A fanciful depiction of a team of AUVs searching the sea floor as proposed by researchers at Virginia Tech
Despite an underwater search of the seabed in the Southern Indian Ocean (SIO) that covered 120,000 sq km, employed five search vessels, lasted 16 months, and cost Malaysia, Australia, and China a total of US$ 226 million, the MH370 wreckage remains elusive. The only debris from MH370 that has been found are parts that have drifted across the Indian Ocean and recovered from the shores of Eastern Africa.
Because of the need for specialized equipment capable of searching as deep as 6,000 m, operated by highly-trained crews in the exceptionally harsh conditions of the SIO, the underwater search is slow, expensive, and dangerous.
First, a bathymetric survey is undertaken to map the topography of the seabed. The bathymetric survey uses “multibeam sonar” transducers mounted on the hull of the survey vessel. By transmitting an acoustic pulse and measuring the time duration to receive an echo from the sea floor, the depth of the sea floor can be mapped with a resolution of about 100 m at a rate of 1200 sq km per day.
After the sea floor is mapped, the seabed is scanned for aircraft debris by pulling a “towfish” behind a search vessel. The towfish is lowered so that it “glides” just 100 – 150 m from the sea floor. The towfish is equipped with “side scan sonar” to search on either side of the towfish, and “multibeam sonar” to search below the towfish. This allows scanning the seabed out to a distance of 1 km to either side of the towfish at a resolution of about 70 cm and a rate of 133 sq km per day. If there is an object of interest, or the seabed is difficult to scan due to challenging topography, an autonomous underwater vehicle (referred to as an AUV or drone) can be deployed to get close to the sea floor and obtain high resolution images. For instance, the drones of the type used in the MH370 search have a resolution of about 10 cm and can scan the seabed at a rate of about 17 sq km per day.
Our estimates of the location of the crash site come mainly from two bodies of evidence: satellite data that was recorded for the brief intervals that MH370 transmitted signals to the Inmarsat communications network, and from drift models that estimate the crash site based on the timing and location of debris that has been recovered from the shores of Eastern Africa. Unfortunately, neither of these data sets is sufficiently precise to provide high confidence in the location of the wreckage.
Due to the expense of seabed searching, combined with the imprecision of using the existing data sets to estimate the location of the wreckage, some are suggesting that it is not economical to do further searching with our current technology. The argument is that further searching should be suspended until we gain additional information or insight that allows us to more precisely estimate the location, or until we develop new technology that allows us to more economically search large areas of the sea floor.
It is the promise of new technology that can more economically search large areas of the sea floor that led me to the work of Dr Dan Stilwell, a professor of electrical engineering at Virginia Tech. Dan’s team conducts research in the area of marine autonomy and robotics, and they have developed small, fast, high-performance, inexpensive AUVs for the US Navy. His research team is using their extensive inventory of technology to compete in the Ocean Discovery XPrize, which aims to accelerate innovations to improve the speed, scale, and image resolution of technologies used to explore the ocean floor.
Dan Stilwell (right) and his team with one of their AUVs
The XPrize contest will require mapping 500 sq km of ocean floor with a resolution of 5 m and at a depth of 4,000 m, and also to produce high resolution photographs of various objects on the seabed, all within 24 hours. That’s quite a challenge with existing technology. Nonetheless, the prospect of winning the US$ 7 million prize has attracted interest from 21 teams from around the world.
Dan’s approach is to use a “team” of small, low-cost AUVs to cooperatively survey and scan the ocean floor. Each AUV can travel at 4 knots for 24 hours on a single battery charge. Rather than using expensive inertial guidance systems to navigate, Dan and his team are using technology developed at the Woods Hole Oceanographic Institute, whereby all the AUVs acoustically communicate and navigate using a low-bandwidth, time-division multiple access (TDMA) network. In this approach, each AUV in the team is assigned a time slice and each AUV has a synchronized clock. Each AUV measures the time delay between transmission and receipt of pulses from each of the other AUVs, and from this and other information, the relative position of all the AUVs may be determined. One node remains at the surface, which provides an absolute GPS position reference. This approach to acoustic navigation has not previously achieved the accuracy that is required for the XPrize contest, but Dan’s team will implement a number of new tricks that they expect will provide a sufficient boost in navigational performance.
Can next generation AUV technology provide an economical way to search for MH370? Consider this: Dan estimates that it would take about four of his drones to match the scan rate of a single towfish. But there are compelling economic benefits to using a team of drones. For one, each drone is relatively inexpensive–Dan and his team can build one for about US$ 125,000. Secondly, a large team of AUVs can be deployed from a single surface vessel and crew, while a towfish requires a dedicated vessel. For instance, if a cooperative team of twelve AUVs is deployed from a single vessel, that vessel would be able to scan three times as much sea floor as a vessel deploying a towfish. As sonar sensors increase in performance and miniaturization, sea floor scanning with AUVs will become even faster and cheaper.
As we struggle to squeeze every last bit of information from the existing MH370 evidence, it may be that some of our resources are better directed to improving our ability to quickly and economically search large expanses of the sea floor. Research on autonomous vehicles like that performed at Virginia Tech by Dan and his team can help us.
Update on May 31, 2017.
I was recently in a discussion that included a well-known ocean explorer who happens to be a judge in the Ocean Discovery XPrize competition. We were having a general discussion about searching for MH370 and ways to scan the ocean floor at high resolution, and he told us about the capabilities of Ocean Infinity. Like the team at Virginia Tech, their approach is to employ a team of AUVs. From their website:
Six HUGIN autonomous underwater vehicles (AUVs) are capable of operating in 6,000 m water depth collecting high resolution data at record breaking speeds. Our AUV fleet is accompanied by six unmanned surface vehicles (USVs) to ensure precise position and constant communication.
With multiple autonomous vehicles working simultaneously utilizing innovative technology, we are able to survey huge swaths of the seabed, quickly and with outstanding accuracy. We can operate in shallow waters but excel in extreme depths, working in dynamic environments ranging from the tropics to the Arctic ice.
Because of the size and complexity of each AUV/USV pair, the capital cost of the technology from Ocean Infinity would greatly exceed the capital cost of Virginia Tech’s technology, which uses small AUVs with innovative navigation systems. On the other hand, both approaches benefit from having a single host vessel supporting multiple underwater vehicles, which offers significant operating cost and scan rate improvements compared to the conventional towfish technology.
Ocean Infinity’s seabed exploration system is commercially available today, including underwater and surface vehicles, on-board support equipment, and the host vessel. This is an exciting possibility for conducting the search for MH370 in the near future.
@Victor Iannello:
Thanks for opening a new thread. It loads in no time at all!
@All
If and when the comments on this thread get to long on this thread because your browser takes to back to the top of page, press the End key on your keypad or keyboard, to go the bottom of page to the recent comments.
@Gysbreght: Try reloading the previous thread and see if it loads faster now.
@Victor Iannello: It’s faster now – 15 seconds – but that has always been variable, like the weather.
@Joseph Coleman
Joseph, the key equation for the BTO is:
Range(satellite to aircraft) =
c.(BTO−bias) / 2 −Range(satellite to Perth GES)
Somehow you should relate your statements to this equation..
