hot. This temperature error “window” is a probably a small fraction of 1C because the AES must control this frequency effect to within several hundred Hz at most, and the slope of frequency versus temperature is quite large (about 3 kHz/C). Over this very small temperature range, the frequency versus temperature characteristic of the crystal may be adequately modeled as being linear.
Additionally, we know that every SDU has a temperature control system with a unique step response curve. So, unfortunately, you can’t take another SDU, test it at various cold starting temperatures, and expect to replicate accurately the unit in 9M-MRO. I have requested ATSB/DSTG and Dr. Holland specifically, on multiple occasions, to arrange cold testing if it were possible to obtain a SDU and tune its servo response to fairly closely match the data we have for the 9M-MRO SDU. That has not happened so far, and it probably won’t be done at this point.
ALSM and I are discussing on the side how to align the 18:25-18:28 log-on CBFOs with the other ground-based log-ons. His alignment may be eye-catching, but, because the data are very sparse, it is not unique, and he did not include the other ground log-on data in the same plot, which do not match either of these two cases. In addition, if you fit one curve to both the 12:50 and the 18:25 log-ons, you are inherently assuming the transient curves always have the exact same shape and amplitude, with only one variable available for each event for alignment in the time direction. This assumption is not borne out in the other ground log-ons. In summary, none of the ground log-ons exactly match each other. With only the one 18:25:34 BFO value, and with no ground log-on data 90 seconds later to compare with, one can always match the first point of the 18:25 Log-on with another curve so long as it also has a high peak error. This apparent match of one pair is suggestive, not but not proof of identical transient curves.
There appear to be two classes of log-on transients. Events #2-5 have about 1-hour depowered times and also have about 60-80 Hz true peak errors (based on my recent fits; note that the observed peak error may be smaller than the true peak error depending on the exact relative delay time of LOR compared to the first zero-crossing of the servo controller). I call this group Class A. Events 1 and 6 have much longer (4-6 hour) depowered times and much larger peak transient errors (130-150 Hz), and these I call Class B events. Event #7 at 18:25 is curious because it has the same amplitude of peak transient error (about 130 Hz) as Class B, but it has the short ~1-hour depowered time of Class A. I am wondering if the thermal time constant for the 18:25 log-on was shortened because there continued to be forced air circulation around the SDU in flight when the left AC bus (and the SDU) was unpowered. That would cause a significant reduction in OCXO thermal time constant. In other words, the crystal cooled below its 75C operating point to about the same temperature after 1 hour in flight that it would cool down to in 4-6 hours with no air circulation around the SDU on the ground. Consider the possibility that the OCXO never got colder than the normal, in-flight cabin air temperature, but it reached close to that same start-up temperature in only one hour instead of the 4-6 hours it normally took when the galley fans and air conditioning were not operating prior to 12:50 (as occurred in Log-on #6) and also for the other ground log-ons.
To summarize:
a. The many examples of ground log-ons we have analyzed do not have identical transient error curves. They vary considerably in frequency-axis peak amplitude or in time-axis alignment, or both.
b. Therefore, one would not generally expect an in-flight log-on to exactly match a ground-based log-on. If it appears to match, at one particular frequency, it might be coincidental; it is also ill-defined because of the sparseness of the data.
c. One can fit very well all the ground log-ons by simultaneously fitting a classical servo transient error curve with two parameters per event, allowing a relative amplitude and a time offset to vary for each event. Thus one “shape” can match all the data, but not just one frequency-axis amplitude parameter nor just one time-axis parameter.
d. If one assumes that there is no ongoing maneuver and therefore the baseline is flat from 18:25-18:28, one can fit the 18:25 log-on event with a very large peak error occurring between the 18:25 and the 18:27 BFO groups.
e. Alternatively, by applying a best-fit ground curve (with a constant shape determined by fitting all the ground events #1-#6) to the 18:25 log-on, one can produce the close time alignment that Mike has proposed between the 18:25 and the 12:50 log-ons, but only if there is an ongoing underlying maneuver that temporarily increases the BFO by about 24 Hz at 18:27. I am favoring this scenario because it can also explain the BTOs (i.e., a SLOP).
You said: “Log-on plotting. Your plotting of log-on 3 differs substantially from Holland’s, which (his figs 8, 9) indicate LOA is about 24 Hz above ‘steady state’ initially, 38 Hz at 8+ mins later. You have this at 62 though doubtless I miss a simple explanation? You can’t get to steady-state in 30 seconds. Itr takes several minutes. That event has a BFO some 1500 seconds later, which is what I use for the steady state value.”
Holland does not show any data for #3 between about 30 seconds and 1500 seconds after LOR. I don’t know where you get 8 minutes. My current fit to that data set has a 70 Hz peak error, about 50 Hz of which is visible from the LOR until 30 seconds after LOR. I am not assuming the LOR always occurs at the same time as the peak of the transient error. I don’t know any reason why this would generally be the case. For smaller initial temperature errors, it appears the boot time plus other BITE tests don’t allow transmission until after the peak transient error has occurred and the transient has begun to decay. For larger start-up temp errors, the LOR may occur close to the peak of the servo overshoot.
You said: “Holland’s fig 9. He has upward divergence from steady state at the 160 secs mark of log-on 7 which your fig 3 does not show. Also there is like divergence in his log-on 6 at the 240 secs mark which does not appear on yours. It is on ALSM’s Beijing 1250 decay which he posted on 18th December which I take to be log-on 6. No-one has made comment on this apparent late divergence from steady state that I recall. Do we just overlook those as outliers?”
This difference is caused in part because Holland did not correct for channel bias differences. I did that in my #7. That compresses the “disconnects” by 4 Hz from about 8 Hz to about 4 Hz. There are still some residual small, and I don’t know their cause.
You said: “Newton’s Law of cooling. I think your equation at your item 10, page 6 needs, ‘Tset point – ’ to be added in front of its current right hand side.”
Yes, thank you for noticing. It should be Tocxo = Tsetpoint – (Tsetpoint – Tambient)*[1-exp(-t/tau) ]. I have revised the posted manuscript, including changing my many dyslexic OXCO’s to OCXO’s.
You said: “Overshoot Amplitude. Some conjecture. To me it is odd that the SDU design permits logs-on and subsequent transmissions well exceeding subsequent steady-state frequency fluctuations (your 3hz, page 1, item 2). Generally, to me there would need to be a good reason for designer toleration of overshoots at booting, unless it simply is unavoidable.”
I suspect the spec on the maximum time from a cold start to communicating was a driver that forced operating before the temperature control had stabilized with small error. The transient frequency errors do not prevent reliable communications.
You said: “Also I am unsure why it would be that overshoot amplitude would increase with OCXO heating time as per your paper.”
For this type of servo controller, the overshoot amplitude is a fixed percentage of the initial error when the loop is closed (i.e., the overshoot depends on the input step size). If the temperature error at loop closure varies, the peak frequency overshoot will be a linear function of the initial temperature error. Longer depowered time equals larger start-up temp error, equals larger frequency overshoot.
@DrB said: For this type of servo controller, the overshoot amplitude is a fixed percentage of the initial error when the loop is closed (i.e., the overshoot depends on the input step size).
I doubt it, as the initial temperature difference is so large that I suspect the heater is fully on for most of the warm up. It might be that thermal gradients between the sensor and the crystal or other thermal masses cause a difference in behavior for different initial conditions.
DrB and Victor: Victor is right. Thermal control loops in OCXO will always use a two step algorithm. When the delat T is large, the heaters are turned on 100%. The loop changes to a classic 2nd or 3rd order loop once the deltaT gets down to a few degrees.
re: Richard Feynmann and the Rogers Commission
We shouldn’t over-emphasize the role of Richard Feynmann in the Challenger inquiry. The technical reason was understood by a group of engineers in Morton Thiokol before the launch. Astronaut Sally Ride passed NASA information on the O-ring thermal problem to General Kutyna, who gave Feynmann a broad hint. Feynmann added the external expert role. The Rogers Commission was more an organised method of understanding how such a situation could have arisen and not been addressed, and presenting the information and a way forward to the public, finally, in an open way. That’s not the case here.
It is Feynman not Feynmann.
@Richard Cole
presenting the information and a way forward to the public, finally, in an open way. That’s not the case here.
Yes, forget my spelling comment. The important thing is credibility. Feynman was not going purjure himself for anyone or for a government agency. He may not have “solved” the Challenger issue, but his contribution was to “telling the truth”. That is huge, IMO, and lacking relative to MH370. I was actually most comfortable with Angus for the same reason.
@DennisW: There has been no lack of brainpower working to solve the disappearance. However, not all the facts have been available to all the interested parties. I do think the ATSB has tried to make publicly available all the information it has, but ICAO protocols have often impeded that. Here are some example that we know about. Perhaps there is other information that has been suppressed or omitted:
1) Digitized military radar data has not been released to the public. The ATSB has sought permission, but those requests have been rejected.
2) Data related to the registration of the cell phone connect was withheld from the ATSB, and only became available to the ATSB through unofficial channels. Despite this, the ATSB chose to include this information in its Final Report.
3) Data related to recovered simulator data found on the captain’s computer was provided to the ATSB by federal investigators. The ATSB chose to include this information, as well as how this information influenced the search, in its Final Report.
4) The flaperon recovered on Reunion Island has not been released to the ATSB for independent evaluation.
It’s possible that if all the known information were to be released to all the interested party, we would be no closer to finding the wreckage of the aircraft. However, I think that if this investigation was not subject to international protocols, and only driven to leave no stones unturned and to doggedly follow the truth, we would be further along.
@VictorI
It’s possible that if all the known information were to be released to all the interested party, we would be no closer to finding the wreckage of the aircraft. However, I think that if this investigation was not subject to international protocols, and only driven to leave no stones unturned and to doggedly follow the truth, we would be further along.
The Malays have been a bottleneck from the get-go. That alone should tell you what the diversion was all about. Would unfettered access to all the data help? I doubt it.
I think the opposite. With unfettered access to the withheld data, the aircraft would have been found by now.
Imagine if the following was made public:
Indonesian Primary Radar showing the point of the final major turn,
Primary Radar shown at the Lido Hotel,
The 10 sec Primary radar estimates,
FO’s phone connection with Banda Aceh,
The actual playback of the Malaysian SSR recording at Igari,
The fuel burn and endurance report,
The Vietnamese ATC transcript with MH88, and
The service history of the crew oxygen bottles.
MH370 memorial for Perth
https://thewest.com.au/news/perth/national-monument-for-victims-of-mh370-to-be-built-at-perths-elizabeth-quay-ng-b88699361z
Seabed Constructor is now passing CapeTown at 14.8 kts, on schedule to be in Durban Dec 27th.
@ALSM: Thanks for all the updates, Mike. How this plays out will be interesting.
@Niels
Back on the ‘blue panel’ image.
We have some hard data:
The camera was a handheld Nikon D2 xs.
The lens used was probably a Nikkor 70/?300mm zoom lens (other imagages in the same sequence were taken with 70mm).
The focal lenght used was 125mm.
The sensor was a 12 mega-pixel 23.5×15.6mm, crop-factor 1.5 (compared to fullframe sensor).
The altitude of the plane was 600ft.
The angle at which the image was shot through the plane’s window is unknown but I estimate about ~40 degrees.
Therefore the distance to the object was the long side of a triangel between the 600ft altitude of the plane and the position of the object shot at that angle of ~40 degrees.
I estimated 700 to 800ft but I think it could be even more; up to 900ft.
I could explain my approuch in further detail but I see this subject of the ‘blue panel’ and its associated debris field is not picked up on by others on this blog (except you).
And I’m not an expert either in this field.
I believe the subject deserves more attention. But maybe it does in the background by some real experts. I hope so and I hope to hear their results.
@All,
I have a new model result for the SDU warm-up transient events that utilizes the two-stage temperature controller concept.
You can get it HERE .
This result seems to explain all the BFO Log-on data, and it does not require the SDU to be cooled below normal cabin ambient temperature. There is only one transient error curve, and it applies to all seven of the events for which we have data. The only difference among events is the timing of the LOR (and subsequent transmissions) relative to the temperature controller settling time, and this, in turn, depends solely on the starting temperature at power-up. The longer the depowered time is, the lower the starting temperature will be (but still not below normal cabin ambient temperature), the longer it takes to warm up, and the closer the first LOR transmission will be to the peak of the temperature/frequency overshoot. Shorter depowered times have higher starting temperatures and shorter heating times, and the LOR transmission will occur well after the overshoot has begun to decay, producing smaller observed transient BFO errors.
Applying the (single) transient curve, which is fitted to Log-Ons #1-6 only, to the 18:25 Log-on #7, shows that the 18:27 BFOs are ≈20 Hz higher than predicted by the transient error alone. This is consistent with a northerly course then, and therefore a right SLOP centered at ≈18:27 is indicated.
Your comments, questions, and suggestions are welcomed.
@DrB: I think your current model with the two-stage heat up is better. Thank you for accepting and including our previous comments. However, in my opinion, a manoeuver is only one possible explanation for the higher BFOs at 18:27. The initial and thermal boundary conditions will be different for the in-flight boot-up of the SDU at 18:25, so you may be comparing apples and oranges. Also, the anomalous BFO values you rejected for the 12:50 log-on (log-on 6) is concerning. Nonetheless, the fits you obtained for log-ons 1-5 are impressive.
@DrB:
I am not convinced the #7 time scale is correct. As you noted in an email to me on Friday, you can get an excellent fit of all the 1825 BFO data, assuming no maneuvers, if the time scale is adjusted slightly from what is presented in the link above. I suggest you share with others both scenarios, not just the one that suggests a maneuver.
Nice work-in-progress, Dr B.
Happy Christmas to all.
@DrB
Excellent work! I admire your persistence. Do you have your data in tabular form, including the assumed Ambient Temperature of the SDU at power up?
@Richard
Temperature? My impression (warning subject to error) is that the responses were a scale to curve fit exercise.
@DennisW
Merry Christmas!
I just thought that the lower the Ambient Temperature at SDU Power Up, the longer the OCXO takes to reach a stable working temperature.
For example, if the SDU Power Up for MH371 was in a cold cabin in Beijing where the outside temperature was -2°C, it might take longer to reach a working temperature than a SDU Power Up for MH370 in Kuala Lumpur, where the outside temperature was +24°C, which might be similar to the time for an inflight SDU Power Up for MH370 at 18:25 UTC in a cabin continually heated to about 21°C.
Just a thought! (warning subject to Christmas excesses)
@Richard
Merry Christmas to you as well. I do agree with your statements above, but I do not think that is how DrB generated his curve fits. Even the initial rise corresponds only to a discarded data point. It could well be that the response was single-ended from near the top of his plot. Who knows?
@ALSM,
Yes, as I indicated privately to you, there is one fit which works with no maneuvers at all from 18:25-18:28, but this requires an explanation of why the 18:25 log-on would occur 25 seconds earlier in the transient error curve than all other log-ons (i.e., the 18:25:27 LOR occurs 25 seconds prior to the peak overshoot error. So far I don’t have this rationale, with the possible exception that the SDU was cooled well below 20C then, so I really don’t know if it is possible. I am fairly confident the peak itself is real because you can see the curvature change in the first several BFOs of a couple of the log-ons.
@All,
The two-stage thermal control design was discussed a year ago, but I did not try to numerically model it then. I initially fit both peak error and time delay simultaneously to each event. These fits are less robust because you have two parameters to fit per event, not just one. I like this latest method much better for a number of reasons.
Does anyone know of any reason why the channel-to-channel BFO bias differences could possibly be different for the 12:50 POR data whereas at least most of the other examples are the IOR satellite?
If one accepts both the BTOs and the last few military radar positions as being accurate, which I do, then we know a maneuver occurred between 18:22 and 18:28, even with no BFO data at all. In this case, finding a transient fit indicating no maneuvers circa 18:25 may just be an academic exercise.