For example the aircraft, even at constant track, is not travelling in a straight line in 3D space (follows the curved earth surface). So it looks to me you found an interesting coincidence. But I still don’t get how
“Equal proportions in percentage over time (between 19:41 and 00:1929 if 18400 BTO is used at 00:1929) showing Sat movement and Plane movement (near constant speed and near constant direction each individually) the BTO separating the range from each other”
@VictorI
Interesting new article and food for thought in different respects. I agree there seems to be a limit reached in what can be squeezed out from all available evidence. However, personally I’m in doubt all relevant evidence has been shared with investigators (including the official team). For example, I still find it hard to believe that Thailand and Indonesia wouldn’t have seen MH370 by radar. And perhaps there are other sources (radar / satellite) which have not been disclosed for geopolitical / security or whatever other reasons. As scientists / engineers we are trained and have the ambition to look for solutions. The disappointing reality, also in the case of MH370, is that in politics different rules apply and different games are played. So the question is if the key in understanding the MH370 tragedy is eventually found through technics or through politics.
@Niels
“Interesting new article and food for thought in different respects.”
It is interesting. Acoustic ranging/navigation has been tried by a lot of different folks including me and people in my past life. No one has ever had any success relative to producing a marketable product. Only bats seem to be able to make it work.
Sitting on my porch years ago I was watching a moth fluttering by the porch light. At that moment a bat came streaking into the porch “tunnel” (porch is enclosed overhead with access at each end), pick off the moth and continue out the tunnel – amazing. Bats use a chirped acoustic emission, and seem to be able to map not only objects in their domain (including Doppler compensation for their own motion), but also pick out and snag insects as well. The amount of digital signal processing required for us to do something like that is staggering. Yet the bat has a brain the size of a small marble.
All I could do was take another sip of scotch, and reflect on the fact that we don’t know jack shit.
@Niels: I agree that data has not been fully disclosed. As you know, I have tried many different avenues to get more released.
@DennisW: Yes, acoustic navigation by bats is amazing. Incredible what a small, biological neural network can do.
@DrB – In your paper you said you received fuel consumption history for a few of 9M-MRO’s preceding flights. Is that something you can share with us? It would be interesting to learn what the real PDA’s were. Does this consumption support about 2700 NM (air) after the 18:22 position?
@DennisW
That is indeed fascinating. We also have bats behind the house, and I enjoy following them in the rather long summer twilight that we have here. The scotch is a good suggestion, also in the light of the other thoughts.
The model here would seem to be a school of porpoises or whales, not a bat — different both in time resolution [you don’t need to put all the hardware and software i a 1-gram package, and in open water, you aren’t dodging branches and chasing fast moving objects in 3D.] it seems that the kind of drones proposed would not be mapping by sonar+inertial input over long distances, but need only to locate [or be located by] and stay in communication with a surface [or shallow-submerged tender], and with the rest of the “school” — right? My guess is that optimally, the school would stay in formation with the tender, all hovering if one of the school finds a feature reuiring a closer look, and surface in usison for recharging.
All likely to attract attention of a James Cameron or Elon Musk in a few years.
Thanks for the new thread Victor.
People today are having some relavent success with smallsat constellations that have at their roots the pioneering efforts of the AMSAT community back in the 70’s. Basically, the concept is to replace a small number of large, expensive spacecraft (and payloads) with a large number of much smaller spacecraft that can be manufactured on a commercial scale at far lower cost. Using the same concept, adapted to a “fleet” of AUV’s, you could simultaneously scan a width of, say, 40 nm using a hundred drones. Deployment/recovery and battery recharging could get interesting.
For those working on developing fuel models, I found the following document to be informative:
http://media.aspeqexams.com/ATPL_Flight_Planning_Data_Booklet.pdf
It covers the performance of a 777-300ER, presumably with GE90-115B engines, and thus cannot be used directly with Trent 892 engines; however, the document includes fuel burn tables for LRC, HOLD, and Cost Index 50 modes, trip fuel tables for LRC and CI50, plus a table showing Mach as a function of cost index and weight. If CI=0 is truly Maximum Range Cruise, we can evaluate the universality of the AERO chart showing mileage v. Mach. It turns out that the function is not universal, but becomes more symmetric as the difference between LRC and HOLD Machs becomes large. I came up with a somewhat ugly but seems to work algorithm for rescaling the function, and, in dimensionless form, I assume that it applies to the Trent engines as well.
Could it be an opportunity to execute this contest in a high probability search area around the 7th arc?
I mean when 21 teams each cover 500km/2 they together would cover 10.500km/2 in one day at a resolution of 5m.
Probably a resolution good enough to find wreckage.
@ALSM: re: my latest attempt to stochastically model end-of-flight: thanks for your encouragement.
You can assist me, actually. I repost a request you may have missed on the prior thread re: MH370 end of flight sims you ran in early 2015:
1) exactly how many trials did you run?
2) of these, how many happened to exhibit descent rates within what the ATSB set in its Nov. 2016 report as hard bounds at each of 00:19:29 and 00:19:37 (which for purposes of your sims we should describe as “2nd engine flameout plus, respectively, 120 and 128 seconds”)?
3) for only those trials which DID satisfy those bounds, how far from Arc7 (FL0) was impact?
Alternatively, I would be thrilled to accept a raw(ish) data dump of your sim runs, and rummage through them myself. Whichever you prefer.
Profuse thanks in advance for your time and attention.
@sk999:
I’ve compared the Cost Index in the ATPL booklet to that derived from AERO Figure 1.
We don’t know the conditions for the Aero figure, but the fuel flow of 3641 kg/hr at M.84 and FL350 would correspond to a weight of 250 tonnes in the ATPL booklet.
On the subject of Cost Index, I wondered why MAS specified CI=42 rather than a round 40 or 50.
Well, increasing the CI from 40 to 42 increases the Mach number by 0.0001, and that brings a theoretical reduction of total operating cost of (rounded up)0.0000001 %. Good for the bottom line!
Brock:
We had 4 hours in the simulator…not a lot of time to conduct trials. We did a number of different experiments first, followed by 3 that were specifically aimed at testing the plane’s response to fuel exhaustion at 35,000 feet, AP & AT engaged, on a 186 magnetic heading at ~480 kts. Unlike some of the tests conducted by others, where fuel exhaustion was simulated by shutting off the fuel, we simulated the event by setting up the initial conditions with 1000 lbs in one tank and 200 lbs in the other (or in one case, 200 lbs of fuel in both tanks), then sitting back and watching. In all three cases, a spiral descent started developing within a minute or two. They all ended quickly with speeds that pegged the IAS (500 kts max indicated) at times, steep turns, phugoids, descent rates up to 29,000 ft/sec. In one sim, the plane made a total of 3 full 360 turns, the last in 25 seconds. We did not attempt to extract a path, but that might be possible from the available data. I’ll take a look. But you can almost do it in your head. The radius of a 10-15 deg/sec turn is small, so the POI was close to the point of fuel exhaustion, and could have been on either side of the 7th arc.
This last point has not received all the attention it should. With all the attention now looking at points between S30 and S35, I went back and looked at the GO PHOENIX data set from Sept 2, 2014 to June 20, 2015. It turns out Phoenix scanned from about S32.8 to S34.8, but the scan width was only 22 nm, with 2 nm inside the surface 7th arc, and 20 nm outside the surface 7th arc. Several of us (including ATSB) have looked at the speed and distance between the 6th and 7th arcs and concluded with high confidence the plane was already in a left turn at 00:19:29. With available data and sim results, it is almost equally probable that the plane is inside or outside the 7th arc.
@Victor
It’s a pity, with a little more foresight, the ATSB didn’t try to coopt/invite the teams to carry out their round 2 tests (Oct 2018, up to a depth of 4000 metres) in the new SIO north zone identified earlier this year. There must be areas of the new search zone that don’t go to 6000 metres. Perhaps an additional bounty prize of $2m dollars donated by Shell/the tripartite governments for any of the teams detecting seabed aircraft debris.
Your thoughts?
As far as the Underwater Drones topic, in the future, one way to find MH370 would be to form a joint industry R&D program to develop technology and possibly work together to search for aircraft. I say this partially because I used to be chairman of a joint industry R&D organization, and I know how effective they can potentially be. MH370 could perhaps be the impetus for forming such a group, especially if the R&D time frame was long enough to allow a legal comfort zone as far as statute of limitations for legal actions if the cause of MH370 is ever actually discovered.
@ALSM
“It turns out Phoenix scanned from about S32.8 to S34.8”
I am working up a proposed MH370 flight path ISBIX 180S CTH or possibly BEDAX 180S CTH whereas MH370 finally goes down (with winds) 94.2E such that S32.5 to S33.5 is my preliminary crash zone estimate.
Does the Go Phoenix search area show in Richard Cole’s map of area searched? Or do we need to add that? I will look myself when I get a chance, but just asking.
@Mick Rooney: The logistics of organizing a structured competition in a part of the world as remote and harsh as the SIO would be challenging. Compounding this is we really can’t be sure where to search. I think the approach of the Ocean Discovery XPrize of testing new technologies in a controlled environment would be more productive at this time.
I hope the XPrize competition will shine the light on promising new technologies that might be used to search for MH370. That was really the objective of my post. I don’t rule out the prospect of a related “free form” competition that awards a prize to the first team that identifies debris from MH370 on the sea floor. (Malaysian officials this past January talked about offering a reward to any private group that finds the fuselage of MH370. When pressed for details, they backed off their offer.) Using the existing evidence to decide where to search might be part of the competition. For instance, a team that believes they can estimate the crash site within a small area might not need cutting-edge scanning technology. Another team might choose to search a larger area with more advanced technology.
Coincidentally, the XPrize was founded by Dr Peter Diamandis, who is a friend from our days at MIT. Peter would be a good person to get ideas about how to organize a competition for finding MH370.
@TBill: I’d be surprised if there are not qualified teams that form to find the plane. Some might be partnerships between industry, research institutions, and private citizens. Finding philanthropists to fund a team would be challenging, but I think it’s possible if the team has the proper qualifications and offers a realistic approach. I have not seen any public announcements from any credible team.
@Gysbreght,
The ECON Cost Index in the Flight Brief is 52 (not 40).
TBill:
My GO Phoenix data covered 2014-09-02 up to 2015-04-16. Richard Cole’s data only covered 2015-04-07 to 2015-06-20 (11 days of overlap that matches). We know Phoenix stopped searching on 2015-06-20, so this should be a complete record of the Phoenix tracks. Of course, it is unknown if the fish was deployed and operating during 100% of the time they were making parallel tracks, but it looks like they covered the circumscribed area well. In the following GE image, my data is in yellow and Richard’s data is in red. these tracks are available as .kmz files if interested.
https://goo.gl/HT77fR
@VictorI
The SIO is not always that harsh throughout the year. It would be a challenge; nothing lost nothing gained for those teams only want to prove their systems work or not. A tremendous bonus would be if they find MH370 wreckage. Above ~32S the sea floor is above 4km.
It would be a perfect enviroment to test their equipment devided over 500km/2 for each team. Covering 10.500km/2.
I think it would be a very good initative to perform the challenge/contest right there along the 7th arc.
I’m sure the NoK and everyone else would support this.
I think this is an opportunity this contest cannot affort to slip away.
Great publicity anyway to profilate themselfs.
I add a picture of the SIO around the latitudes concerned..:
http://static.panoramio.com/photos/large/37312554.jpg
@ALSM: thanks for the reply, and for offering to dig up the data on those sims.
I agree the Go Phoenix search was asymmetrical with respect to Arc 7 at FL(0). However, it was materially less skewed with respect to Arc 7 at FL(350). Determining just how far inside of Arc 7 impact could have occurred in theory is a key goal of my integrated stochastic modelling.
In the limit where Arc 7 intersection occurs at precisely FL(0), maximum distance inside this arc is precisely zero, because intersection occurs at the precise moment of impact. It is not clear from anyone’s work – yours or the ATSB’s – just how far inside this Arc a plane could impact GIVEN descent at 00:19 was at BFO-consistent rates.
If you graciously publish your detailed sim data, not only will it allow me (and others, I hope) to analyze this critical question, it may help shame the ATSB/Boeing into similar publication of their own sim trials – something I believe we all want quite badly.
As may of us (including our esteemed host) have pointed out, 33-35S is already strongly counter-indicated by the empty Oz shorelines, “Roy”, and other evidence – it would be a shame to launch another multi-million dollar search further inside Arc 7 if the sim data could have shown us before the fact that it was pointless. I am keener than ever to develop this integrated model, to help sort this out.
@Gysbreght,
Thank you for providing additional comments on the possibility of developing a generalized fuel model. The question in my mind is how much the shape of the FF versus speed curve varies with weight and flight level. If not drastically, then perhaps an approximate equation can still be useful, if not exact.
I have several questions:
1. I note that you are connecting the LRC and Holding points on your TSFC plot with straight lines. Why straight? Is this an assumption equivalent to assuming a particular interpolation equation for fuel flow versus speed?
2. If you have previously defined the parameters used in your TSFC plot, such as delta, theta (temp?), etc., would you please just list the date and time so we non-aerodynamicists can look up what they are. If not previously referenced here, then please explain them to us ordinary folks or give us a reference that does that.
3. It would be helpful if you would give one numerical example (perhaps using LRC at W = 240 tonnes and FL350) and show how the abscissa and ordinate values in your TSFC plot are calculated from the Boeing LRC tables for Mach and Fuel Flow. Are you assuming anything else not in those tables?
@ALSM: what do you have as the precise longitude of Arc 7 (FL(0)) at each of S32.8 and S34.8S? Just syncing up. Thanks.
@DrB: “The ECON Cost Index in the Flight Brief is 52 (not 40).”
Why 52 (not 50) ?
@DrB: RE YR Questions at 2:37 pm today:
1. As stated when I first posted that chart, I connected two points with the same Mach number on the LRC line and the Holding line by straight lines. Actually both point lie on a curve in the generalized plot of TSFC=f(thrust, Mach) for which I provided an example for the PW2040 engine at FL350. Since we don’t have that chart for the Trent 892 engine you just have to imagine the curvature of those line segments or take the straight line as first order approximation.