The only single or pair of maneuvers I can find which matches radar/BTO/BFO is a right SLOP. I am now looking at how soon a SLOP would need to begin circa 18:25, and if the 18:25:34 and the 18:27 BFOs could both be affected by the SLOP, with perhaps a smaller effect at 18:25 compared to 18:27. This situation may come about because we now think the turn rate then was only 0.6 deg/sec with a 15-degree bank angle limit at high speed. That low turn rate means the SLOP takes longer to execute, and it could affect more than just the 18:27 BFOs. I’ll post those results later on.
I’ll also post a table of all the retrieved log-on BFOs from Holland’s paper.
In my temperature servo model , I assume an ambient of 25 C and an operating point of 75C. I don’t try to guess the starting temps for each event. The exact solution is quite complicated, including the fact that temperature gradient through the crystal continues to change with time even if one crystal surface temperature is held constant (per ALSM). The generic equation I use with two shape parameters is a simple way to get a reasonably accurate shape to match the time -shifted transients, in line with DennisW’s comment. This approximate method seems to be borne out by the precise overall fit of about 1-2 Hz RMS.
@DrB
Could you also tell something about the implications your findings could have on the 00:19 log-on?
Especially on when the IFE log-on could be expected?
@Ge Rijn asked “when the [00:19] IFE log-on could be expected?”
I’ll be describing that in a day or two, hold tight.
Bobby: Re “…the 18:25:27 LOR occurs 25 seconds prior to the peak overshoot error…”
As we have discussed, this might happen if the SDU POST logic works as I have described several times here. Depending on the order in which the several POST logical input conditions turn true, the LOR could occur at different times relative to the BFO transient. If the OCXO flag goes true before other tests complete, the LOR could occur later during the transient than for the case where the OCXO test is the last input to go true. The OCXO test might ordinarily turn true early when the ambient temp is +25C, but last if the ambient was, say -30C. We have no way of knowing what the ambient temp was at 18:22, but given everything else we do know, a cold cabin is certainly one possibility. I have been recommending an environmental chamber test for the last 11 months, but so far, no results. If David Linsdall is following this blog, please consider running this test David. It could shed some light on what happened circa 1825.
@ALSM, @DrB: It is admirable to attempt to extract all the details from our sparse information, but we have to be careful to not over-analyze the data in light of our unknowns. We know neither the initial temperature condition of the distributed thermal masses, nor the thermal boundary conditions for the power-up transient at 18:22. In my opinion, it is possible but it would also be surprising, if the transient thermal behavior was exactly as it was for the other power-ups. To attribute deviations in the transient behavior to a specific manoeuver that occurred in that short time interval is one of many possibilities.
Victor: I agree. That was essentially my point. That said, if Thales runs a test, and determins that the 18:25 LOR is consistent with a very cold cabin, that would be helpful.
@Don Thompson
I hold tight..
As I see it with non-specialist eyes and understanding the crystal will only resonate acurate within very specific frequency bounderies at a specific temparature.
The colder the crystal is the longer it will take to reach that specific state. But as soon the OXCO (and the crystal) reaches this temparature all should be fine with only a frequency overshoot as AlSM and DrB state (if I understand them well).
I would be curious if such an ‘overshoot’ would also be present in the 00:19 log-on sequence. And if it could be figured out the IFE log-on should be expected earlier than 89 seconds at that time.
Considering the SDU was powered after 18:25 and there would have been no OXCO warm-up at the 00:19 log-on.
Ge Rijn: As noted many time here, the OCXO temperature would not decrease much at all in the one minute the power was off between 00:17:30 and 00:18:30. Therefore, there would be very little frequency error at the 00:19 LOR.
@Ge Rijn,
Just to ensure a basic understanding of the issues that currently have focus in DrB’s work, and David’s comments.
OCXO warmup and settling time: particular to BFO interpretation at 18:25. As ALSM notes, while the AES is without power for approx 60s, at 00:19, as the APU starts and its generator comes online is not significant.
Datalink events after GES Log On: particular to whether the elapsed time, following LOA-LOA at 12:50 aircraft power up, or restoration of the L AC Main Bus at 18:25, would be representative for the scenario at 00:19.
@Ge Rijn,
Just to ensure a basic understanding of the issues that currently have focus in DrB’s work, and David’s comments.
OCXO warmup and settling time: particular to BFO interpretation at 18:25. As ALSM notes, while the AES is without power for approx 60s, at 00:19, as the APU starts and its generator comes online is not significant.
Datalink events following GES Log On: particular to whether the elapsed time, following LOA-LOA at 12:50 aircraft power up, or restoration of the L AC Main Bus at 18:25, would be representative for the scenario at 00:19.
@All,
I have put a number of OCXO items into a single EXCEL spreadsheet. First, I list the BFOs and times for the seven warm-up events. Events #1-5 are the estimates I made (and recently did again) from Holland’s Figures 8 & 9. I think the accuracy of my transcription is generally about +/- 1 Hz in BFO and +/- 1% of the time reading in seconds.
The spreadsheet also lists the transient curve parameters, and the best-fit time and frequency offsets for the seven events. The transient curve is fitted to Events #1-6, and then I use that curve to correct Event #7 at 18:25.
In order to determine the best timing of #7 relative to the peak overshoot, one has to simultaneously fit the SLOP maneuver to the CBTOs and the CBFOs. That is, this combination fit finds both the SLOP parameters and the timing offset of the #7 transient curve. There is a very good solution, and the parameters are listed on the sheet, along with many figures displaying the sequence of all relevant flight parameters with time from 18:20 through 18:30. I constrained the SLOP offset distance to be 15 NM per MAS procedures.
The best-fit SLOP begins circa 18:24:01 (shortly after the left AC bus power is restored and while the SDU is rebooting) and ends circa 18:27:42.
You can get this spreadsheet HERE .
As always, comments, questions, and criticisms are welcomed.
@Andrew & Mick Gilbert,
I tried fitting the several versions of the Holding patterns, which we discussed previously, to the 18:25-18:28 data without any credible success. I don’t think a Hold occurred near NILAM for that and for other reasons.
@All,
Looking at the Inmarsat data logs for the 12:50 warm-up Event #6, the Bit Error Rate for the data points I excluded from my fitting process (as outliers) display high BERs – from 14 to 30 (units = ?). In addition, during that entire warm-up event the received power was low, and the carrier to noise ratio was also low.
The BERs for the 18:25 event are all zero, except for the 18:25:27 BFO, for which the BER is 5. As we know, that BFO was in error due to overlapping signals.
I think it is likely that at least some of larger scatter in the #6 BFOs circa 12:54 is due to degraded accuracy as indicated by the high BERs then.
@Dr B. Thank you for your consideration of my comments.
A few of points on your response.
• You and others have requested ATSB/DSTG to arrange cold testing without result. Were there confidence in Taus, calculation could quantify the interchangeability of the effect on OCXO temperature of depowered (ordinate) and time to heat (abscissa), with a series of exponentials rising from the origin as a fan, based on different ambients, to the right being the lowest ambient. However Tau complications due to convective cooling prior to ground reboot vs forced ventilation in the air, plus varying ambient over several hours of cooling would complicate that.
Perhaps OI with their new contacts would now see it in their interests to promote and support that?
• When you say, “Events #2-5 have about 1-hour depowered times” I think log-on 2 at, “between 295 and 354 minutes” (your Table 2) is in the 4 – 6 group. That may trouble your class A vs B peak transient error comparisons.
• My 8 minutes to reach steady state in Holland’s 3rd log-on’s, at 38 Hz beneath LOA, was in error, my mistaking 1500 secs for 500. The comparison with your 62 remains though. I see in your new graph it is 40 so the difference now is unremarkable.
Your new graph is a great development generally.
• Your response about the effect of servo controller on overshoot amplitude has been overtaken by events as I understand it. I see note 3 of the Honeywell manual on the 4200 SDU model has a constant 10W for heating, for up to 10 mins which looks pertinent as to the open loop first stage.
• You make no mention of a 16 min down time for the SDU as a possibility prior to APU autostart at 00:17:30. Do you think that is remote?
Your presentation of your work remains admirable thank you.
About your new plot, the coincidence of the 7th arc LOA hitting peak overshoot amplitude is striking, LOR being assumed as below LOA. Probably of no consequence, I notice that drawing a curve between the top of the LOA error bars, the bottom of the IFE 1st completion and the top of the 2nd gives a higher gives a higher transient.
You have found the 6th log-on BERs I see. You may well be aware that in the immediate next BFO, 25 secs later at 13:19:55.747, the BFO had risen again to 358. That was followed by 2 more within 4 secs, at 360. The BERs there were 0 and 1. Channels were 13 and 24.
I will have more to say on the other sections of what I raised, those about the possibility of the 7th arc being lower, when more settled on a couple of points.
Dr B. Second last para 25 mins vice seconds
@DrB
Accepting the radar data from Pulau Perak at 18:02:59 UTC and the last radar point at 18:22:12 UTC, then the average speed of MH370 in the interval was around 512 knots. Interpreting the positions in the 18:25 / 18:28 timeframe from your chart of latitude vs longitude, the speed that best fits the BTO and BFO data is around 420 knots. This implies MH370 was slowing down, which in turn might also indicate it was descending.
The Malaysian Factual Information report states that MH370 disappeared abruptly from radar. I understand “abrupt” to mean, that once the aircraft was off the radar screen, there were no further intermittent captures. This could also imply the aircraft was descending. “The tracking by the Military continued as the radar return was observed to be heading towards waypoint MEKAR, a waypoint on Airways N571 when it disappeared abruptly at 1822:12 UTC [0222:12 MYT],10 nautical miles (Nm) after waypoint MEKAR.”
I agree with you that irrespective of changes in speed and altitude, a lateral offset to the north fits the BTO and BFO data that you have carefully corrected. I just want to point out that the scenario at 18:25 / 18:28 could possibly be more complex with a reduction in speed and altitude as well as a lateral offset all happening at the same time. Change in speed will affect the position, which in turn affects the BTO fit. Change in altitude also affects the BTO fit. Speed, Rate of Descent and Track will affect the BFO fit. It is quite difficult to model all the possibilities, if you remove the assumptions that MH370 remained at 34,000 feet and around 500 knots.
However, your excellent work provides a well argued baseline scenario for the 18:22 / 18:28 timeframe. I note that your chart shows MH370 continuing the lateral offset parallel to N571 after the first call at 18:40 UTC and that the FMT occurred later.
Per @Don Thompson, Ocean Infinity’s Seabed Constructor is now docked in Durban, South Africa. Whether OI will reach agreement with Malaysia, and whether OI would re-start the search without an agreement, are unknowns.
If Malaysia do not sign a contract with Ocean Infinity, then OI is not bound to search the ATSB First Principles Review 25,000 km2. If OI decide to go ahead without a contract, then they can search where they think best.
As I understand the legal implications of going ahead without a contract, OI can search where they want in international waters, but they cannot recover any wreckage without an agreement.
For OI to be the company to find MH370 is worth a lot in marketing terms, as proof of their technology.
But what we want is the black boxes and other key pieces of wreckage to be recovered and analysed, so that the mystery can be resolved.
@Richard
As I understand the legal implications of going ahead without a contract, OI can search where they want in international waters, but they cannot recover any wreckage without an agreement.
Actually I do not believe there is any “legal” impediment to OI recovering wreckage or even selling recovered wreckage to a third party.
@Dennis I think when you look up maritime salvage law you’ll find it’s not quite as simple as that.
@DennisW
I was not just thinking of the fact that MH370 comes under the Maritime Salvage Convention 1989, but also that MH370 is subject of an ongoing aircraft accident investigation under the international standards and practices of the ICAO Annex 13, as well as a site of a police investigation by authorities from several countries and subject to numerous court cases in various jurisdictions.
@Paul
@Dennis I think when you look up maritime salvage law you’ll find it’s not quite as simple as that.
Actually, I spent quite a lot of time looking at Maritime Salvage Law. I am not in the habit of posting personal opinions (except in the designation of whackos). The simple truth is that OI does not need a contract to salvage and take possession of the aircraft debris. They could claim both recovery cost and opportunity cost if someone wanted to acquire it. The risk is that it is (my opinon here) not worth very much. FDR will likely show aircraft was flown to its recovered location and the VDR will be blank.
The law of salvage is a concept in maritime law which states that a person who recovers another person’s ship or cargo after peril or loss at sea is entitled to a reward commensurate with the value of the property so saved. The concept has its origins in antiquity, with the basis that a person would be putting himself and his own vessel at risk to recover another and thus should be appropriately rewarded.
OI’s good faith attempt to negotiate an award ahead of time is merely a sensible attempt to avoid a lengthy procedural process after the fact.
@Richard
Yes, it is complicated, but salvage law has a very long history.
My guess reamins that the snag in the negotiations with Malaysia revolves around two issues:
1> Cost, of course. It is hard to put a value on the black boxes. A sensible conclusion is that the value is relatively low.
2> Escrow of reward. The Malays are simply not trustworthy, and the appointment of an escrow agent and the escrowing of funds would be mandatory.
@Richard
My mind flashed to a recurring dialog I have with Chilean friends who accuse the US of not paying Chile enough for agricultural products (including wine). Since I am regarded as an “ugly american”, I do not dissappoint them when I tell them they are lucky we pay them anything at all. We could simply come down there and take it.
Another facet of the above is chess. Every time I go down there to visit (mostly hike/climb in Torres del Paine) they drag another chess player out of the woodwork in response to my assertion that there is probably not a Chilean citizen who can defeat me. So far I am up on that one as well. They love me.
@Richard, you said: December 27, 2017 at 10:23 am
I was not just thinking of:
(1) the fact that MH370 comes under the Maritime Salvage Convention 1989,
but also that:
(2) MH370 is subject of an ongoing aircraft accident investigation under the international standards and practices of the ICAO Annex 13,
as well as:
(3) a site of a police investigation by authorities from several countries
and:
(4) subject to numerous court cases in various jurisdictions.
Four points.
(1) You claim “fact” that MH370 “does” come under
the convention.
I suspect that there would be an army of lawyers who would disagree with you.
(2) Do you still consider MH370 to be “an accident”, within in the genuine meaning of the word ?
You must be the only one.
Even Najib said “deliberate act”.
Deliberate = criminal = legal investigation(s) = ICAO Annex 13 is “out the window”. See (3) & (4).
(3) True.
(4) True.
It would be interesting to determine which country / legal jurisdiction Seabed Constructor is operating under.
If Seabed Constructor did “discover” the wreckage of MH370, it would be interesting to know which country / legal jurisdiction they would be legally bound to report the fact to.
@Ventus
It would be interesting to have a lawyer with extensive salvage experience weigh in here. I bounced it off my SO who is a licensed attorney (Cali bar) and as usual (she is also an excellent mathematician – PhD EE), but she demurred. She regards MH370 as a complete waste of time both from a legal and mathematical standpoint. Like other colleagues, she looked at the ISAT data for 30 minutes or so and laughed.
@ventus45;
“If Seabed Constructor did “discover” the wreckage of MH370, it would be interesting to know which country / legal jurisdiction they would be legally bound to report the fact to.”
The “Seabed Constructor” is registered in Bergen, Norway.
Therefore the legal jurisdiction is the Norwegian state, which would need to act in accordance with ICAO Annex 13. The interests of Malaysia would be best served if it (Malaysia) appointed Norway to act on its behalf and a commercial agreement was made with Norway, with OI contracted through that conduit.
The charter agreement between OI and Swire Seabed (the vessel’s owners) is probably governed by the laws of Norway, but could also be delegated for dispute settlement in the Admiralty Court of the UK. Something often found in maritime charter-party agreements.
ah, Ventus, your shows you’ll be all lols when/if OI finds MH370. No doubt you’re spending your holidays plotting the layout of phony debris in Seabed Constructor in preparation for that day.