2. I’m quite sure I have defined delta and theta earlier, but here they are again:
Delta = ambient pressure ratio, i.e. the ratio between the static pressure at altitude to the standard sealevel pressure.
Theta = ambient temperature ratio, i.e. the ratio of absolute static temperature at altitude to standard sealevel absolute temperature.
If you want a reference I would recommend Boeing’s “Jet Transport Performance Methods” but there must be many others.
3. The FCOM Long Range Cruise Control table for W = 240 tonnes, FL350 gives:
– M.84 and FF/ENG 3645 kg/hr.
– W, FL and M define the lift coefficient cL = 0.46714
– Professor Obert’s chart then gives the drag coefficient cD = 0.024078
– The total thrust equal to the drag is then W*cD/cL = 12,370 kg
– The thrust-specific fuel consumption TSFC is then total FF divided by drag.
I’ve not assumed anything else.
@ALSM: quick observations on your linked graphic:
1) you appear to overlay a square-coord map (Richard’s, and thus presumably yours) onto a GE backdrop. While reasonable in general, the stat you quoted – “amount searched inside ground level Arc 7” – appears to be sensitive to this approximation error. Arc 7 FL(0) is nearest the outermost GP track at the southern edge of its zone, but nearest the 4th track in if we take our measurement nearer the zone’s centre. (This is why I asked you above for benchmark coordinates for your Arc 7’s: I have the search tracks at very good fidelity, and wish to place Arcs just as accurately.)
2) extremely minor: coverage actually extends roughly a km beyond the tracks, due to towfish lateral range.
3) at the southern edge of GP’s zone, Fugro Equator and Discovery later widened the area scanned – roughly 16 tracks inside, and 4 tracks outside. The tracks are spaced for towfish scanning (~1nm), not bathy surveying. This changes the total width scanned from 22 to 42 – and the inside/outside Arc7 FL(0) split you quoted above from 2/20 to 18/24. After adding towfish scan width at the margins: 18.5/24.5. After redefining Arc7 to a more plausible intersection altitude: 21/23. This additional non-GP coverage does bleed down rapidly as you head NE along the GP zone, but it is coverage we are unwise to ignore.
@Gysbreght,
Thank you for the additional information. That is helpful to my understanding what you have done.
As for why MAS used CI = 52, it’s only a guess but I assume they have someone whose job is to optimize overall profit by occasionally adjusting CI, especially when fuel costs change noticeably. They may even use a formula with various costs as inputs, and this could lead to a specific numerical result, such as 52.
@Lauren H,
You said: “@DrB – In your paper you said you received fuel consumption history for a few of 9M-MRO’s preceding flights. Is that something you can share with us? It would be interesting to learn what the real PDA’s were. Does this consumption support about 2700 NM (air) after the 18:22 position?”
1. I was provided average Fuel Flow values during cruise from previous flights. I cannot release these because of a Non-Disclosure Agreement I have with ATSB. However, I am now of the opinion that at least the right engine fuel flow sensor values, if not both, have sensor errors between 1-4%.
2. I trust the MAS average PDA value of 1.5% from the flight brief more than the PDAs I estimated from the ATSB Fuel Flow sensor values (which are both higher than +1.5%).
3. At MRC from 18:36 onward at FL350, fuel exhaustion is 5 minutes early. At LRC and FL350 it is 13 minutes early. At Best Holding at FL350 it is 21 minutes late (5%). You can figure the ESAD’s.
4. Obviously, you can match the endurance if you fly less than half the time at Best Holding and more than half the time at ECON.
5. Alternatively, a fixed KIAS at an inefficient altitude might also match the endurance. That’s why I am trying to come up with a fuel model that can estimate fuel flow as a function of IAS, weight, and flight level.
6. One possibility could be ~250 KIAS at an inefficient altitude – either very high at ~FL380 or low at ~FL150. The idea is to identify combinations of Flight Level and airspeed that produce zero endurance error. I think it is correct to say that such scenarios exist that are consistent with the observed fuel burn, but then one also has to demonstrate consistency with an available navigation mode and BTO/BFO, and this has not yet been done.
@DrB: The real point of the exercise is to illustrate that the FCOM LRC and Holding tables do not contain the information you need to calculate fuel flow at speeds other than LRC and Holding. Modelling, curve-fitting or “physical equations” cannot fill that void.
@Gysbreght,
You said: “@DrB: The real point of the exercise is to illustrate that the FCOM LRC and Holding tables do not contain the information you need to calculate fuel flow at speeds other than LRC and Holding. Modelling, curve-fitting or “physical equations” cannot fill that void.”
If this is true, then on what basis did you connect the LRC and Holding points in your graph with straight lines?
Besides, we do have some additional information – namely Aero Figure 1. That plot allows us to interpolate between LRC and Holding for one combination of weight and altitude. My question is, how representative is the “shape” of that curve to other weight/altitude combinations. This will only be an approximation, but it is better than nothing and it is the best we have at the moment. In addition, we also know the FFs are monotonic and they do pass through the LRC and the Holding points. So we are not without useful information to fill in the void, albeit with much less accuracy than the Boeing tables.
@sk999,
The booklet you posted is helpful in that it allows one to check the ECON model equations for FF and Mach as functions of CI, as you have done. I have always wondered how large the errors are from assuming the CI equations derived from the Aero Figure 1 curve also apply to other weights and altitudes. Now we have the ability to do an assessment of those errors.
As you (and then Gysbreght) have demonstrated, there are noticeable differences on the order of 1%. I’ll also have a go at modeling the variation with altitude and weight, making the assumption that the general trends will apply to 9M-MRO. If I can obtain a fairly simple result, I’ll post it and you can compare my result with yours.
@DrB said, “This will only be an approximation, but it is better than nothing and it is the best we have at the moment.”
I think we already have better models. The semi-empirical model I developed uses Obert’s drag curves combined with an engine model that includes the effect of ram drag, parasitic losses, and temperature effects. It does not use Aero Figure 1 in the formulation, yet matches it quite well, as you can see in this figure.
No, I have not yet documented the model. If there is interest, I will at some time in the future.
@DrB: “on what basis did you connect the LRC and Holding points in your graph with straight lines?”
As I explained yesterday: none whatsoever.
“My question is, how representative is the “shape” of that curve to other weight/altitude combinations.”
Remember this chart?
@Victor Iannello: I would be interested in your engine model and the considerations on which it is based.
All,
At the top of my index file here:
https://docs.google.com/document/d/14hleZyx1pUPL44yaeHKt6jnSQ3DbgRq2zibbKkFLq2c/edit?pref=2&pli=1
I have links to two spreadsheets. One is a set of tables for fuel burn as a function of mass, altitude, and mach, and the other is my prediction for MRC based on scaling from the ATPL tables. The methodology is explained briefly. No physics. No guarantee that they are correct. However, they are in a form that can be used for comparison by others.
@Gysbreght,
No, I have not seen that chart before. If I understand it correctly, the curve for the A330 is a different shape than the Aero curve for the B777. That doesn’t surprise me, and it also doesn’t answer my question, which is how much the B777 curve changes with weight and altitude.
The “parabolic” curve is not explained, so I don’t know what to make of it. Looks like a simple model, and it does not match either aircraft.
@VictorI,
The match between your model, after scaling, to the Aero curve is impressive and reassuring.
Obviously one could simply start with the Aero curve with its single set of conditions, and try scaling it to other MRCs by simple multipliers applied to Fuel Mileage and to Mach. My question was how well that resulting shape works for different weights and altitudes. It seems you have a model which can answer that question. What say you?
@all
I really struggle reading through the fuel consumption discussions, and to a lesser extent the aircraft aerodynamic issues. We cannot forget that Boeing was a member of the SSWG and participated in the selection of the high priority search area. Does it make any sense at all to second guess the manufacturer of the aircraft? Do you think Boeing is ignorant or was non-cooperative in the search planning?
@DrB: We discussed that chart in an email exchange back in July 2016. The Airbus curve is for an A330-300 with GE engines, 180,000 kg, FL310, ISA. The curve for the A330 is a different shape than the Aero curve for the B777 because it is at a lower weight and altitude. The B777 curve shape is likely to change similarly at lower weight and altitude, where the Mach effect is less pronounced.
The “parabolic” curve is is what you get without Mach drag rise, drag coefficient linear with lift coefficient squared, and constant thrust specific fuel consumption.
The red line in this chart shows the MRC Mach number (CI=0) from the ATPL booklet in comparison to LRC and Holding versus Weight/delta (Weight corrected to sealevel pressure).
Interesting to note in above chart that the ratio between LRC and MRC speeds approaches the “parabolic” value of 1.0877 at weight/altitude 20% below optimum weight/altitude (approximately W/delta = 820 and 1,020 tons, respectively).
@DennisW: “Do you think Boeing is ignorant or was non-cooperative in the search planning?”
I think it is reasonable to assume that Boeing’s participation consists of providing the information they are asked to provide. They will not volunteer information or express an opinion unless the investigation goes in a direction that could be contrary to Boeing’s interests.
I think resuming there are two main questions left that dispite all calculations and other efforts are still left unanswered with definite proof.
1; Was the flight controlled by a pilot till the and or a ghost-flight (after FMT)?
2; Was it a high speed impact in the water or a low speed ditch-like event?
I think as long as those two questions can not be answered with conclusive proof first, there is no use in further calculations based on assumptions.
According the ATSB-report the plane also stayed airborne in several simulations for 20 minutes. If this is what MH370 did it could be 100 miles away from the 7th arc. Controlled or uncontrolled. No one knows.
The debris tells a conflicting story regarding the final BFO’s I’m still sure. It does not show the signs of a high speed impact but clear signs of a ditch-like event. Final 100% proof one way or another is still not given by the ATSB or anyone else.
These basic questions have to be answered with proof first now IMO.
Otherwise we stay on a road to nowhere I’m afraid.
@DennisW: You asked a good question that I have asked myself. I also agree with @Gysbreght’s response. Boeing has the capability to do the fuel calculations with more accuracy than we can ever hope to achieve. That doesn’t mean their capabilities have been fully utilized.
When I first investigated the fuel consumption by assuming a flight at LRC and FL350, I discovered that the fuel consumption of MH370 was about 1.8% higher than the LRC fuel flow table suggests. I attributed this to a combination of PDA and temperature effects, and was not too concerned. However, @DrB admirably pressed this point, reasoning that for the temperatures expected during the flight, the PDAs would have to be unrealistically low. We now know that the average PDA for the engines for fuel planning was about 1.5% based on what is probably a fairly sophisticated model used by MAS. We only know about the value for the PDA because of the leaked RMP report. Yet, we have been asking the ATSB and Malaysia for the PDA for 3 years!
The bottom line is that the plane likely did not fly at LRC speeds for the duration of the flight. This is another blow to the 37.5S latitude implied by the DSTG study, and pushes the probable crash site further north. But we already knew this based on the failed search. Why didn’t the DSTG use Boeing’s assistance to incorporate an accurate fuel flow model to help refine the search area? We can only guess.
If the plane flew at holding speed and at FL200 after 18:28, the fuel endurance would be about an hour longer than observed. Therefore, there are many possibilities of altitude and speed that result in the observed fuel exhaustion at 00:17. In my opinion, this is consistent with a possible descent at 18:40, which has been discounted by the ATSB.
I think there is value in developing better fuel models because it helps to eliminate certain paths. Unfortunately, I think most of those paths have already been eliminated by the failed search.
@DrB asked, “My question was how well that resulting shape works for different weights and altitudes. It seems you have a model which can answer that question. What say you?”
Good question. I’ll post some plots later that could shed some light on this, first at FL350 at various weights.
@Ge Rijn
I do not know if the the pilot was alive at the end of the flight. But I do think it is very possible the pilot was alive after FMT to set a CTH True Heading course south and change speed between Arc2 and Arc3. The Inmarsat Arcs are obviously (based on many proposed flight paths including for example DrB’s) consistent with True Heading southerly with some wind effects.
@TBill
That’s the problem. I don’t know either and nobody does. Logically and statistically speaking the plane must have been pilot controlled after FMT too, probably till the end.
No reason or proof at all that it didn’t. But nobody knows for sure.
Deliberate peed changes between arcs etc.; if it only could be proven..
“peed” ?
@Gysbreght
‘Peed’; dictionary definition; an act of urination.
What do you mean with it?
@DrB
Re your earlier comment: “at MRC from 18:36 onward at FL350, fuel exhaustion is 5 minutes early. At LRC and FL350 it is 13 minutes early. At Best Holding at FL350 it is 21 minutes late (5%). You can figure the ESAD’s” – I agree with Dennis, the subject of fuel consumption can be a minefield for the unwary (I include myself as one of the unwary) but as I understand it from the online Boeing article, Fuel Conservation Strategies, LRC is flown at a higher speed than MRC, and fuel would thus be used at a correspondingly higher rate, so how come LRC lasts 8 minutes longer that MRC? If I have misunderstood or misinterpreted your comment, I apologise right now, and will exit quietly stage left with a suitably red face.
I find it interesting that (from 18:36) the aircraft appears to have flown at or very close to a speed that maximizes the range. This suggests to me at any rate, a pilot deliberately wanting to maximize the range following the FMT.
@Victor
You said “The bottom line is that the plane likely did not fly at LRC speeds for the duration of the flight. This is another blow to the 37.5S latitude implied by the DSTG study, and pushes the probable crash site further north. But we already knew this based on the failed search. Why didn’t the DSTG use Boeing’s assistance to incorporate an accurate fuel flow model to help refine the search area? We can only guess” – I would respectfully point out another possible reason that the search failed, and that is the aircraft was glided to an impact point outside the search area southern boundary. Yes I know the drift studies seem to say otherwise, but I still think the CSIRO drift study results are suspect, and are designed to back up the S35 area, and bail the ATSB out of its predicament.
Something else.
While the subject has changed, still also the obvious conflicting CSIRO conclusions on their own latest drift-data remain unresolved.
IMO they seem so obvious confirmation-biased towards the latest ATSB high probability area which is still in their ‘pre-debris’ area based on Inmarsat and conflicting the CSIRO drift-data.
It gives the impression to me all involved parties are avoiding responsibility and being called at their 200 million dollar failure for maybe heads will roll.