Personally I find it better to be eating fruitcake than to be one at this time of the year.
whoops — used an illegal character in last post. Meant to say “your scarequoted “discover” gives away your plan..
@lkr
Don’t be rediculous. I was not suggesting any such thing.
What I had in mind, is, a hypothetical.
What if, OI is working on some other theory, unknown to any of us ?
For arguments sake – say they have secretly contracted Metron, and they have receomended “some other area(s)” other than the CSIRO Area.
Malaysia has been quite stupid, or very astute, depending on how you look at it, in specifying:-
(a) a rediculously small area to search, and
(b) a rediculously short time constraint for the search.
By doing so, they have boxed themselves into a very small “credibility space”.
If Malaysia genuinely beleive it is in such a small area, and that it could be found there in such a short time, why have they not gone all out to find it themmselves ?
They haven’t, so obviously, they don’t.
We all know the “official tripartite line” = precise location etc.
None of the governments want to be up for any further expenditure for searches, even though they say (for public and media consumption) “money is no object”.
The truth is, they failed with their own search, and it will be highly embarresing if someone else finds it, when they couldn’t.
I don’t think any of the governments ever considered, let alone expected, that an independent orgainisation would front up with a “no cure no fee” offer for MH370. It clearly took them by surprise, and they are having difficulty handling it. They have been “shadow boxing” with OI, probably hopeing that they will simply give up and go away.
But OI are not going away.
If I were OI, I would plan on searching “the CSIRO Area” first, just for show, publicly, and then, after all the media press conferences in Perth, “we tried, but we failed”, simply reprovision, and quietly sail off into the sunset, obstensively back towards Africa, but instead, go and search where “my advisors” really think it is.
What if they found it then, well away from “the oficially ordained search area(s)”.
That would open many cans of worms, plenty of bait for a swarm of lawyers.
All,
Yesterday I met with some relatives whom I had seen at most a few times over the years. As it happens, one was a pilot with a major airline for many years who had flown all over the world in aircraft of all sorts, including the 747 (but not the 777). I had primed him that I was going to ask questions r.e. MH370. The following are some notes on what I gleaned from him (incomplete on account of the fact that lots of family talk was interspersed, and all of this took place over lunch). Any errors are mine.
1. R.E. the Operational Flight Plan. Remember that, in Folder 5, page 1 of the OFP has two handwritten numbers for fuel, plus a signature that we now know is Zaharie’s. Who filled in the numbers? According to my relative, the first stop for captain on a flight is the dispatcher. The dispatcher hands the captain an envelope with all the information for the upcoming flight (and what he described is a lot like what is in the MH371 OFP). It sounds like the fuel numbers are written in by the dispatcher, and if the captain wants them changed (e.g., if more fuel needed for a worst-case engine out diversion), they are crossed out and new ones written in; cargo can be off-loaded if needed to balance out the extra fuel. But it depends on the airline.
2. When I described how “Mode S” dropped off ATC radar, his immediate response was “Altitude reporting off”. I hinted that the transponder switch had been rotated left but not to the standby position, and he didn’t disagree.
3. Manual flying – he indicated that the aircraft could fly level if it was well-trimmed, but it would be blown about by the winds. Further, if the pitch trim was off, it would go through fugoid motions.
4. Transfer of control – I was curious about the precise time when a pilot would contact the next ATC after passing a Transfer of Control point. (He seemed unfamiliar with the concept of “transfer of control” – to him, it was a reporting waypoint.) He said that it depended where you were – in US airspace, he would contact the next ATC center as he overflew the reporting waypoint. In Vietnam, he said that it might be possible to overfly the entire airspace without reporting in and nothing would happen. When I said it took 4 hours for MH370 to be declared missing, he was not surprised.
5. How would a pilot know the precise momment of “passing waypoint IGARI”? He said that, if watching the legs page, a particular waypoint turns magenta as you fly past it.
6. He said that if the autopilot is engaged with nothing initialized, it would fly constant heading. When queried, he agreed that it would be magnetic or true depending on the setting of the norm/true switch. He said that it was normal practice to switch to true heading in polar regions (where he has flown a lot.)
7. When I asked why Boeing would say the Mach needed to be greater than 0.73, his first response was that it was to avoid stalls [but he was thinking of normal cruise altitude], then said it was to avoid excessive fuel burn. He also mentioned turbulence penetration [which I think means you need to allow that speeds can fluctuate up and down so you want to say away from both min/max limits to provide margin.]
8. He was not aware of any Antarctic runways being in the navigation database and was skeptical that they would be.
Finally, here are three things I learned that are not MH370-specific.
1. On a 747 there is a buzzer that goes off if no action (e.g., pushing a button) has been taken for 3 minutes.
2. Pilots were instructed not to fly over the North Pole. But if you did (hee, hee), no one was there to watch what you were doing.
3. Back when flights started flying over polar regions, there were concerns about fuel freezing. So, just before takeoff, a sample was drawn from the fuel tank, and the freezing point was measured; this info was uploaded to the pilots after takeoff. Fuel temp was monitored in-flight, and if it came close to the freezing point, speed was increased (.01 Mach, e.g.) to increase skin friction and heat up the fuel.
@Ventus. The point you make need not entail any misleading. Suppose OI does its best in a contracted search in good faith. If unsuccessful and it wants to search on without reward, or others do, what is to stop them, international waters, site unknown?
A corollary is that it may be that they may be free to go ahead now, contract-less.
There might be a little cat and mouse afoot, they apparently applying pressure on Malaysia yet M with an unpaid reward to gain by not contracting. I doubt OI would be in a strong bargaining position if then it found the wreckage.
Not entirely hypothetical given the current apparent standoff.
@sk999
Interesting, and it follows my own experience.
I recently interacted with my daugeter’s (Professorr of CS) colleagues at a party in San Luis Obispo, home of Cal Poly, The physics profs were totally unaware of the recent advances in geomotric calculus, and could care less – despite the fact that they will revolutionize physics eductation going forward.
Likewise the people here. They have no interest in that learning curve.
@David
“I doubt OI would be in a strong bargaining position if then it found the wreckage.”
Malaysia clearly can not be trusted.
If I were OI, and planned to scan say 100,000 square kilometres, and say I found it in the 30,000th square kilometre, I would continue to scan the full 100,000 square kilometres, then, having finished all the 100,000 square kilometres, sail home, and then announce I had found it, with all the sonar and ROV evidence, but refuse to release the location details, until the money had been paid, in full.
@sk999,
Re additional point 1. crew activity monitor: yes, a function of the FMC. If no interaction is detected the 777 sets an EICAS warning, if that doesn’t solicit a response a klaxon sounds via the flight compartment speakers. However, it’s readily disabled by means of a cct breaker on the overhead panel.
@DennisW expects the CVR recording to be silent. That is quite possible, if that breaker was pulled.
@sk999
Expanding on Don’s comment re the crew alertness monitor, the FMC monitors pilot interaction with switches on the MCP, EFIS control panel, display select panel, CDUs and radio transmitter microphone switches. There are three levels of annunciation available if no pilot interaction occurs within a specified time interval:
Advisory – EICAS advisory message with no aural warning.
Caution – EICAS caution message with an aural beeper.
Warning – EICAS warning message with an aural siren.
The crew alertness parameters can be modified by the airline. Any or all of the annunciations can be enabled; the default setting is ‘disable’. The default times and the available ranges for the cruise phase are:
Advisory – 15 min (15-60 min)
Caution – 20 min (15-75 min)
Warning – 25 min (15-75 min)
Fuel temperature certainly can be an issue on the B744, especially when departing North America for high latitude routes with a load of Jet A fuel, which has a higher freezing point specification than the Jet A1 used in other parts of the world (-40°C vs -47°C). At the airline where I work, the actual freezing point of the fuel being loaded into the aircraft was determined and given to the crew before departure. In my experience, increasing speed doesn’t do much to increase the fuel temperature – it takes too long to have any measurable effect, despite Boeing’s advice. If it became an issue, we nearly always had to descend (or sometimes climb!) to find warmer air. I never encountered the same problem in the B777; the fuel always stayed much warmer than it did in the B744.
@Andrew,
Thank you for relaying the FMC CAM detail, I didn’t have it to hand earlier.
The MAS FCOM states “the EICAS alert message PILOT RESPONSE is displayed“. It specifically states ‘alert’ rather than Advisory, Caution or Warning, whereas other airline FCOMs do use the defined terms.
@Andrew
Re: Jet Fuel “North America for high latitude routes with a load of Jet A fuel, which has a higher freezing point specification than the Jet A1 used in other parts of the world (-40°C vs -47°C)”
That’s why I keep asking about the energy content of the KLIA jet fuel for the endurance calcs. Presumably the North American jet fuel is a bit higher energy.
The quality difference was a bit of a factor in the TWA Flt 800, whereas the USA jet fuels also trend towards higher flash points. But the center tank was filled with residual from Europe, I believe.
In the first week of MH370, when there were reports of flaming aircraft off-shore Vietnam, of course my initial MH370 theory was fuel tank fire.
@sk999 said He was not aware of any Antarctic runways being in the navigation database and was skeptical that they would be.
I don’t know whether or not the airports were in the navigation database. Malaysia says they were not, which agrees with what your relative said. However, there might be some confusion, depending on how the question was asked. What takes up a lot of memory in the navigation database are the arrival and departure procedures for an airport, e.g., the SIDS and STARS, and an airline is not likely to include this data for airports outside of their serviced geography. On the other hand, the memory size allocated to the coordinates of an airport waypoint (or a nearby navigational aid) would be much smaller, and could possibly be in the navigation database. I’m not sure that question has ever been properly posed to Malaysia. Malaysian officials tend to answer the question they want to answer, and not the question that was asked.
@VictorI
To jump in. I think if Malaysia states those runways were not in the MH370 database, they were not there at all. Malaysian officials can answer such a question with double meaning but it would be easily confirmed or refuted by any MAS B777 pilot and any official investigation team.
Therefore I rather take it for fact that those runways were not in the database at all.
It would make no sence either imo to fly towards one of those runways. Why deliberately flying there when they are not reachable anyway? Then SP makes more sence as a simple track/heading from a PIC who has no specific goal in mind other than to let the plane drop out of the sky at fuel exhaustion.
I just don’t buy this. Overwiewing the whole flight till 18:40 it has all the signs of a well planned, controlled undertaking.
Therefore I believe the end-of-flight and it’s impact area were as well planned too.
I believe the end-waypoints were set just before- or after FMT. At a specific location that suites criteria best for a most favourable disappearance of evidence.
Maybe the best evidence is; he succeeded so well to date.
@Ge Rijn: The theory is that the waypoint would serve as a distant, unreachable target. Pilots fly long distances in LNAV mode, not by dialing in a course. What actually occurred, I do not know.
At this point, it probably doesn’t matter. The first 25,000 sq km recommended by the ATSB will be searched by OI. After that, if required, the search will proceed along the 7th arc to the northeast. How far along the arc will be decided by OI. Using the scanning rates supplied by Richard Cole, I think 1,000 sq km per day is possible, and I’d be surprised if the search stops short of 29S.
@David,
You are correct in that the Log-on Event #2 has a similarly long depowered time as #1 and #6. I don’t have any theory as to why its frequency decay appears to be comparable to the short depowered times, other than possibly there is a significant transcription error. The data points for #2 are particularly hard to decipher in Holland’s figures.
You said: “You make no mention of a 16 min down time for the SDU as a possibility prior to APU auto-start at 00:17:30. Do you think that is remote?”
I tend to agree with the ATSB’s reconstruction of events at fuel exhaustion, in which the SDU only loses power when the left engine runs out of fuel and the APU auto-starts. I think it is much less likely that the SDU lost power when the right engine failed.
@Victor/@all
I am struggling to express my thoughts on the OI effort, and I am not doing very well. Try again.
People here and on JW refer to the OI effort as a “continuation” of the search. It is not a continuation at all. The original search can be regarded as a government funded public service. The service being for humanitarian reasons and for ICAO reasons. Failure scenario advocates would add public safety to the list of reasons. At no time was any regard given to cost/benefit considerations. Well, that effort was executed well (my opinion), but came up empty. The parties involved have more than satisfied humanitarian and ICAO obligations. Public safety might still be put in the mix by those so inclined.
The new effort proposed is quite different. It is a for profit salvage operation. In this case cost/benefit has a significant if not dominant role to play in the decision making. I am not aware of any contributions being proposed by China or Australia. Perhaps Australia will allow the use of a Perth base at no or low cost. Don’t know about that. Whether Malaysia goes ahead with a contract would probably influence Australia’s attitude. In any case OI is free to proceed on the basis of salvage law. Although proceeding without a contract would have enormous risk, IMO. If the black boxes are recovered it would be difficult for OI to hold them hostage for some negotiated payment.
@DennisW
It seems to me a new company in this business like OI with developing and testing new equipment and technologies it’s not about short-term profits but mainly about longer term investment.
Their trials in the Atlantic also must have cost millions. This is not wasted money ofcourse.
I think a ~50 million dollar investment in the search for MH370 won’t be wasted money either for OI anyway. Gaining tremendous experience and marketing in the process. And if they succeed in finding MH370 they’ll hit the jackpot. Maybe not with a Malaysian fee but in the world-wide industry for maybe decades to come.
@Don Thompson
@sk999
Re: ”The MAS FCOM states “the EICAS alert message PILOT RESPONSE is displayed“. It specifically states ‘alert’ rather than Advisory, Caution or Warning, whereas other airline FCOMs do use the defined terms.”
Our FCOM says the same, but all three levels of alert are enabled on our aircraft. Bored pilots crossing the North Pacific have been known to avoid touching any buttons to see for themselves how the system works!
My previous description wasn’t strictly correct. The PILOT RESPONSE EICAS alert remains displayed if there is no response after the initial advisory is triggered. The alert is then escalated by the addition of an aural caution beeper if there is no response after an additional time interval, followed by an aural warning siren if there is still no response.
@DennisW: I think OI’s motivation for searching is a mix of economic, humanitarian, and personal interest. I don’t know the relative value to assign to each, nor does it really matter at this point. Ultimately, the economic interests will determine the duration of the search.
I suspect that Australia’s contribution will be “in-kind”, with no cash contribution–the Final Report has already been written and published. I am confident that the ATSB does really want to help find the plane, and will assist Ocean Infinity in any way it can. As CSIRO’s David Griffin has said, already the ATSB has provided OI with advice about where to search.
As for international salvage laws, I don’t know how that would work if OI is not under contract with Malaysia. While under contract to search, if OI finds the wreckage, there would probably a separate agreement with Malaysia to recover parts. I haven’t heard that OI is less confident that a contract will be signed, but I won’t make a prediction, either.
@Andrew and Don,
And if all the alerts / warnings are ignored, or not responded to because of an unresponsive crew . . . . . .?
@Richard,
You said: “Interpreting the positions in the 18:25 / 18:28 timeframe from your chart of latitude vs longitude, the speed that best fits the BTO and BFO data is around 420 knots. This implies MH370 was slowing down, which in turn might also indicate it was descending.”
My plot of position versus time in my most recent posting is generated assuming a constant speed of M0.840 at FL340 with +10.9C SAT above ISA. This is 498 KTAS, and it is consistent with the military radar track speed. I don’t know how you get 420 KTAS. I am not aware of any evidence of 9M-MRO slowing down or descending prior to 18:29. Using only a slow-down or only a descent cannot fit all the 18:22-18:28 data. In particular, a descent then forces the BFOs far below the 18:25-18:28 BFO measurements. I have also tried to fit a slow-down with the SLOP, but this produces no benefit over a SLOP alone.
You said: “The Malaysian Factual Information report states that MH370 disappeared abruptly from radar. I understand “abrupt” to mean, that once the aircraft was off the radar screen, there were no further intermittent captures. This could also imply the aircraft was descending.”
I think a better explanation is that the radar set simply clipped off all returns at time delays beyond 250 NM (its specified maximum range). That explains the lack of further contacts. You don’t need a descent to explain this.