@Rob
You put the words right out of my mouth before I read your comment.. 😉
@ROB,
You said: “. . . but as I understand it from the online Boeing article, Fuel Conservation Strategies, LRC is flown at a higher speed than MRC, and fuel would thus be used at a correspondingly higher rate, so how come LRC lasts 8 minutes longer that MRC? If I have misunderstood or misinterpreted your comment, I apologise right now, and will exit quietly stage left with a suitably red face.”
Indeed you have misunderstood me. I said at MRC, fuel exhaustion is 5 minutes early. Early means before 00:17 (when MEFE occurred). LRC fuel exhaustion occurs 13 minutes earlier than 00:17, or 8 minutes before it occurs at MRC. As you correctly pointed out, LRC has higher fuel flows so its endurance is shorter than MRC (as I indicated). Holding fuel exhaustion occurs 21 minutes late (after 00:17). Best Holding has the longest endurance (by definition).
There is no need to apologize. We all misunderstand occasionally.
@ROB,
You said: “I still think the CSIRO drift study results are suspect, and are designed to back up the S35 area, and bail the ATSB out of its predicament.”
Based on a few communications with Dr. Griffin, in my opinion the CSIRO studies are not “designed to back up the S35 area.” They may be wrong, but I think they are CSIRO’s best and unbiased effort.
It is important to keep in mind that the CSIRO study uses data not incorporated in any other drift study. They used satellite measurements of ocean surface height to estimate the near-surface current flow patterns. Surprisingly, the scale for directional changes is as small as 50 NM. Near 35S this has a significant effect on the probability of WA landings. Again, I’m not saying their conclusions are correct, just that their results will vary somewhat from other drift studies because of their different treatment of surface currents.
@VictorI,
Thanks Victor for using your new fuel model to address my question.
By the way, I have some new Boeing FF/Mach data I will post soon that covers a wider range of weights and altitudes, and has closer spacing.
It seems that an accurate FF model needs to incorporate two findings:
1. The relationship between LRC and MRC Fuel Flows and Machs (as a function of weight and altitude) cannot be assumed to be constant in an ECON model. The data provided by sk999 for the B777-300ER as a function of Cost Index should be a big help. I would think one can assume the variation with altitude and weight would be very similar for the B777-200ER as for the B777-300ER.
2. The “shape” of the remainder of the “Aero Figure 1” curve for speeds lower than MRC (again, as a function of weight and altitude) also needs to be determined. You have a semi-empirical model for this which has been validated (with a fairly small scaling) by Aero Figure 1. Independent of that, it seems we have good prospects for determining 3 points on each of those curves: LRC, MRC, and Holding. It’s not yet clear to me if that is sufficient to determine the entire curve, but your model runs should tell us.
@Gysbreght,
Thanks for the notes and the new plot, which is very informative. I am studying it and may have more questions later.
One question I did have now is whether/how you scaled the B777-300ER MRC data. Near Weight/delta = 1000 the red MRC curve seems inconsistent with the Holding curve. To my eye it looks like the MRC curve should be shifted a bit to the left (i.e., to lower weight). Are all the curves in this plot (LRC, MRC, and Holding) for the 300ER, or is it a mixture of 200ER and 300ER data? If so, which is which?
@DrB:
The LRC and Holding data in the chart are all the conditions contained in the FCOM tables for the B777-200ER with RR Trent engines.
The MRC speeds are from the ATPL booklet at Cost Index = 0 and FL350 as follows:
ATPL(A) Flight Planning Data Booklet
FL 350 delta = 0,235305
CI = 0
Weight (t) Mach W/delta LRC Mach LRC/MRC
180 0,73 764,9649 0,796 1,090
200 0,761 849,961 0,824 1,083
220 0,791 934,957 0,840 1,062
240 0,816 1019,953 0,840 1,029
260 0,83 1104,949 0,840 1,012
280 0,83 1189,945 0,840 1,012
300 0,829 1274,941 0,840 1,013
320 0,828 1359,938 0,840 1,014
Weight (t); Mach; W/delta; LRC Mach; LRC/MRC
180; 0,73; 764,9649; 0,796; 1,090
200; 0,761; 849,961; 0,824; 1,083
220; 0,791; 934,957; 0,840; 1,062
240; 0,816; 1019,953; 0,840; 1,029
260; 0,83; 1104,949; 0,840; 1,012
280; 0,83; 1189,945; 0,840; 1,012
300; 0,829; 1274,941; 0,840; 1,013
320; 0,828; 1359,938; 0,840; 1,014
@DrB: “Near Weight/delta = 1000 the red MRC curve seems inconsistent with the Holding curve.”
Why inconsistent? MRC and Holding speeds are based on different criteria.
Holding is based on unknown criteria because it is obviously not the “maximum endurance speed” stated in the explanatory text.
@DrB
I am wondering what your latest pin location is?
I am becoming increasingly convicted that MH370 may have flown 180S CTH down around 94E and if true that says the unsearched area just inside Arc7 at S32-34 is where MH370 is located (I give optional leeway for a short Easterly stint at the end as my scenario would include the possibility of intentional grounding at certain spot or trench).
@GeRijn
“While the subject has changed, still also the obvious conflicting CSIRO conclusions on their own latest drift-data remain unresolved.”
You and I have disagreed on how drift analytics are best done, and I am not trying to revisit that argument. This post is to simply point out how conclusions are deeply warped by how the question is posed. Tossing stuff in the water and observing where it goes is just not the way to get answers. For example, the ways to pose the flaperon question are:
1> If I toss debris in the ocean at such and such a place can it arrive on ReUnion in a time frame compatible with the actual flaperon find?
2> Given the finding of the flaperon on ReUnion at time X what is the most probable position of the flaperon on March 8, 2014?
Both questions are in the drift study domain. The first way to ask the question is how CSIRO does it, and it is just plain bad science. A properly configured reverse drift model is the only proper way to proceed. Doing it using the question 1> approach only determines what is possible not what is probable.
@Rob said, “I would respectfully point out another possible reason that the search failed, and that is the aircraft was glided to an impact point outside the search area southern boundary. Yes I know the drift studies seem to say otherwise, but I still think the CSIRO drift study results are suspect, and are designed to back up the S35 area, and bail the ATSB out of its predicament.”
Independent of drift studies, in my opinion, the BFO is a strong indication of a rapid descent, and the damaged parts are consistent with that. Others here disagree.
@Victor Iannello: The BFO could be a strong indication of a rapid descent if there was a credible scenario that resulted in those rates of descent at the time the 7th arc SATCOM transmission took place. Sofar no one has been able to produce such a scenario.
@Gysbreght
The CSIRO and other drifter based forward drift studies (based on starting points on the 7th arc) are the only sensible drift studies.
Studies based on debris as a starting point reversing in time turn reality up side down asif the plane crashed where the debris landed. Reversing the chaos the wrong direction (to the east) leading to everywhere and with that to nowhere regarding predicting a possible crash area.
@VictorI
Indeed the BFO is a strong indication of a rapid descent but still no proof of a high speed impact. Nobody has delivered that proof yet.
And the same goes for the damaged parts.