You also said: “I note that your chart shows MH370 continuing the lateral offset parallel to N571 after the first call at 18:40 UTC and that the FMT occurred later.”
I was not intending to imply that the FMT occurred later than 18:40. In this case, I was only looking at the 18:22-18:29 data. My opinion is that it is more likely the FMT occurred before 18:40, and next I plan to fit that turn along with the 19:41-00:11 data.
@Brian Anderson
Re: ”And if all the alerts / warnings are ignored, or not responded to because of an unresponsive crew . . . . . .?”
I’m not sure, that case isn’t mentioned in the manuals. I suspect nothing happens, except that the alert might be recorded by the DFDR.
@Brian, & @Andrew
My understanding of the FMC CAM function is that it’s an alarm, the FMC function does not change as a consequence if the alerts get no response.
Two cct breakers on the overhead panel will isolate the aural warning speakers, another two disable the stick shakers. If the CVR is ever located & its data recovered, I’ll be interested to know if the aural warning and/or stick shaker is evident.
@Don Thompson
That’s my understanding too. There’s no ‘fault’ so I doubt the alert would be detected by the CMCS and consequently the CMCS won’t generate an automatic report to the ground. The aural alert would be recorded by the CVR, assuming it wasn’t subsequently overwritten (or the warning speaker circuit breakers pulled). The EICAS advisory might be recorded by the DFDR, but I’m not sure.
@DrB
My apologies if I read too much into your excellent work.
I look forward to your next part of the analysis.
@DrB
I was not intending to imply that the FMT occurred later than 18:40. In this case, I was only looking at the 18:22-18:29 data. My opinion is that it is more likely the FMT occurred before 18:40, and next I plan to fit that turn along with the 19:41-00:11 data.
Like Richard, I look forward to the next chapter. My feeling is that an FMT at or prior to 18:40 will produce terminal locations too far South, but perhaps not.
@Don and Andrew,
That’s where I was heading too . . .
The CVR might reveal something. While not a “fault” I would have thought that ground might be pretty interested in the situation where all alerts/warnings were not acted upon.
Do you have CONFIDENCE in the MH370 search investigation?
=====================================================
JIT, Final Report, Refinement of analysis, Mar 2014:
“The LAGOG waypoint was included as it was close to a supplied, HIGH CONFIDENCE position labelled ’19:12Z’.
——
ATSB, Final Report, Initial Search Area, June 14, 2014:
“1912 NW point … reassessed as INVALID …”
—————————————————–
Tony Abbott, Australian Prime Minister, April 11, 2014, r.e. Ocean Shield “pings”:
“We are CONFIDENT that we know the position of the black box flight recorder to within some kilometres”
—————————————————–
Angus Houston, May 3, 2014:
“… I’m CONFIDENT that with an effective search, we will eventually find the aircraft …”
—————————————————–
Warren Truss, Transport Minister, May 5, 2014:
“We are still VERY CONFIDENT that the resting place of the aircraft is in the southern (Indian) Ocean, and along the seventh ping line …”
—————————————————–
Martin Dolan, ATSB, Jul 31, 2015 r.e. flaperon:
“… authorities were ‘INCREASINGLY CONFIDENT’ the debris came from MH370.
—————————————————–
Darren Chester, Assistant Minister for Defense, Dec 3, 2015, as the underwater search approached the peak of the DSTG heat map:
“We have a HIGH LEVEL OF CONFIDENCE that we ARE SEARCHING in the right AREA …”
———
Greg Hood, ATSB, Jan 17, 2017:
“… authorities were CONFIDENT it was NOT in the AREA that has been SEARCHED.”
—————————————————–
David Griffin, CSIRO, Apr 21, 2017, after testing genuine flaperon at sea:
“… this new work leaves us MORE CONFIDENT in our findings.”
@sk999: Yes, the turn of events has shown that Australian authorities have made statements that have been overly optimistic about the probability of success. Part of that is because they had to justify the proposed search budgets to politicians and the general public.
@All: So along those lines, let me pose a question to the group. What do you think is the probability of finding the debris field in the 25,000 sq km area suggested by the ATSB?
@Victor. 1 in 2.
Esky-nines: Are you not “confident” that the Reunion flaperon came from MH370? If you aren’t, there’s probably a better campfire elsewhere on the web.
Per most of the rest of your “confident” quotes — you can ignore the politicians [no different in Oz, Malaysia or US] — who will always hear what they want to from the techies and spin it out of recognition..
And eat your Xmas fruitcake — it’s getting stale.
@VictorI
I’ll give it a 1 in 10.
The ‘1’ only representing the most northern part of the ATSB 25.000km2 area between ~95.3/96.2E and ~32.3/33S.
The leaving ‘9’ I put outside of this area between ~95.3/97.5E and ~32.3/33S.
@Victor
I am not a fan of the new CSIRO area as you know. I lean more in Richard’s direction of 30S being the peak of a normal distribution. If I assign a liberal sigma value of 2 degrees, then the area below 33S has a cumulative probability of around 0.07. Give it 10% to round things off. Richard would probably assign an an even smaller sigma value.
@DrB, Thank you for sharing your detailed work 12/26 (OCXO Transient Errors and 1825 SLOP Fit). Housekeeping comment only, delta description in O:3…
@Victor “So along those lines, let me pose a question to the group. What do you think is the probability of finding the debris field in the 25,000 sq km area suggested by the ATSB?”
You might not consider me one of the group, but to be honest, Victor, the chances are absolutely zero. The 25,000 sq km search area has no basis in reality, as will become clear if/when OI get round to searching it.
Happy New Year to everyone.
@Victor, @DennisW
I am not a fan of the new CSIRO area either, but Victor’s proposal of an end point around 34.3°S 93.7°E, based on an ultimate waypoint of the South Pole or constant true track of 180°, fits the BTO and BFO with relatively small errors. This end point does not fit the drift analysis as well as 30°S however, in my view. I assume that David Griffin would disagree with my statement.
If I include an impact point along the 180°T path, in the 25,000 sq km area, I assign a probability of 14.3% to the ATSB residual search area between 36.0°S and 32.5°S. Victor’s end point at 34.3°S has a probability of 7.0%, in my view. The entire ATSB First Principles Review search area of 25,000 sq km, excluding Victor’s end point at 34.3°S has a probability of 7.3%. On the other hand, my preferred end point around 30°S has a probability of 10.3%, in my view.
The higher probability at around 30°S, based on an ultimate waypoint of 40°S 100°E, is because of:
1. The transoceanic drift analysis results for all floating debris finds,
2. The fit of the RNZAF aerial search floating debris locations to the local drift analysis as demonstrated above by Victor.
Have OI agreed to first search the ATSB First Principle Review search area of 25,000 sq km because they believe it is more likely to be where MH370 will be found, or is the OI search priority based on political expediency to ensure ongoing ATSB support?
@Victor
Here was my first pass probability assumptions, from the prior thread:
>Prelim. Assumptions/Guesstimates:
95% Chance MH370 is between 20-40S +- 200 nm of Arc7
90% Chance MH370 is between 20-40S +- 60 nm of Arc7
80% Chance MH370 is between 20-40S +- 30 nm of Arc7
15% Chance MH370 is Unfindable due to Undersea Mountains/Trenches/etc.
>Results derived from the above assumptions:
20% of Arc7 searched to date between 20-40S +-30 nm of Arc7
68% Chance of finding MH370 between 20-40S +- 30 nm of Arc7
77% Chance of finding MH370 between 20-40S +- 60 nm of Arc7
5% Chance finding MH370 outside of 20-40S
10% Chance finding MH370 between 20-40S but further away than +-60 nm
So I am saying only 68% chance of finding MH370 +-30 km of Arc7 between 20-40S . The proposed 25000 km2 area is (rounding up) at most 10% of the remaining unsearched area, so that would be about 6.8% chance of finding MH370 in that territory assuming equal weighting of the remaining area.
As I do consider 180S a hot spot area, I would apply a 2x or 3x multiplier on the 6.8% chance.
@sk999 @Victor @Joseph Coleman
“The LAGOG waypoint was included as it was close to a supplied, HIGH CONFIDENCE position labelled ’19:12Z’. (Later ruled out)
Not sure if I might have a mini-finding, but if I compare UAE343 flight path from Flight Aware (courtesy Joseph) at about 19:12 UAE343 would be at LAGOG.
I feel I should motivate my probability estimates more;
1. The most northen part of the ATSB-priority area between ~32.2 and 33s is mostly not searched yet at all. South of this area everything close to the 7th arc has been searched with the high speed descent/dive impact scenario in mind.
Based on this assumption the plane is not south of ~32.6S imo; it has not been found there.
Based on the same assumption it could be in this not searched ATSB-priority area. I have great doubts about the high speed descent/dive impact scenario.
That’s mostly why I give only this area a 1 in 10 probability.
2.IMO Richard’s drift-studies can explain well the Roy/Mosselbay piece and lack of debris finds in WA but show a historical average buoys pattern passing too far north above Reunion and buoy arrival times half a year to a year too soon at Reunion and Madagaskar if coming from ~30S.
IMO the delays in arrival and finding times are too large with his model.
3. CSIRO/Griffin’s latest drift-analysis can explain the arrival/discovery time of the flaperon well but falls months too short of explaining the arrival/discovery time of the Roy/Mosselbay piece.
And their latest model predicts quite some pieces of debris should have landed on WA shores if coming from between ~34S and ~31S.
This is not conform GDP-historical buoy arrivals on WA from this region (none arrived) and nothing has been found there to date.
The satelite images are too uncertain/undetailed to serve as evidence/guide of possible hotspots. Then the RNZAF images taken on 28 and 29 March 2014, although also quite uncertain, do a much better job.
4.Using the aerial search between ~30s and ~34S as evidence the plane could not have crashed there can be refuted for the fact actually many objects were spotted and photographed in this area but never recovered.
@Victor’s images and the ‘bue panel’ with it’s associated debris field are proving examples.
5. The flaperon-like ‘blue panel’ and it’s reported associated debris field at ~32.4S/~97.8E are about smack in the middle of Richard’s and Griffin’s high probability latitudes. They were ~130 miles west from the 7th arc in the westward current calculated by CSIRO/Griffin before this current turns north-east.
This ~latitude could explain both Richard’s Roy/Mosselbay piece arrival/finding time and CRIRO/Griffin’s flaperon arrival/finding time.
And could explain the lack of debris found on WA shores.
6.Considering the disappearance was well planned from the start (which it has all signs of) also the end-of-flight and destination must have been well planned. The only reason I can think of regarding a planned end-destination would be an area that could serve hiding all evidence of the culprit as best as possible.
7. The best possible areas to accomplish this all are the trenches at Broken Ridge between ~95.3E/~97.5E and ~32.2S/~33S.
That’s why I give this area (outside the ATSB-area) the other 9 out of 10.
“They were ~130 miles west from the 7th arc in the westward current calculated by CSIRO/Griffin before this current turns north-east.”
My God.. have this west and east problem again.. please turn it around again..sorry
@Tbill
For EK343 to be passing LAGOG at 19:12 on the date in question along the N571, 49 Minutes 48 seconds later from a Predicted approx location at 18:22:12 UTC (6 23 29.26N 96 52 6.31E) EK343 would have to be covering a constant ground speed of approx 711.08kph. Please note these are only calculations at request, as we don’t know with 100% accuracy where at what time EK343 actually was. After 18:07.
Not sure if you’ve noticed flightaware have added Estimated and ADS-B tabs in the tracklogs.
@Kenyon,
Thank you for noticing the sign difference in the residuals label. I’ll fix that and post a corrected version at the same link.
@Joseph Coleman, @TBill: If we trust the FR24 data posted on March 22, 2014, by “Beggarman”, at 18:22, EK343 was on airway N571 at position N6.37,E96.94, with a groundspeed of 493 kn. The distance along N571 from this point to LAGOG is 323.5 NM, which would take 39.4 min at 493 kn. This would put EK343 at LAGOG at about 19:01, not 19:12, unless there was a slowdown from 493 kn to 388 kn. This is similar to the speed calculated by Joe using different methods. This makes it unlikely that EK343 was the 19:12 target that was first seen and then not seen.
@Victor. Taking you back to my low 7th arc log-on possibility, Don Thompson and I have discussed the lack of BFOs after the 18:25 log-on and before its subsequent IFE SMS/email connection, and whether any could have been expected early after the 7th arc log-on.
Citing the 12:50:26.768 cold ground log-on as a comparable example, the 4 transmissions between that and what in fact was that first IFE connection, at 12:53:02.729, were all ACARS related. ACARS being off at the 18:25 log-on would then explain the transmission lack there before IFE connection. It is likely that ACARS also was off at the 7th arc log-on at 00:19, similarly explaining lack of early transmissions after that.
So lack of these is not evidence of a crash from a low altitude 7th arc.
What also emerged was that though this subject has indeed been done almost to death earlier, including why the ACARS connection failed, the time lapse to be expected after the 00:19 log-on to the first IFE connection remains an outstanding question still.
In my opinion, the ATSB’s assumption that that should have been the same 89 secs as for 18:25, nothing more or less, remains dubious. To be the same implies the same circumstances when, just as examples, SDU power outage and so warm-up times were different. Of note also you will see the time from the 12:50 log-on to that first connection was 156 secs (though of course to reach that or beyond after the 00:19 log-on would require sufficient pipeline and residual fuel for the APU and retention of the satellite link).
Don Thompson leans to that 156 secs as the more likely for 00:19 but is looking into first IFE connection times some more.
As it now stands the elapsed time in which the first IFE connection should have been expected is anywhere it seems between the 60 secs he mentioned in a post above and 156 or more.
However what is missing still is an explanation for whether the first IFE connection necessarily takes so long, the circumstances that would alter that and supporting test results. In passing, while it is improbable that ACARS de-selection would affect that time, there is no ground log-on example (or, again, testing) which confirms that.
With more work, minimum IFE connection elapsed time for likely sets of 00:19 circumstances could be established with more confidence. In my view there remains a possibility, albeit small, of a rationale for a lower 7th arc boundary than as of now.
@ALSM
I’m working on the possible paths between an assumed early FMT and possible 19:41 positions, which requires me to look at the 18:41 BFO’s. Based on your summary (“BFO Bias as a function of Channel Type and Channel Unit”), is it correct that you suggest to use a BFO bias of 147 HZ for the first phone call C6 data?
@Richard
On the other hand, my preferred end point around 30°S has a probability of 10.3%, in my view.
Not sure what area you are implying above, but with 30S +/- 1 degree I would put the probability at around 38%. All somewhat subjective with arm waving, of course.
@David: Thank you for the update.
We expect OI to search a width of +/-25 NM from the 7th arc (assuming some altitude). If the altitude was low at 00:19, then the maximum distance from that time until impact is also small. Even though the 7th arc moves inward for a lower altitude, because the distance traveled gets shortened, searching +/-25 NM from the 7th arc, let’s say at some intermediate altitude like 20,000 ft, should adequately cover all possible altitudes at 00:19.
Do you have reason to believe that searching +/-25 NM from the 7th arc will not be sufficient?
Niels: Yes. 146 or 147 for the C channel BFO Bias. Note that the evidence is sparse, but that is the best estimate.
Mike
Niels: Here is the BFO Bias summary for all channel types: https://goo.gl/v6SvVt
@Victor @Joseph Coleman
OK thanks for the comments re: UAE343 I will look at it again to see.
http://mpegmedia.abc.net.au/radio/local_sydney/audio/201712/pam-2017-12-30-mh370-angry.mp3
@ALSM
Thank you, Mike. I understand the calibration is a bit tricky, if it had to be based on the second phone call data.
I like to stress once more to OI in particular for the sake of finding the plane the ‘Blue panel’ and it’s reported associated debris field has to be taken seriously.