The ATSB and MOT-reports have given indications and suggestions but none of them has stated yet it’s 100% certain it must have been high speed impact damage. None of them even mentioned it as a conclusion yet.
@Gysbreght: We don’t know the details for the Boeing runs, but we do know that Mike E. in his simulator runs observed similar steep descents combined with steep banks. We also know pilot inputs could have caused steep descents even if there was no bank.
@Ge Rijn: You may wish to hang your hat on the possibility that there was an alive pilot that first allowed the plane to develop a descent rate of 15,000 fpm and then later decided to pull the plane out of the descent and gently land the plane. I think that is very unlikely.
@GeRijn
I put the words right out of your mouth. Great minds think alike, as they say.
@Gysbreght,
You said: “Why inconsistent? MRC and Holding speeds are based on different criteria. Holding is based on unknown criteria because it is obviously not the “maximum endurance speed” stated in the explanatory text.”
Thanks for contributing to this discussion of fuel models. Your knowledge and experience is certainly valuable to me.
The reason your plot appears inconsistent to me is that it shows the Machs are equal for MRC and Holding at Weight/delta = 1000. That implies the Fuel Flows are equal, too. For a given aircraft configuration, how can MRC have the same Speed and Fuel Flow as Holding? If MRC is Maximum Range, and Holding is maximum endurance, these will not occur at the same speed/Fuel Flow.
I suggest the problem is that you are comparing 300-ER MRC with 200-ER LRC and with 200-ER Holding. There must be a difference between these two aircraft such that their MRC curves are different. You have not attempted to “scale” the 300-ER MRC curve to the 200-ER configuration (I don’t know how to do this either), but I think you have compared apples and oranges.
Perhaps a better approach is to start by making the same plot using only 300-ER data. Then the MRC plot should always be in between LRC and Holding speeds. That is, at all Weight/deltas, the MRC speed should always be noticeably higher than Holding and noticeably lower than LRC. If the 300-ER plot shows this to be true, then a 200-ER plot must also show this (and therefore some scaling of the 300-ER MRC plot is required for it to be consistent with the 200-ER configuration).
@Victor Iannello: “we do know that Mike E. in his simulator runs observed similar steep descents combined with steep banks.”
Mike E. has shown details of only one og his simulator runs. In that run the airplane reached a rate of descent of 15,000 fpm 7 minutes after the second flameout, 5 minutes after the time of the 7th arc, just before crashing. I have strong reasons to believe that the ATSB/Boeing simulations similarly failed to back up the BFO’s.
@Dennis,
“Both questions are in the drift study domain. The first way to ask the question is how CSIRO does it, and it is just plain bad science. A properly configured reverse drift model is the only proper way to proceed. Doing it using the question 1> approach only determines what is possible not what is probable.”
Absolutely wrong understanding of the drift modeling.
1. CSIRO does a good job, but they have to justify what they are paid for. Besides being consistent with the DSTG, there is a problem with the aerial search for 30-33S, which was well covered. Now I understand better why CSIRO points on 35S.
2. There is no such a thing as “A properly configured reverse drift model”. We have discussed this in details. You can fool the public, but you can’t fool the nature.
3. Meteo France reverse study suggested the likely crash area south of 35S, peaking at 40S or so for the actual flaperon characteristics observed in the hydraulic lab. Are you satisfied with this result?
@DrB
Thank you for clarifying the respective burnout times. The figures now make more sense to me.
@DrB: “Holding is maximum endurance”
As I said only a few minutes ago, the speed scheduled in the FCOM table for Holding is not the maximum endurance speed. We do not know on which criteria it is based, but it is not the speed for minimum fuel flow. You yourself, together with Victor have pointed out to me that in the AERO Figure 1 the fuel decreases monotonously until the curve ends at left border of the chart. I can also show you a chart of the lift coefficients that show that the Holding speeds are scheduled somewhat arbitrarily, rather than following the natural characteristic of the airplane.
“Perhaps a better approach is to start by making the same plot using only 300-ER data.”
Please go ahead and show me what you got.
@Andrew,
Thanks for your comment in the previous thread and diagram. Interesting.
How do you know that the “GP 1 data” is only time/date data provided by the #1 general purpose bus data from the L AIMS cabinet. GP – general purpose? Or GP – global position? In general, the system time is synchronized with GPS data, so can position be.
There are two channels of IRS data. Are both sourced from ADIRU? Or the left one – from ADIRU, and the right one – from SAARU?
@Gysbreght: I am aware of three simulator runs. In the run where the left engine failed first combined with rudder trim of 0.5 units, the plane crashed 4.5 min after fuel exhaustion. In the run where the right engine failed first combined with rudder trim of 0.5 units, the plane crashed 7 min after fuel exhaustion. In the run where both engines failed together combined with no rudder trim, the plane crashed about 14.5 min after fuel exhaustion. Rapid descents developed in all runs combined with steep banks, and the initial bank seems to be related to the rudder trim. The rudder trim represents the out-of-trim since the simulator plane is perfectly trimmed at zero rudder trim. We don’t know what the rudder out-of-trim was for MH370. That trim could also change with altitude and speed.
I agree with Mike’s assessment that the simulator runs he witnessed are consistent with the observed BFOs.
Gysbreght:
Re your statement: “Mike E. has shown details of only one og his simulator runs. In that run the airplane reached a rate of descent of 15,000 fpm 7 minutes after the second flameout, 5 minutes after the time of the 7th arc, just before crashing. I have strong reasons to believe that the ATSB/Boeing simulations similarly failed to back up the BFO’s.”
WTF! Everything about your statement is deliberately, factually incorrect. First, I have shared details on three complete end-of-flight sims, not just one. Maybe not with you, for reasons clear to most, but with at least 30 others that have confirmed the conclusions I reached in Nov 2014 about a steep, rapid descent, ending close to the 7th arc. Second, the descent rates in our sims were as high as 29,000 ft/min, and all of them exceeded 15,000 ft/min as I recall. Thirdly, ATSB confirmed to me in 2014 that all our sims were consistent with what Boeing found in their simulator, and all these sims were 100% consistent with the BFO data. We would all appreciate it if you would stop making up patently false statements.
@ALSM: Perhaps I should refrain from commenting on your post.
@Victor Iannello: “the initial bank seems to be related to the rudder trim.”
Thank you for confirming the important rôle that the rudder trim played in those simulations.
@ALSM,
@Victor,
@Gysbreght,
I am also interested in the simulations, which are consistent with both the last BFOs. Could you post respective spreadsheets with BFO included to put dot in this?
@Gysbreght
“There is no such a thing as “A properly configured reverse drift model”. We have discussed this in details. You can fool the public, but you can’t fool the nature.”
Not sure where you are getting that nonsense. I stand by my statement.
@Oleksandr
“How do you know that the ‘GP 1 data’ is only time/date data provided by the #1 general purpose bus data from the L AIMS cabinet. GP – general purpose? Or GP – global position? In general, the system time is synchronized with GPS data, so can position be.”