This is the only one photographed piece in the aerial search of 28 and 29 March 2014 that can be positively identified as a probable aircraft piece.
I looked at every possible angle at it. I find no evidence it could not be the left wing flaperon of MH370. On the contrary. And no one supplied convincing evidence it wasn’t or wasn’t aircraft debris to date.
It’s location fits all possibilities including CSIRO’s latest drift analysis ~130 miles east ot he 7th arc and a solution between Richard’s and Griffin’s problems with African arrival/finding times and lack of debris finds in WA.
And it can fit the Isat-data too.
I know this would be a one in a million, for if true it would actually ‘pinpoint’ the crash site.
But it’s certainly not a reason to ignore this.
Add the image once again;
https://www.dropbox.com/s/stc5oqydhyysl2o/Xflaperon1.jpg?dl=0
@Ge Rijn: That area will likely be searched by OI. It’s not being ignored.
@Brian Anderson
Re: ”While not a “fault” I would have thought that ground might be pretty interested in the situation where all alerts/warnings were not acted upon.”
As Don mentioned, the crew alertness monitor is simply an ‘alarm’, designed to wake up the pilots if they become drowsy and fail to interact with the aircraft systems within a set period. The CMCS does downlink reports to the ground in the event certain maintenance messages are triggered, but those reports are for engineering purposes and are not a means of monitoring pilot activity. As far as I’m aware, the PILOT RESPONSE alerts are not monitored by the system. Regardless, no messages would be sent if the ACARS was disabled by a failure or deliberate action.
@DennisW
My probability estimate is based on the area from 27°S to 37°S with a width of +/- 25 NM being 80% probable.
Theoretically, MH370 could be anywhere from 6°S to 40°S and further than 25 NM from the 7th Arc.
The area 30°S +/- 0.5° with a width of +/- 25 NM is 10.3% of the total theoretical possible area.
Accepting that MH370 has not been found and not been missed in the ATSB 120,000 sq km, then the area 30°S +/- 0.5° with a width of +/- 25 NM has a probability of 15.5%.
Extending this to 30°S +/- 1°, gives a probability of 25.5%.
So we are nearly aligned, bar the subjective assessment and arm waving.
in any case, 30°S +/- 1° is more probable than any other latitude +/- 1°, in my view.
Hi Richard: I would pick slightly different numbers, but I agree with your overall assessment. For the topline probability (for your total area), I would say higher…maybe 85-90%. I would assign 80% to the ±20 nm width. I agree, S29.7±1° is the highest probability ±1° box, but there are a few others not far behind, including Griffin’s area(s). Given the Durban start + operational, logistical, weather and other considerations, starting on the SW end of the unsearched arc (cover what Phoenix missed) and moving steadily up the arc, covering ±20 or ±25 nm may be the most economic approach. (If we had a single hot spot with a much higher probability than any other ±1° box, I would go there first. Otherwise, mow from SW to NE (~S36 to S27).
I am just finishing David Mearns new book, “The Shipwreck Hunter”. It’s a great read for anyone with a bit of adventure in his/her sole. (Thanks Don.) But it will be especially interesting to MH370 hunters following the movements of OI’s Seabed Constructor. David goes into some detail about how he plans a search, down to the individual lines he will search, and the order of the lines to be searched. Of course, OI’s AUV technology changes the game. And the area to be searched for MH370 is much greater. But you can really get a sense of how many practical, logistical, political, regulatory and other considerations get folded in to the detailed search strategy. I highly recommend the book.
@Victor @Joseph Coleman
Re: EK343
So what I am saying is Flight Aware *estimated* EK343 path hits 92E (actually well north of LAGOG/N571) at about 19:12 at about 426 knots (after passing MEKAR at 18:30), and you guys are countering that the FlightAware *estimated* data is probably quite inaccurate re: flight path, times, and speed. I can see that you are probably correct about that.
In any case it remains interesting that ATSB originally claimed to have a fix on MH370 around LAGOG at 19:12, implying there was some radar data out there.
All: Seabed Constructor will leave the Durban dry dock early tomorrow morning. It now has 8 Hugan AUVs onboard and two new high speed Inmarsat terminals installed. They also cleaned the hull to comply with Australian Customs. General provisions would be next. Still expecting departure for the search area NLT Jan 4.
Make that 8 Hugins, not Hugans.
@TBill: I think what we know is there was a mobile Singapore radar source in the vicinity. What we ultimately learned is there was no capture at the NW point at 19:12. I think there was confusion because the information was passed to the ATSB cryptically, probably by the US, due to security and confidentiality concerns. Ultimately, the ATSB completely ignored the NW point.
For you @Tbill
EK343 Estimates at 929kph continued.
Estimated EK343 at 02:22:12 to MEKAR approx 2 minutes 48 seconds =02:25
02:25 MEKAR in 35 minutes 20 seconds arrive LAGOG AT APPROX 03:00:20
MEKAR TO LAGOG approx 547.1km (295.41037NM)
Bearing 295 along N571
Again these are estimates, and may not be 100% accurate as larger distances over longer times, are covered where possible changes of altitude and/or speed could have occurred. If you want anything close to specific information, try and put in a request for it to the investigation teams or even Tim Clark at Emirates. Not sure if you’ll get the reply you want though.
This may be wishful thinking but hopefully the MY investigation team include the EK343 path in final report as it could have been the closest aircraft to where MH370’s location is thought to have been recorded last on the radar. Even though perhaps no-one on EK343 had seen or heard another aircraft around that area at that time.
Here is the description of the 1912 position from the Final Report, Table 15, with my highlights:
“The 1B scenario had the aircraft passing close to a northwest point (8d 35.719’N, 92d 35.145’E) at 1912. This was an initial QUALITATIVE assessment of the POSSIBLE radar coverage from multiple data sources. This point was SPECULATED and used as the furthest point west the aircraft was likely to have flown.”
Cryptic, but in no way implying a positive detection.
The whole episode with scenario “1B” is a great case of confirmation bias at work (at least, that is how I would describe it). Initially it was intended simply as the route with the latest possible turn South (with 1A being the earliest turn South.) The first BFO analyses of the 1B route seemed to show a good match. The route also passed close to waypoints (which is where LAGOG came in, even though it was 35 nm from the 1912 point itself) and, in part, an airway. An independent Monte Carlo of routes starting at the 2nd arc and best matching the BFOs also came up with a similar route. The performance models (both from Boeing and from MAS) showed that there was enough fuel. The highest priority zone was picked to be where airway M641 crosses the 7th arc (even though it wasn’t the best fit to the BFO or BTO data.) Ocean Shield detected pings, right near M641 as predicted. We must be close! Except we now know that the BFO analysis was flawed, the pings were spurious, the association with waypoints and airways was wishful thinking, and the performance bounds are very broad and encompass countless plausible routes.
Joseph Coleman, LAGOG has no meaning, whether or not EK343 overflew it at or near 1912.
@SK999
I’m not suggesting any meaning for anything regards LAGOG, just worked out some estimates for @Tbill that’s all.
@All
Before I tried to obtain unique path solutions to the BTO/BFO data by introducing path curvature minimization in my path generation tool, see the link I posted back in September for the document:
https://www.dropbox.com/s/p477ng11mrezhc7/Straight%20paths_V0_8.pdf?dl=0
The pristine path (without offsets to the D(t) function) ends at -34.9 degrees, see figure 3 and app. B (l = 0 and m = 0). This path has a distinct curvature in the first 1 – 2 hours after 19:41, as well as a lower than average GS. So far I attributed this to a possible error in the 19:41 D value used.
What I have tried now is to take the 19:41 GS and track, and extrapolate the path “backwards” towards a possible point of early MT, estimate near IGOGU. The calculated 19:41 GS, track and position are 417 knots, 186 degrees and (0.8, 93.6), see the graphs in the document. The estimated distance from the 19:41 position to IGOGU is 407 nm (SkyVector), while the required track is 187 degrees (SkyVector). This looks like in reasonable agreement, except perhaps that the average groundspeed should be slightly lower than 417 knots.
Next I looked at the BFO for the first phone call, which again shows good agreement:
The calculated Doppler Residual D = -37 HZ (ANOKO, 18:41, 417 knots, 186 degrees)
The measured D = -37 Hz (based on 88 Hz for the BFO and a 147 Hz bias value)
All together this could be indicative that the calculated 19:41 track and speed is quite representative for the whole interval 18:40 – 19:41.
Note also that there are other paths, namely for l = -m (zero mean offset to the D(t) curve), that show a reasonable match in speed/distance/track (however with a worse fit for the first phone call BFOs).
What emerges as a possible scenario based on the calculated pristine path in combination with “backward extrapolation”:
An early MT near IGOGU, followed by an interval at 400 – 420 knots connecting to a 19:41 position around N0.8. Possibly this track is continued for a while before turning straight south, following the E93 meridian towards a crash location in the S35 area. Note that this is a typical “best fit” scenario quite similar to the old “Inmarsat” scenario (with a slightly different speed profile).
@sk999: I’m not going to belabor the point, but I’ve had enough private conversations regarding this NW point that I believe my account is correct. There was confusion within the ATSB surrounding the meaning of the data point, clarification was difficult to obtain, and it was ultimately ignored.
@Niels
I agree that a flight path that goes to the vicinity of ISBIX and turns due south between 93-94E to 33-35S is one hypothesis that seems to fit.
@Victor. You asked about sufficiency of a ±25 NM search width. The First Principles Review Report conclusions, p23, state that, “99% of the DST Group analysis results lie within 25 NM to the east and west of the arc”.
The 25 NM comes from its earlier p17 where it says the final two BFOs, “indicated that the aircraft was likely to be on an unstable flight path” and, noting, “recent simulations showed high rates of descent” the participants agreed, “that the aircraft was likely to be within 15 nm of the 7th arc”, and that searching either side of the 7th arc could be reduced to 25 NM. Curiously it drew those remarks from this the ATSB’s Search and Debris Examination Update of 2nd November, 2016 which was released coincident with the Review’s meeting, 2-4 November. Yet the 2nd November report at its page 14 described the previous search width of 40NM as an “appropriate width to encompass all uncontrolled decent scenarios from the simulations, ie the more recent simulations.
There was a change of heart at that meeting it seems. In fact the 25 NM seems to originate from the 15 NM mentioned above, plus a 10 NM margin, mentioned at page 13 of ATSB’s Definition of Underwater Search Areas of 3rd December 2015, which appears arbitrary.
In summary, the First Principles Review meeting reduced the ±40 NM assessed immediately before as “appropriate”, and which included unsimulated possibilities such as APU start up and main engine brief restarts. It reduced this to its 25 NM based on the 15 NM DSTG assessment of 99% probability within that, plus an arbitrary 10 NM.
The 15 NM sounds reasonable if one accepts the implied probability that divergence outside the simulator results is unlikely. Personally I fear that is wishful thinking but then again that august body really should have a better idea about than I (at least) do, and so I defer to that. However, the 10 NM round figure, not being supported by argument, looks excessive to me.
About your 20,000 ft intermediate altitude for the 7th arc, the Review settled on a 2½ NM offset (see the report’s footnote, page 17 again and then the note bottom right figure 14, page 23). The note seems to overlook that the 25 NM SE width should be reduced by 2½ NM but I leave that aside. If one allows 10,000 ft below their 40,000 ft 7th arc datum for cruise altitude and a further 7,000 for height loss from MEFE to the log-on (around 1,000 ft during a ½ min descent after MEFE, 4,000ft over 1¼ minutes while banking and then 2,000ft during 15 secs downwards acceleration to the log-on of the last BFO) you end up with 23,000 ft.
To me that offset looks about right in that context.
The 25 NM search width is likely to be contracted for the 25,000 sq km search. If unsuccessful should it apply to subsequent spot searches?
Here optimisation intervenes. First, the 10 NM part of that is arguable. Second if there are to be spot searches to that width, each one will in my opinion deny search elsewhere along the 7th arc. It seems possible that the arc could be searched from the 25,000 sq km to 28˚S, with 1½ sweeps each side, say 8 miles or possibly even more. Search time will be limited either by weather, cost or other jobs on offer. To select spots for individual searches to 25NM is to me an unaffordable luxury. Each one will entail using that precious search effort on progressively lower probabilities as the 25NM limit width is approached. The length of each spot (and there is talk of margins of 1˚ along the arc at these) will mean about four times their aggregate length is searching foregone along the 7th arc hot line to an 8 NM width, that is with transit and turning times included. It just seems to me that spot searching risks being an indulgence, ie picking favourites when there are no outsiders.
Extend the season by doing the arc in more than one section to the 8 miles or whatever else is adjudged achievable, starting at the southern end as you say. An even number of sweeps would allow that to be reasonably efficient, ie minimise transits between sections.
All that said, practical factors like having to return to harbour might intrude. I hope they have a ferrying vessel available to minimise that.
@Victor. Please delete, “DSTG assessment of 99% probability within that” from the 4th para second last line.
Joesph Coleman, you say “I’m not suggesting any meaning for anything regards LAGOG, just worked out some estimates for @Tbill that’s all.”
Fair enough! Even if LAGOG has no meaning, it can be useful to work out thse estimates anyway, just in case a situation arises. You never know.
Victor, you complain, “I’m not going to belabor the point, but I’ve had enough private conversations regarding this NW point …”
Sorry, I’ve not been privy to any of said conversations. All I did was quote verbatim from the Final Report and then add my own commentary. Note that I did so after rereading your article from Feb 11, 2017 titled “Singapore Radar and MH370”.
I should be more blunt. The 1912 point was introduced and given high prominence by the JIT. The JIT no longer exists – such is life. However, there remains as a stern warning – beware confirmation bias (CB). It is rampant. You asked where OI should search. The vast majority of the responses are tainted by CB. All of David Griffin’s recent analyses suffer from it as well. Mowing North is the best answer we can give. There are no hot spots.
@Victor. You asked about the 25,000 sq km search prospects. The same can be asked for a search after that. Not that you did ask but if not found there I would then extend my 1 in 2 chance for that to the same for a search up the 7th arc to 28˚S, the total amounting to 3 in 4.
The other quarter is in the vast low probability area.
@sk999: I don’t disagree with the approach of proceeding north at a constant width. It would be incorrect to assign high probabilities to the assumptions used to determine the hot spots. Paraphrasing one commenter, the hot spots are barely warm.
@David: My probabilities are similar to yours and what others have estimated. I’d say about 40% success in the 25,000 sq km area, and cumulatively up to about 70% success if the search proceeds to 26S, all at a width of around +/-25 NM. Possible reasons to not succeed include a long glide with an impact further than 25 NM from the 7th arc, failure to detect the debris field during the upcoming or during the past searches, or a crash further north along the arc than 26S (which is more probable if the BFO bias drifted). There is also a very small chance our interpretation of the satellite data is incorrect, either due to equipment failure, unknown behaviors, or tampering with the data.
Is OI going to feed live data to ATSB the same as Fugro did ?
ventus45: The AUV real-time bandwidth is limited, so there is no high fidelity sonar imagery available to transmit in real-time. The data has to be retrieved at each battery swap, which will occur approximately every 50-60 hrs. Data will be sent to the onshore OI data processing center (at least first), not ATSB directly. I’m sure data will be shared as needed.
@Victor. Included in my one quarter chance of failure of a new search is the possibility the aircraft is in fact lying within the probability density function of the previous search but in the low probability area not in fact searched, which constituted about an eighth of the PDF.
Not much of that would be picked up in the 25,000 sq km search.
@All: I think there is a lot (but not 100%) agreement about how the search should proceed and the probability of success. So, let’s re-visit a question that’s fundamental to guiding the new search: What assumptions (if any) were incorrect in the DSTG’s Bayesian analysis that led to a posterior distribution that most of us now believe is incorrect, or at least has been modified by the null search results to date? I’ve already expressed my thoughts on this subject matter, but it is helpful to understand what others believe.
@Victor
In my view, the single biggest error in the DSTG’s Bayesian analysis is that they did not consider that a descent at 18:39/18:40 UTC was possible. They worked out that a descent at 00:19 UTC was possible, why was the same mathematics not applied earlier in the flight?
@Victor
I would make two points:
1. The model used a simple model of the aircraft motion that allowed few manoeuvres. This allowed it to be validated, since it corresponded to commercial flying, but might not be applicable to the accident flight. This is an observation, rather than a criticism, since constraints had to be applied to get a useful result.
2. In the application of the model to the data, it appears that the data was allowed to select the autopilot modes, or at least to bias the autopilot modes that appeared in the final path set selection. The alternative would have been to only ignore mode/path selections that were statistically excluded, at some level of significance. The LNAV/CTT modes might have given the best fits to the data, but that’s not relevant.
However, in mid-2014 the target size for a search area was 60,000sq.km. As of now, if the search is to extend north of 30S, the total area will become 200,000sq.km. It would be unfair to criticise post-facto the search authority for being too conservative, when their target was so much smaller.
@Victor
The basic incorrect assumption was a passive pilot pilot after about 18:40. The Bayesian analysis gave priority to simple paths with one FMT turn at 18:40 without maneuvers. In my work, the only ping period I see possibly without maneuvers is Arc3 to Arc5. Part of the ATSB rationale for de-emphasizing maneuvers was that those solutuons would be too difficult to model, and not yield an obvious “hot spot” (eg; if we re-onsider starting from anywhere on Arc2).
Yet I feel the path could have been relatively straight including some maneuvers. I feel there are probably “exact” solutions to the Arc3 to Arc5 period, and then we have to figure out what maneuvers could give exact solutions to the remaining pings. Then of course it is possible the “exact” solution is wrong (if the BTO/BFO data has error bars) but at least we would know what the data says, assuming it is correct.
@Victor
I agree with others that one of the biggest ‘mistakes’ has been to assume the 18:40 BFO’s represented a FMT. The coincidence of that first call being just in that FMT was/is just unrealistic imo.
The call caught the plane in a descent that started somewhere after 18:25 and before 18:40 is far more logical.
Consequently the FMT or MT happened much later and I argue at- or just before LAGOG (around the 19:12 point).
Then the assumption the flight went into a ‘Ghost-flight’ (I know you don’t like the term but imo it’s a clear call) after 18:40 is not based on any logic or evidence. On the contrary; this assumption is against all statistical odds.
But this assumption led to the believe the end-of-flight was also uncontrolled which could turn out to be the major ‘mistake’ OI could make also.
@Richard G, @Richard C: I’ll play the devil’s advocate. The BFO values at 18:40 approximately matched the BFO trend from 19:41 to 00:11 for unpiloted, southern, level flight, so the assumption of a southern, level flight was extended to 18:40. Also, although it probably was not considered, the variation of BFO values at 18:40 is low for a typical, idle “pitch controlling speed” descent, also suggesting a level flight at that time. I’ve made the argument that a V/S descent, where pitch controlling vertical speed, could produce stable BFO values, but we don’t know for sure.
But even ignoring all BFO values, DSTG’s analysis stills calculates a peak in the posterior distribution in the previously searched area. Only by eliminating certain navigation modes (e.g., LNAV, CTT) does the distribution shift to the north, probably requiring speed changes along the way along the curved path. But we also know that there are LNAV and CTT paths that terminate in the new search area if we leave the latitude at 19:41 to be unconstrained. However, these paths require a “loiter” of some kind between 18:28 and 19:41. Again, since a holding pattern, for instance, was considered to be multiple manoeuvers as opposed to a single, structured manoeuver, it was modeled as low probability in the DSTG analysis.
So, basically agree with both of you that assumptions of level flight at 18:40 AND assumptions about the number and kind of manoeuvers resulted in a peak in the PDF near 38S latitude.
By the way, I don’t offer this as criticism of the DSTG’s work. Rather, it is an attempt to understand what assumptions led to the previously failed search.
@TBill said: The basic incorrect assumption was a passive pilot pilot after about 18:40. The Bayesian analysis gave priority to simple paths with one FMT turn at 18:40 without maneuvers.
I would say that the Bayesian analysis assigned higher probabilities to paths with fewer manoeuvers. An assumption about a passive pilot was not really invoked.
@Ge Rijn: Again, in the DSTG analysis, there was not really an assumption of no pilot inputs after 18:40. Rather, the fewer the pilot inputs, the more likely the scenario.
I think the evidence leans towards an uncontrolled flight after fuel exhaustion and the search will likely be conducted on this basis. However, in my opinion, there is a small, but non-zero, probability that the plane entered a controlled, long glide, and crashed outside of the planned +/-25 NM zone. If this occurred, likely the plane will not be found by the upcoming search effort. That’s one of the reasons I assigned a success probability only 70% if the area all the way to 26S is searched at the planned width.
@Victor
However, these paths require a “loiter” of some kind between 18:28 and 19:41. Again, since a holding pattern, for instance, was considered to be multiple manoeuvers as opposed to a single, structured manoeuver, it was modeled as low probability in the DSTG analysis.
With the assumption that the flight was diverted for a reason, the loiter makes sense. Also the turn South under those circumstances would be at a preselected cut-off time relative to a negotiation scenario. It would be logical to pull the plug at an even hour time i.e. 19:00. A turn at 18:40 +/- would be more compatible with a position threshold, not a time threshold. I know, weak, but something to consider.
@Victor
The combination of a very controlled flight from 19:41 UTC to 00:11 UTC followed by a very uncontrolled flight at 00:19 UTC, would suggest there was no pilot in command. As long as fuel was available then all aircraft systems function perfectly and the aircraft proceeds under autopilot, but when the fuel is gone, the inevitable happens and an uncontrolled dive occurs.
@Richard,
I agree. The landscape changed around 19:00, and my personal opinion is an autopilot flightpath (lots of choices) from that point in time until fuel exhaustion.
@DennisW,
Your “threshold” of an even hour is possible, but I would not call it a ‘threshold” that he “waited” for at some geographic location, (ie, “held” somewhere), before heading south.
In his planning, he would have had to have considered it a very high probability that negotiations would fail anyway.
In reality, he had to head south from the start, on the basis that it was the most probable end game anyway.
In that scenario, he would have headed south, and could have waited until about 21:00 before making his “final” decision.
Up until that time, he would have either “passivly” monitored HF radio, (co-conspirators would communicate with him by HF radio, {either ham radio or marine yacht equipment}, not satphone, for a whole host of reasons, ie, equipment availability, ease of use, and “security” for them, i.e. non traceability / locateability / arrestability) to see if negotiations succeeded or not, or he could even have “actively” communicated with them, perhaps a number of times, between 18:30 to 21:00.
Effectively, this “time” became his “point of no return”, since up until this “time”, he was still able to “return to land”, quite safely. After that, not, so the “die was thus cast”.
I therefore consider 21:00 (approx) to be a “decision point” to “CONTINUE” south.
For these reasons, I think TBill and I are on the same page (conceptually) regarding how the modelling should have been done, with respect to the 3rd, 4th, and 5th arcs (20:41, 21:41, 22:41).
Instead of working “both ends against the middle”, as has been done up until now, we should work “from the middle out”, ie, “backwards” to get the northern and southern limits that it was possible to reach on the 3rd arc (20:41) from the Penang Hamid phone detection, without being constrained by the Lido Radar Slide at all – i.e. totally ignore it. Only then, can we work “foreward” off the third arc.
@Richard: Yes, likely that happened. But in the event that the search is unsuccessful, the possibility of a long, controlled glide would have to be considered, as would the possibility that the plane is further north along the arc, as well as the possibility that the wreckage was scanned and not recognized. Of these possibilities, I think a more northerly impact would be most likely, but that’s only an opinion.
@Ventus
Valid points, IMO.
@All
Happy New Year to you all 🙂
Hopfully the New Year will bring the answers we are all looking for.
And thanks to Victor for creating and continuing this informative discussion forum.
Happy New Year to all from Boulder.
@ALSM
Boulder is in the USA. right? I think I flew over it a few times. Just kidding from Silicon Valley.
Indeed, Happy New Year to all from Sydney (Australia).
We had our big bang up fireworks show 12 hours ago now.
Today – the klean up.
Allowing the possibility of a descent at 18:40 will likely not change anything – the timing of the FMT is NOT driven by the BFO of the 18:40 phone call but rather it is driven by the BTO data and the requirement that the models account for the aircraft at all times (Fig 10.3 of BM). The BFO data serve mainly to exclude Kazakhstan.
The BTOs vary very smoothly after the restart of the SATCOM, unlike, say, the BTO’s of the MH371 flight. There is little, if any, room for a fidgety pilot.
As has been discussed before, the DSTG heat map can be constructed to first order by combining three narrower Gaussians – one for LNAV routes, one for CTT routes, and one (somewhat broader) for CTH routes. In the DSTG analysis, all five autopilot modes are considered equally probable, but there is no foundation for this assumption other than ignorance. Because the heading/track is nearly N-S, the LNAV and CTT routes pile on top of each other, giving rise to a really sharp peak around latitude -37 degrees, and perhaps giving the artificial appearance of a higher likehood in this region than is warranted.
Conversely, magnetic track routes are disallowed because the lower Mach limit is 0.73, even though that value has never been explained properly. In fact, in March and April of 2014, Boeing was merrily calculating performance curves for speeds as low as 350 knots (if not lower). Why the change?
Are there nuances in track and hading mode that are not included in the DSTG models? E.G. a sudden gust of wind blowing an aircraft sideways will cause it to follow a new path that is offset from the original. Is this effect important?
The SS Central America search was considered a success, but it required searching out to the 94% (basically 2-sigma) boundary to find it. The search for MH370 is still far from that boundary. Which situation has the more uncertain data?
@sk999
As you said earlier, go North. There is not much more to say than that.
First; Happy New Year to all from Amsterdam Holland 🙂
@VictorI
I would argue a long glide is not necessary. If a deliberate glide was involved the objective would have been to ditch the plane at a specific location. Not to glide a long distance.
Rather it would have been done in the shorterst possible time coming from altitude.
In this light a steep descent first and then a pull-out and (short) glide could make sence imo.
@All,
I have made a few minor revisions to my OCXO Transient fitting program.
Thanks to David and Richard and others who made helpful suggestions.
Here is what’s new:
1. I made the residuals column sign consistent with the label.
2. I added a note on Event 2 highlighting the fact that it was determined by Holland to have a long depowered time, but it only fits with a late LOR time delay w.r.t. the peak overshoot, which is consistent with the shorter-depowered-time events with starting temperatures closer to the high set-point temperature. The cause of this discrepancy for Event #2 is unknown.
3. I updated the graphs showing the BTO and BFO errors and the aircraft track with a FMT just before 18:40 which is consistent with the phone call BFOs then.
4. The notional FMT I used here occurs at 18:37:15 and goes from 296 degrees true to 180.0 degrees true, passing through IGOGU/ANOKO/NOPEK.
5. At the same time as the FMT begins, the airspeed is reduced to 418 KTAS (which is M0.700 and 240 KCAS at FL340 at +10.9C).
6. Note that fitting the BFOs constrains the TAS. Then, depending on the flight level assumed, one can find the equivalent Mach and KCAS.
You can get the new Rev. 1 version HERE .
I have also been working on fitting the BFOs at 18:40. The difficulty here is to find a combination of Flight Level, true airspeed, course, and ROC/ROD which is set just prior to 18:40 that is consistent with: (a) the BFOs then, (b) the subsequent BTOs/BFOs, and (c) the MEFE time. If you can find it, then you have a candidate route for an auto-piloted flight into the SIO for which no pilot inputs are required after about 19:00. I have been analyzing single-FMT routes, which are a separate class than the extended-loiter case proposed by Richard and VictorI.
One complication is that the BTOs and BFOs depend on the TAS, whereas the fuel consumption is more directly dependent on KCAS. In addition, the auto-pilot speed control is never based on TAS, but rather on Mach or KCAS (and Flight Level and weight, of course). Thus, there is a requirement to find the combination of Flight Level and speed control mode, which simultaneously matches the TAS-dependent parameters BTO and BFO while also matching the known endurance (which is roughly dependent on KCAS). I have been studying this, and I am preparing some graphs to illustrate the problem using about six solutions to fitting the 18:40 BFOs. These 18:40 BFO fits include descents and climbs. We know the Flight Level had to be within a fairly limited range during the military radar track because of the very high TAS (about 498 KTAS, which is achievable at M0.84 at FL340). However, there is no Flight Level at that TAS that matches the MEFE time because of the high fuel flow. Therefore, a slow-down must have occurred after 18:22 (and before 19:41). My SLOP fit appears to indicate that no slow-down occurred until after 18:29, and it may have occurred prior to 18:40.
A critical question is, can the 18:40 BFOs be fitted with a single FMT and simultaneous slow-down which also fits the MEFE using a standard speed mode at an achievable Flight Level? Stay tuned for more on this subject.
Happy new year to you all from Bandera, Texas!
@All,
If you have a problem with the link above to my revised OCXO fitting program, use this:
https://drive.google.com/file/d/1d8Td73ayKCQyRzGwUBEKCPyYA0XYJtz2/view?usp=sharing
The “Seabed Constructor” is berthed at T Jetty west in the port of Durban. Close to the town, so a few sore heads can be expected following New Year’s Eve celebrations.
As I write, the bunker barge “Smit Lipuma” is alongside, presumably undertaking the refueling operation.
This new search for MH370 appears nearly ready to go.
Happy New Year to All from Frankfurt, Germany!
A video of the fireworks going off across the Frankfurt skyline shows how the Germans spend €137M in the space of a few minutes and dump up to 10 tonnes of particulates on each city.
https://www.dropbox.com/s/6igqitx44azxukz/Frankfurt.mov?dl=0
Best wishes to all for the coming year from a very blustery Northern Ireland.
Here is a useful reference, given a significiant topic in this comment stream.
@Don Thompson
Your reference does not show up (here).
Btw, Holland had an all time high record on new years day; 14.5 degrees C.
@Richard
The Dutch spent nearly E70M with only ~1/5 of the German population..
It’s a poluting waste but a beautifull one imo. Like a vulcanic eruption we’re all in awe to look at too..
@Ge Rijn
No fireworks in California this year even at our beach place. The state is as dry as a bone. Fire danger is still very high even in Northern California.
Link, yes, I may have omitted to paste the URL into the href tag.
Trying again. Some additional, referenced, reading might be useful.
@Don T
Interesting. Thanks.
Kahneman and Tversky wrote a lenghty paper related to the same subect as your link. It is a long read, but the important take-away for me is below.
The production of a compelling scenario is likely to constrain future
thinking. There is much evidence showing that, once an uncertain situation has been perceived or interpreted in a particular fashion, it is quite difficult to view it in any other way.
Does that sound about right for our problem? Of course, that does not apply to me.
https://msu.edu/~ema/803/Ch11-JDM/2/TverskyKahneman73.pdf
@Don
We believe almost certainly there is a very good chance and highly likely better than even, that the probable outcome is likely to be probably the best. We doubt there is almost no chance that it could be worse. Chances are slight that the outcome is probably not what we want and it is unlikely that the outcome is improbable. There are, of course, those who say it is highly unlikely and we have little chance.
@DennisW
Yes, I’ve read you got your fireworks plenty down in California. Nature doesn’t give a damm about polution or destruction doesn’t it. And the sight of it (at a distance..) is the same kind of breathtaking as new years fireworks above Amsterdam, Frankfurt or Sydney. It reminds us how small we are against the forces of nature. Fireworks at new years eve are in a way more or less controlled childs play to remind us of those forces..
@All: Happy New Year, and thank you for your diligent efforts to solve this mystery. I am confident that we will learn more in 2018, even if simply eliminate some possibilities.
@Don Thompson
This makes me think about the catogories of found debris. Confirmed, highly likely, likely, probably, maybe.
The same with drift-analyzis between ~40S and ~26S. By now; 40S/38S impossible, 38S/36S highly unlikely, 36S/34S likely, 34S/32S highly likely, 32S/30S likely, 30S/28S probably, 28S/26S maybe. Beyond impossible (imo).
This kind of probability-statistics in this case is based upon known facts and data ofcourse not just random believes.
@Victor @Niels
As you know, we just do not know what maneuvers happened between Arc1 and Arc2. I would argue, as an analogy, we also may not know what maneuvers happened between Arc5 and Arc6.
If 18:40 was a turn south, that is controversial to me, as I feel that is an unlikely flight path. To me that implies there was either (1) some data/knowledge at ATSB that MH370 indeed went close to Samatra, or worse (2) some knowledge MH370 instead went up towards LAGOG and the path via Sumatra is false to placate Malaysia, because anything towards LAGOG semi-proves intentional flight. Besides loiter, the other reason to head towards LAGOG or DOTEN is to decoy a flight north with an IGARI-style U-turn.
Kind of crazy to apparently have a creative pilot making all kinds of turns from IGARI to MEKAR, and then put a “modeler’s convenience” kibosh saying no creativity after 18:40, because “commercial jets normally fly maneuver-less paths”. Gimme a break.
Maybe there was an operational reason to turn south at IGARI (like getting away from EK343 or the Singapore radar – two flights we know in the area). In that case Niels recent case makes sense to me, in so many words, continue westerly after 18:40 southern jog, followed by due South 180S track at approx. ISBIX or BEDAX.
If we could go back to your past article about Hollands paper, what I would now say we should not be expecting a straight BFO line for a straight flight. If you do a sample calc of a straight flight south after 18:40, I believe the shape of that BFO line is a curve, especially as the aircraft (MH370) is passing the close-approach to the satellite. The fact that the BFO line is straight actually infers some maneuvers in the time period, which most models are assuming either (1) a big slow down happened before 19:41 and/or (2) also assuming the 19:41 BFO is in error a little bit, and taking the liberty to assume a preferred BFO is more correct. Not to mention I am totally uncomfortable assuming 1840 and 2314 channel C BFO can also be adjusted manually to fit the other BFO’s from the main channel.
Correction above:
Maybe there was an operational reason to turn south at *IGOGU* (not IGARI)
@Ge Rijn:
We had one “fire works barge” explode just off-shore from Terrigal Beach just north of Sydney last night.
Both explosives techs went swimming – but were recovered unharmed – fortunately.
A few wayward projectiles impacted the crowded beach area – fortunately no serious injuries.
https://www.youtube.com/watch?v=IXW8N7rlbb4
https://www.youtube.com/watch?v=AXZz6w3PKBk
@Ge Rijn: Regarding glide distance, I was only referring to reasons why the new search might not succeed. Even if the search proceeds at +/-25 NM to 26S, there is still a significant chance the wreckage will not be found. I give it better than even odds that a search as far north as 26S will find the plane, but not finding it does not mean that much less likely theories, such as a termination in the Maldives, Diego Garcia, Bay of Bengal, South China Sea, or Kazakhstan, should be considered unless new evidence surfaces to support one of those locations, or new evidence arises which contraindicates a crash in the SIO. Contributors to this blog implicitly understand this as the contributors here are earnest and informed. Much less so other places.
@TBill: If we allow a loiter, there are autopilot, constant mode paths after 19:41 that satisfy the satellite data over a range of termination points along the 7th arc. If we remove the constraint of no pilot inputs after 19:41, then that only increases the range of termination points. We believe that the approach chosen by OI is to first search the 25,000 sq km advised by the ATSB, and then proceed north along the 7th arc at +/- 25 NM to 29S or further north, depending on time, weather, and allocated resources. I hope the search can proceed to 26S if necessary.
I don’t think the possibility of pilot inputs after 19:41 provides rationale for changing that search strategy. At this point, it really doesn’t matter what the models predict, as long as predictions that have the highest probabilities are searched.
@VictorI
Most we have is the found debris. From the confirmed, highly likely and likely items of them their (basic) damage analysis show: only tension damage none- or very little (impact) compression damage. You can bend this facts up side down the way you like it. But this clearly indicates there was no high speed dive impact.
Combined with the fact more than 90% of all debris found to date is related to trailing edge, surface control, engine cowling, nose gear door and flap fairing parts evenly spread along its wings it’s absolutly clear the plane went into the ocean with pitch-up, wings level and relatively low speed.
Deliberatily or not the debris shows something like this happened.
Most important is OI is starting the search again but ignoring these facts again could turn out to be a great mistake imo.
@DrB
Many thanks for the updated spreadsheet with the results from your OCXO Transient fitting program.
The SLOP locations and timings you propose all align with my MH370 flight model between 18:22:12 UTC and 18:28:15 UTC.
An FMT at 18:37:15 UTC onto 180°T and reduction in speed to Mach 0.700 aligns with an End Point at around 33.7°S near the 7th Arc.
However, the interim BTOs from 19:41 UTC to 22:41 UTC appear to be out of line or I have made a mistake somewhere.
@Victor
I agree allowing for pilot input does not change the search, because I am thinking pilot input could have been to follow an approx. 180S path. Not sure but that might become more clear if the Bayesian analysis was done to look for that possibility. The way they did the Bayesian was presumably valid for the assumptions made, but if it was me, I would have explored other sets of assumptions…maybe they did behind the scenes.
@Ge Rijn,
I am seriously concerned for your health, IMO you clearly demonstrate symptoms consistent with an absolutely chronic infection of Vanceritis.
It is a fact that relief from this condition is likely with increasingly frequent exposure to disconfirming evidence.
Ignoring the symptoms of this chronic condition, exacerbated by long-term bias constipation, would be a great mistake.
@TBill: The Bayesian analysis does not assume there were no pilot inputs after 19:41. Rather, there is the assumption (as shown in the prior distributions) that more manoeuvers are less likely than fewer manoeuvers. At the risk of my being repetitive, the implication is that a hold would be a low probability event, as it would require a number of manoeuvers. In reality, a hold should represent a single manoeuver, and could be described by a single parameter representing the length of the hold. One consequence of allowing for a hold manoeuver is the peaks flatten in the posterior distribution as a function of latitude.
@Don Thompson: Speaking of Larry Vance, I will disclose that I exchanged some email messages with him regarding his theory that the recovered parts of the flaperon and flap unequivocally indicate that there was a ditching. I asked him to share the analysis behind the statements he made to the media on this topic. I explained that his theory would benefit from peer review, and I offered this blog as a medium for him to expound on his theory in a way that it is not possible in other venues. He explained that he is publishing a book on this topic and he would not release details prior to publication. He said he would share more details with me if I signed an NDA, which I refused to do.
@Richard,
You said: “Many thanks for the updated spreadsheet with the results from your OCXO Transient fitting program. The SLOP locations and timings you propose all align with my MH370 flight model between 18:22:12 UTC and 18:28:15 UTC. An FMT at 18:37:15 UTC onto 180°T and reduction in speed to Mach 0.700 aligns with an End Point at around 33.7°S near the 7th Arc. However, the interim BTOs from 19:41 UTC to 22:41 UTC appear to be out of line or I have made a mistake somewhere.”
First, I am glad to hear your SLOP predictions using my timings also match the CBTOs/CBFOs, confirming my model calculations.
Next, you did not make a mistake. I said that the FMT to exactly 180 degrees was a notional example. It was not intended to be a full solution to the subsequent CBTOs/CBFOs. You apparently assumed, quite reasonably, the lateral navigation mode was CTT after the FMT in order to compute the CBTOs and get the 33.7S intersection.
However, it is easy to show that a solution at 180 degrees true track from 94.4E (i.e., the IGOGU/ANOKO/NOPEK N-S line) cannot possibly fit the 19:41 CBTO because it misses that arc altogether by too wide a margin.
One has to get as far west as circa 93.75E after the FMT to just touch the 19:41 Arc. Therefore, either a due south turn had to occur from a point west of ≈93.75E, or the initial track southward was at a bearing greater than 180 degrees true.
The first possibility, that the FMT occurred west of circa 93.75E, can be refuted because, if one starts with the SLOP track after 18:28:15, then the farthest west one can get at the end of a FMT before 18:40 is only circa 94.2E even if a due west turn is made immediately after 18:28:15. So two turns after the SLOP to maximize westward travel still cannot touch the 19:41 Arc with a 180-degree track.
The second possibility (a final track > 180 degrees true) is what I showed last year (2016.08.12) – a candidate CMT route with a first FMT to ≈220 degrees true and a second FMT to 186/187 degrees true/magnetic near NOPEK.
My conclusion is that if one accepts the SLOP maneuver, which I do, and if one does not use an extended loiter after the SLOP, then there are only two possibilities for the FMT: (a) if the FMT was a single turn, it was to a course greater than 180 degrees true (and more likely circa 186 degrees), or (b) if the FMT was actually two turns, the first turn was before 18:40 (and it determined the 18:40 BFOs) but a second turn further toward the south also occurred between 18:40 and 19:41.
For the two-turn scenario, the first turn would be to circa 215 degrees if there were no changes in speed and no climbs/descents. With a slow-down to Holding speed, the first turn would be to circa 204 degrees. You can easily get 204-215-degree tracks after the SLOP by simply making a left turn to ANOKO (or, less likely, to NOPEK), with the exact timing of the turn establishing the course to ANOKO. That part is simple, and ANOKO is a standard arrival waypoint for WITT in Banda Aceh. Normally there would also be a slow-down by the time ANOKO were reached. Thus, there is a potential rationale for a turn to ANOKO and a slow-down. But what could have happened next to create the “final” SSW course?
I think the single-turn FMT scenario is more plausible because of its simplicity, than the two-turn FMT, but then one has to explain a curved track which begins at greater than 180 degrees true bearing. This is possible, as I have previously demonstrated (2016.09.01) with a CTH route, because of the local winds. That CTH candidate route has two FMT’s, the second one being at BEDAX. However, for a CTH route, one has to flip the NORM/TRUE switch to TRUE (even with an EOR error), and we currently have no rationale for the PIC to do this.
For a (curved) CMT route, why would (an unimpaired) PIC dial in an odd number like 187 using the MCP TRK SEL/HLD? I could accept the possibility of the PIC (or even a cabin crew member) manually setting exactly 180 degrees in the MCP, but why 187 degrees, and why near BEDAX?
@DrB: We believe that OI will search the 25,000 sq km recommended by the ATSB. After that, it will continue north along the 7th arc, probably at a width of around +/- 25NM. I question the value in selecting one scenario among the many possible scenarios that result in a termination in the 25,000 sq km area. On the other hand, if you believe the search area is incorrect or should be modified, that would be helpful input.
@Victor,
These two photographs, the left wing and the right wing, illustrate the dilemma (a sclerosis?) that is vanceritis. The images depict the destruction wrought on the wings of an MD-11 airliner, HB-IWF, as a result of its impact with the ocean.
The dilemma is reconciling how a B777 flaperon and, to a lesser extent, its adjacent outboard flap do not exhibit the extent of damage that is evident on flight surfaces of similar construction in these images: yet, every other recovered piece of 9M-MRO exhibits at least similar, or more catastrophic, damage as exhibited by the wreckage items collected in those images.
@Don Thompson
At least I’m one of few not infected yet with the wide spread chronic form of ‘diveritis’ which causes an irreversible condition called ‘tunnel-vision’ if not treated with enough antidote in time.
@Ge Rijn
Medical advice is always that self-diagnosis should be avoided (“At least I’m one of few not infected“)
@DennisW
Thanks for that reference. Your quote…
“There is much evidence showing that, once an uncertain situation has been perceived or interpreted in a particular fashion, it is quite difficult to view it in any other way.”
That statement may be true for an uncritical observer, however, the critical analyst must view a situation or scenario from many possible perspectives, no one of those perspectives may be complete but the composite will be more informative than only one. It can be challenging to understand that composite picture.
All,
What better way to start the New Year than to finish off old business. In my case, it was to clean up the BFO values from the MH371 flight to better understand drift, statistical, and systematic errors. Here’s my index once again:
https://docs.google.com/document/d/14hleZyx1pUPL44yaeHKt6jnSQ3DbgRq2zibbKkFLq2c/edit?pref=2&pli=1#
Scroll down to the link for “MH371 BTO/BFO Models”, which I have updated a bit. There are new figures 4 and 8, along with some explanatory text.
What was cleaned? Mainly I did a more careful job of eliminating portions of the flight during step climbs, of which there were several. I also did a more careful job of determining channel-dependent biases (actually, I do it as a function of frequency alone). For frequencies in common with MH370, I used the old values determined for that flight. I include R600, R1200, and T1200 values all together.
Figure 4 is probably the most informative. It gives the residuals in BFO values for signals that passed through the IOR satellite. I find a raw rms of 2.3 hz relative to a FFB determined from pre-takeoff and early post-takeoff data (replicating exactly what we can do for MH370). The peak-peak bounds are 12 hz, smaller than, but still consistent with, the JON range of 15 hz. If one makes a linear fit to the data, the slope is consistent with a change in zero-point of 3 hz from beginning to end (7 hr span). The “Monster of Mumbai” effect (i.e., Figure 5.4 of Bayesian Methods) is nowhere to be seen. In the latter, the peak-peak range is 37 hz, three times that of my calculations for MH371.
If I were asked to defend the DSTG’s assignment of 7 hz rms to to the BFO, I would fail miserably.
@Ge Rijn, Don Thompson. Reconciling the wreckage. IMO the immediate relevance of its character is to search prospects.
The wing wreckage generally looks less severely damaged that the rest. If the right wing in particular broke in mid-air close to the junction of its flaperon and outer flap, that would be a ready explanation for the collision between the two, as they separated in shock; and would help explain its quite large pieces of closing panels and flap mechanism fairings.
Of itself though it would not explain the somewhat similar left wing flotsam or the size of the engine fan right cowling piece.
A possibility in a mid-air break up is the the engines could be widely separated, though perhaps not the the main gear unless very high speed fuselage impact resulted in momentum-related dispersion.
I think some have had it in mind that the debris field would be compact, like AF447s but that may not be so.
However any wide dispersion of the major sonar targets will increase prospects of wreckage location, not the converse, IMO.
@DrB
“…The first possibility, that the FMT occurred west of circa 93.75E, can be refuted because, if one starts with the SLOP track after 18:28:15, then the farthest west one can get at the end of a FMT before 18:40 is only circa 94.2E even if a due west turn is made immediately after 18:28:15. So two turns after the SLOP to maximize westward travel still cannot touch the 19:41 Arc with a 180-degree track.”
Could you elaborate on how we can refute 180S paths (say at say BEDAX or ISBIX) due to SLOP? There is plenty of time to get there, right?
Further on wreckage assessment, some mumblings.
The missing appraisal of the Gibson/Madagascar engine vortex generator part, still apparently in Madagascar.
Malaysia appears disinterested. Its apparent passivity and lack of initiative in obtaining it, or access to it, or more information about it, is galling.
Also, the time for the Madagascan inquiry and, separately, its continuing retention of this item, without any progress report, appears sluggish if not worse. They must be aware of the circumstances. The French have at least kept Malaysia and Australia informed about the flaperon.
Even a preliminary assessment by Malaysia based on the Gibson photographs would have been be politic, even if not required by ICAO’s Annex 13.
Maybe OI can stir some stumps, in their own interests.
@sk999
Good stuff. We (you and I) still get a bit orthogonal on how statistics apply to BFO errors. I have given up on Exner and the IG. The notions of stationarity and ergodicity are truly of overwhelming importance. It is the very reason that Dr. Allan developed his variance. Oh well, given a set of numbers we are axiomatically driven to compute a mean and variance. These operations are well defined, and we all learned them in school, even if we fell asleep when and if stationarity and ergodicity were discussed.
@sk999,
Your analyses of MH371 data are both useful and illuminating. I was struck by the small RMS BFO residuals (2-3 Hz, not 5-7) you obtained. This result is consistent both with Inmarsat’s recommendation (7 Hz peak) and with the residuals for candidate paths near due south for MH370 that some of us have obtained. It is not compatible with the larger errors which are required for many of the extended loiter paths not fairly close to due south. Despite the DSTG’s analysis (which no one else can even attempt to duplicate because they have not released the raw data), I think DSTG substantially overestimated the typical single-flight BFO error dispersion. The small RMS value implies paths fairly close to due south.
@Don T
That statement may be true for an uncritical observer, however, the critical analyst must view a situation or scenario from many possible perspectives, no one of those perspectives may be complete but the composite will be more informative than only one. It can be challenging to understand that composite picture.
So I aam sitting outside with my evening martini (Plymouth gin, BTW) and looking at the full moon on the horizon. It is enormous. You people think you have an explanation for everything, but the “moon illusion” defies a creible explanation. Yet, you would have me believe that you understand the beginning of the universe, and it started with a “big bang”. Do you have any idea of how f’up you are?
@DrB said: I think DSTG substantially overestimated the typical single-flight BFO error dispersion.
I don’t know whether or not the DSTG overestimated the BFO error variation, but that would be quite a big mistake for them to make. The Inmarsat estimate of BFO error was based on just one flight (MH21 from KL to Amsterdam) for a different terminal than MH370. Similarly, @sk999’s results are based on a single flight (MH371 from Beijing to KL) for the same terminal as MH370. On the other hand, the DSTG’s results are based on 20 previous flights from the same terminal as MH370.
I agree that more northern paths would require drift in the BFO bias. If there was no drift, then the plane should be found in the 25,000 sq km area identified by the ATSB. For the assumption of small drift of the BFO bias, I am fond of 34.3S latitude because it fits a trajectory that is due south (180T), and the BFO error is minimized near this latitude. However, I recommend to OI a plan to continue the search further north if the first 25,000 sq km yields a null set. I think there are less than even odds that the search of the 25,000 sq km will be successful.
@Victor
The probablity of finding the aircraft in the 25,000 km area recommended by CSIRO is next to zero. The area around 30S is a much better likelihood. But hey, we have disagreed before, and I don’t have a pin in the map at 38S with name on it.
@VictorI,
You said: “The Inmarsat estimate of BFO error was based on just one flight (MH21 from KL to Amsterdam) . . . .”
Inmarsat built the satcom system and has operated it for years. They have unfettered access to all the BFO data for every terminal. I think it is highly unlikely (to put it mildly) that their estimate of BFO variation is based on a single flight. Rather, that flight was an example to illustrate their point. In my view Inmarsat are in the position of having the best possible understanding of this issue, and they knew full well that their reputation would not be helped if their analysis was flawed. I think you seriously underestimate their level of understanding of the BFOs. Their path prediction, as far as I can tell, can only be criticized by their lack of constraints to match FMS autopilot modes, and in this area they had no expertise. I think the aircraft will be found somewhat to the east of their prediction, but all in all, I give them great credit for a good effort that is in the ballpark.
@TBill,
You said:” Could you elaborate on how we can refute 180S paths (say at say BEDAX or ISBIX) due to SLOP? There is plenty of time to get there, right?”
What I said was, that with only one turn before 18:40, or even with two turns before 18:40, you can’t hit the 19:41 Arc with a 180 degree true track. To do that, you must have the final turn to 180 degrees after the 18:40 phone call. Yes, there is sufficient time after 18:40 to get far enough west to hit the 19:41 Arc at 180 degrees, but the final turn to 180 degrees must happen after about 18:45 at the earliest.
@DrB
You are simply ignorant. Gaussian stats applied to BFO errors are wrong.
@all
I am appalled by the “comprehension” of advanced degree people relative to BFO statistics. I have no idea where the disconnect occurred.
@DrB: By their own admission, Inmarsat only fully understood the BFOs weeks after the disappearance, independent of how many years they operated their system. The JON paper was released in Sept 2014. The DSTG study was released Dec 2015. I assure you that the DSTG study of the BFO was conducted in collaboration with Inmarsat. The DSTG presented the results of 20 flights. Inmarsat presented the results of just one. I suppose it’s possible that prior to Sept 2014, Inmarsat had compiled the BFO statistics for more than 20 flights. Considering the timeline, I seriously doubt it.
To be clear, I never said the Inmarsat analysis “was flawed”. Rather, it may be overly constrained. In the JON paper, Inmarsat’s approach in selecting a path is based on minimizing the BFO error. I know this based on interactions I’ve had with them. Reconstructions using this methodology will terminate near their end point. (My great circle path with the lowest BFO error falls almost exactly on their end point.) If there was no BFO bias drift, this approach is appropriate. However, the DSTG says there is the potential for significant bias drift, and therefore the BFO does not tightly constrain their reconstructed paths. I don’t know who is right. Rather than choose, I consider both possibilities, and I advocate a search strategy in which it doesn’t matter who is right.
If you believe that OI’s planned search area is incorrect or should be modified, that would be helpful input. To strongly advocate for an end point that falls in the 25,000 sq km is academic, as that area will surely be searched.
@VictorI,
You said: “I question the value in selecting one scenario among the many possible scenarios that result in a termination in the 25,000 sq km area. On the other hand, if you believe the search area is incorrect or should be modified, that would be helpful input.”
You seem to have a different goal in mind than I have.
My goal is to assist, if possible, in understanding what caused the loss of MH370. This is essential in minimizing the possibility of a similar occurrence in the future. Based on your comments, your “end game” appears to be limited to finding the aircraft debris field.
There are several things which can contribute to understanding the cause and therefore contribute to enhancing overall aviation safety, if appropriate actions are taken as a result of this new knowledge. The basic information sets we have, or hope to get, to meet this goal include the following (not necessarily in order of priority):
1. The published and unpublished factual information and investigative reports.
2. The final location of the aircraft (TBD).
3. Flight path analyses of aircraft and satcom data to determine how the aircraft was flown to its destination. This is my area of focus, not just to find the crash site, but also to understand as much as possible regarding the cause.
4. Recovered drifting aircraft debris.
5. Aircraft debris recovered from the sea floor, if possible.
6. FDR and CVR data, if possible.
7. Examination of recovered debris for evidence of pre-crash damage, including smoke and fire, especially for MEC equipment items and flight deck windshields, etc.
Others can probably add to this list.
Finding the aircraft debris field is #2 above, and knowing that location enables enhanced flight path analyses (#3) to be performed, as well as enabling #5/7 (debris recovery and analysis) and #6 (FDR/CVR). Obviously the FDR may provide sufficient data to render the computer flight path analyses redundant as to how the aircraft was flown, but what if the FDR is not found or is unreadable? Then we need the flight path analyses (#3) to determine the sequence of flight commands that led to the crash site, and these can help understand the motive and the intent of the PIC. If sufficient debris is recovered from the sea floor, then it may be possible to reach a definitive determination of whether or not equipment failure (#7) caused, or contributed to, the aircraft loss.
Predicting the crash site location prior to sea floor searches is certainly worthwhile, but highly detailed flight path modeling may add value beyond that. Consider this scenario. On a certain date, OI announces it has found the debris field at XX.X Lat, YY.Y Lon. What does that tell us about the cause of the crash? Very little, actually, except that we can now constrain a flight path model to end there and see what the intermediate path might look like. In particular, I think it will likely allow the identification of the lateral navigation mode. Depending on that result, there may be only one or perhaps a few options to use that lateral navigation mode, and this may suggest whether or not the SIO route was clearly intentional or perhaps may have been unintentional due to an incompletely executed flight strategy.
Once the crash site is located, it may take a considerable time to recover and analyze recorders and certain debris items. If the crash site location is made known publicly, then it is worthwhile in my opinion to continue the flight path modeling to understand how the aircraft was flown there. Suppose there are no definitive positive evidence for equipment failure and no FDR data. What caused the crash? Sure, people have a wide variety of opinions on that subject, but to me the cause would still be unknown with any reasonable degree of certainty, unless the flight path model itself suggested otherwise.
@DennisW said: The probablity of finding the aircraft in the 25,000 km area recommended by CSIRO is next to zero. The area around 30S is a much better likelihood.
Perhaps 30S is better. However, I would not put the probability at zero that it is in the 25,000 sq km area.
But hey, we have disagreed before, and I don’t have a pin in the map at 38S with name on it.
Nor do I have a pin near Christmas Island with my name on it.
DrB: No, finding the debris field is not my “end game”. But it does represent the start of solving the mystery with any level of confidence. You are extremely optimistic if you think you can solve this mystery without finding the debris field, and hopefully recover the data on the FDR. Even finding the debris field tells us little about the cause, as multiple scenarios produce the same impact point.
@Victor. You said, “However, I recommend to OI a plan to continue the search further north if the first 25,000 sq km yields a null set.”
Many of us would agree with you but I remind that gauged by their reports the ATSB/SSWG etc doesn’t. Participants of the First Principles Review meeting last December stated at page 23, “Based on the analysis to date, completion of this area would exhaust all prospective areas for the presence of MH370”, “this area” being the 25,000 sq km.
Developments since have been principally CSIRO reports 3 & 4, which are unlikely to change that assessment and the October, 2017 Operational Search Report makes no reference to a change in viewpoint.
A curiosity then is what encouragement the ATSB gave OI to search more than the 25,000 this at the recent London briefing.
@VictorI,
You said: “You are extremely optimistic if you think you can solve this mystery without finding the debris field, and hopefully recover the data on the FDR. Even finding the debris field tells us little about the cause, as multiple scenarios produce the same impact point.”
I never said I could “. . . solve this mystery without finding the debris field . . . .” Where did you get that notion?
What I said was that I think the crash location itself is likely to uniquely determine the lateral navigation mode, and this provides insight into the motive/intent of the PIC and may therefore (independently of other evidence) suggest a cause for the aircraft loss.
As an example, suppose one could demonstrate that the only way to reach the actual crash location was using LNAV with an Antarctic airport waypoint. Then I would say that it would appear likely that the PIC intended to crash the aircraft in the middle of nowhere and kill everyone on board, including himself. On the other hand, suppose the only way to get there is a Constant Magnetic Track entered using the MCP. In that case a direction, not a destination, was selected, and the motive is unclear.
I question your statement that, in general, an arbitrary impact point is reachable under autopilot control using multiple scenarios, each of which must, in my opinion, match BTO/BFO/endurance within Inmarsat (not DSTG) and fuel model tolerances.
Judging from the analysis on this website I would say the probability of finding the aircraft is zero for anywhere in the IO region.
@DennisW,
Is boutique gin distilling a ‘thing’ in CA? The market’s awash with them here: ShortCross, and Boatyard to name only two popular local producers. Just curious.
@DrB
Many thanks for the clarifications regarding your latest spreadsheet with the results from your OCXO Transient fitting program.
If the SLOP ended at waypoint IGOGU, I estimate the time at IGOGU as 18:38:56 UTC at a speed of Mach 0.84. This is just before the SATCOM Call(s) starting at 18:39:53 UTC.
In the attached chart, I show your SLOP connected to my Reconstructed Flight Path from waypoint IGOGU in LRC mode to an ultimate waypoint of 40S 100E. The MH370 End Point is at fuel exhaustion around 30°S near the 7th Arc.
https://www.dropbox.com/s/i6l38wjv65mwrxb/DrB%20SLOP%20plus%20Flight%20Path%20Model%20LRC%2040S%20100E%20173.png?dl=0
This scenario requires a loiter of 49.8 minutes.
This scenario would allow a possible fit to Kate Tee’s sighting at 6.6261°N 94.3872°E at 19:25:20 UTC.
@DrB: You are making this too complicated. I suggest that OI continue to the north if not found in the first 25,000 sq km because I believe the BFO values may allow it due to some drift of the BFO bias. You think the BFO bias drift is very small. I’ve asked the group for input as to where OI should look if not found in the first 25,000 sq km. You’ve provided none. That’s your option.
@David asked A curiosity then is what encouragement the ATSB gave OI to search more than the 25,000 this at the recent London briefing.
I suspect there was no discouragement from the ATSB to continue searching to the north if not found in the first 25,000 sq km, although they remain hopeful it will be found there. I think OI understands there is a reasonable chance it won’t be found in the first 25,000 sq km, and is prepared to continue. I think the question is only how far to the north OI will search.
@Victor. As previously stated, I propose the “beyond 38S” solution. My scenario for getting there is a CTH path at econ descent speed and then [for reasons unclear] flight more than 30NM beyond 7th arc. Having said this, I think the 180S CTT/LNAV terminus also has good prospects.
@Paul Smithson: Thank you for the input.
To be clear, do you advocate searching the first 25,000 sq km and then searching wider near 38S? Basically, given the option of searching either to the northeast along the arc or further from the arc, it appears your choice is further from the arc.
@Victor
If OI does not find MH370 between about 26 to 36S, then I would move up to flight Path L894…quite a few proposed paths e.g; Richard’s paper the Long Hunt for a Diversion Airport, Ed Baker’s path, etc. end up there
@TBill: Thanks. I suspect that the first 25,000 sq km could be completed within a month of searching, so there is some immediacy in planning the next phase. It will take months to get to 26S at +/-25 NM. We have time to think about what we would do next, if required, and if the option to continue the search exists. I suspect there would be robust debate as to whether to search further north, further out from the arc, or re-scan previous areas.
@Richard: I agree that a termination at 30S ticks a lot of boxes, and remains an area of high interest.
@Don T
Is boutique gin distilling a ‘thing’ in CA? The market’s awash with them here: ShortCross, and Boatyard to name only two popular local producers. Just curious.
No. Craft beer is a big deal here. The only local gin I am aware of in CA is Junipero made by Anchor Distilling (same company as steam beer) in San Francisco. So far Plymouth is the best gin I have found.
Still seeking the cure for vanceritis.
When the 777 Major Static Destructions Test was carried out on the aircraft wings, Boeing’s manager for the Major Static Test Lab related how “additional analysis after that [test] showed that the right wing failed beginning in the area of rib 25, and the left wing also failed at rib 25 and also at rib 21 on the upper surface of the wing. This was as they expected.”
The location of ribs 21 and 25 defines area of the torsion box, rear spar, where the outboard flap is attached, and the flaperon is inboard of that. Left wing view, right is obviously the mirror image.
Victor, Dr B., There is a difference in the analysis when it is directed towards defining the search area vs. understanding why the plane ended up there. For the moment it is evident that a specific search area is next, and that this is unlikely to change. In the event that the search is unsuccessful, then the assumptions that led to its selection would need to be revisited. Richard has provided the basis for doing that.
Why this event happened may depend on actually recovering the flight data recorder (and we have seen events where even this has not proved conclusive). So, trying to understand the implications of the data we do have is a worthwhile endeavor.
Sid
@Richard,
Your suggested connection from the SLOP to a holding pattern circa Ito a LNAV path is interesting. Based on some recent work in progress, I have concluded that it is fairly unlikely that the 18:40 BFOs were caused by an ongoing descent. However, that does not preclude an extended loiter from occurring after 18:41. If one occurred, I think it would be much farther east than VOCX, possibly in the vicinity of IGOGU/MEMAK. Using your path as an example, the aircraft might be seen when heading NE at 45 degrees on the racetrack to the left side of your proposed 225-degree leg.
Would you please post the data or a link to the data for this route (lat/lon/time/speed/flight level/track and BTO/BFO residuals)? Thanks.
@Don
As you may know, the B777 wing destructive test is on youtube. I recall watching it 20 years ago in the documentary series of the B777 programme.
https://www.youtube.com/watch?v=Ai2HmvAXcU0
@Richard,
Yes, I have watched that video on YouTube.
The transcript of the interview with Thomas A Briggs puts some useful detail to the video.
[Comments here are closed. Please continue the discussion under the new article.]