From the AMM:
“The SDU gets the following data from AIMS:
* Airplane position (inertial reference system (IRS) bus from left and right AIMS cabinets)
* Airplane identification number and digital communication data (data communication management function (DCMF) bus from left and right AIMS cabinets)
* Time and date (general purpose (GP) 1 bus from left AIMS cabinet)
* Data load data (data load bus from the left AIMS cabinet)
* Data load discrete (from the left AIMS cabinet)
* Control and BITE signals (central maintenance computer (CMC) bus from the left AIMS cabinet).”
“There are two channels of IRS data. Are both sourced from ADIRU? Or the left one – from ADIRU, and the right one – from SAARU?”
Both channels of IRS data are sourced from the ADIRU. One channel is from the L AIMS cabinet and the other is from the R AIMS cabinet. Both AIMS cabinets receive ADIRU data via the Flight Controls ARINC 629 bus. That data is then converted into ARINC 429 by the DCGF in each AIMS cabinet and then sent to the SDU via the L & R IRS data buses.
@Ge Rijn
“The debris tells a conflicting story regarding the final BFO’s I’m still sure. It does not show the signs of a high speed impact but clear signs of a ditch-like event.”
Debris items 2 (piece of right wing No. 7 flap track fairing), 3 (piece of right horizontal stabiliser panel), 4 (piece of engine nose cowl bearing RR logo), 5 (part of door R1 stowage closet), 6 (piece of right engine fan cowling), 7 (part of a panel of the wing to body fairing), 8 (part of the tail cone of the No. 1 flap track fairing), 9 (part of the outboard section of the “Upper Fixed Panel forward of the flaperon” off the left wing), 11 (seat back trim panel for encasing the IFE monitor), 15 (part of the outboard section of the “Upper Fixed Panel forward of the flaperon” off the right wing), 16 (piece of cabin interior panel), 20 (part of the right aft wing to body fairing panel and 22 (part of vertical stabilizer panel) are either consistent with a high speed, high energy impact or, at the very least, not inconsistent with a high speed, high energy impact.
The best that you can say about the debris is that it is inconclusive; collectively it neither conclusively supports nor conclusively eliminates either a high speed, high energy impact or a ditch-like event.
@Gysbreght
Sorry for my last comment. It was Oleksandr, not you, who railed against reverse drift modeling.
@Oleksandr
“1. CSIRO does a good job, but they have to justify what they are paid for. Besides being consistent with the DSTG, there is a problem with the aerial search for 30-33S, which was well covered. Now I understand better why CSIRO points on 35S.”
So it takes an advanced degree in oceanography to toss stuff in the water and to watch where it goes? Sad, really.
@Oleksandr
Re: “Besides being consistent with the DSTG, there is a problem with the aerial search for 30-33S, which was well covered.”
Not entirely well covered. As illustrated in the the CSIRO’s “The search for MH370 and ocean surface drift – Part II”, Fig. 4.2. Efficacy analysis of the surface search, the surface search coverage was not completely efficacious between 30°S – 33°S; there is a sizeable lacuna at about 30° 10’S 98°E and there is a smaller patch rated at less than 50% probability of detection at 32° 50’S 95°E. Moreover, on the basis that the surface search of the area between 30°S – 33°S didn’t commence until 20 days after the airplane went in, I think that assigning a 100% probability of detection for large tracts of that region is overly optimistic.
@ALSM. Yes, your simulations resulted in a high rate of descent. However when you say that, “all of them exceeded 15,000 ft/min as I recall”, that is enough just to exceed the minimum 2nd BFO descent rate of 14,600fpm but not the bulk of the bandwidth to the maximum, 25,000fpm.
Overlooking that though, still there remains a twofold challenge in proving consistency with the BFOs. One, did the descents result in the 10,823 fpm rate increase in the 8 secs between them, that applying to all descent rates from minimum to maximum and being needed for BFO consistency? (Note that this derives from the 182 +2 Hz difference in the two BFO readings. Rounding of those two figures, adding to a maximum gross ±1 Hz, provides a tolerance of just ±60fpm to the 10,823fpm). Two, did the elapsed times of these descent rates from fuel exhaustion to the BFOs match the 2 minutes and 2:08 of the BFOs?
And even if they did, what difference could the unknown MH370 rudder trim have made? Did the SSWG/ATSB/simulations allow for that uncertainty as a variable in a matrix of them and others? As another part of that matrix, the ATSB has described both parts which add to the 2 minutes as “approximate” (and in how approximate I leave aside my own views; and also whether fuel exhaustion or AC power loss was the datum). Without such a matrix how could they imply that the Boeing simulations match the estimated timings when it is uncertain what they are matching, ie when there is a spread of conditions? Even the recent introduction of different configurations into simulations, with just one or two to look into each, suggests to this outsider that their approach might not be that rigorous and might not embrace all reasonable circumstances.
Also, as a Parthian shot, did they confirm that the aircraft attitude at both BFOs allowed SDU connection with the satellite?
You say, “Thirdly, ATSB confirmed to me in 2014 that all our sims were consistent with what Boeing found in their simulator, and all these sims were 100% consistent with the BFO data.” Now, as to that last bit. Can you say how they did that back then? Their detailed final BFO analysis, leading to the minimum/ maximum descent rate band, was published in November 2016. Besides, “100%” means they all matched in all detail and in each case. Really?
I think it most likely that given a full explanation as to how the SSWG/ATSB/Boeing, including MAS (which did its own simulations), concluded that the various simulations were consistent with reality, ie the BFOs, we would be satisfied. In the meantime though, conjecture has popped up some reasonable questions.
One further thought is that your simulations exhibited a high descent rate at the bottom whereas the most recent Boeing simulations at least do not really point to that as highly likely. The appearance (I stress) currently offering is of the SSWG/ATSB cherry picking and demonstrating at best that descents COULD have been met BFO criteria. There has been no conclusion that the high BFO descent rates continued; and as Ge Rijn correctly points out there has been no claim of a high speed descent ending from recovered flotsam evidence. Unfortunately even if there were, that of itself does not demonstrate the crash was close to the 7th arc since there could have been a dive after a glide.
For my part, along DennisW’s lines, I presume good faith and competence are behind the ATSB claims for simulations, even including fuel availability for the APU. Extending that theme I for one accept that it is reasonable to assume the aircraft did not go further from the final log-on than they allow. However Gysbreght has some difficulty with simulation interpretation and would like proof. For my part I do not think that unreasonable, or for him to question the relevance of the simulation of yours that he has seen.
@Victor. On the subject of this post one worry is that the final report will not contain proprietary information. Without that there could well be insufficient to allow future searchers the background they will need.
@Ge Rijn. The 20 minutes duration of the more recent Boeing simulation was in a slow spiral so would have needed a pilot to gain distance.
@ALSM. My second para, second last line please delete, “to the BFOs”.
@Mick Gilbert
Re your post to GeRijn iro the debris evidence, I am going to take issue with you when you suggest the debris is core consistent with a high speed, uncontrolled impact, because this is most definitely Not the case. Items 9 and 15 (plus the RH flaperon) are particularly significant, imho. The condition of the flaperon hinge attachment points (on the flaperon underside) suggest the flaperon was sheared off cleanly at the hinge points, by a suddenly applied force to the flaperon underside. It has already been established by the ATSB that the RH outboard flap was stowed at the time, and that the flaperon was in its neutral position. Now consider the flaperon upper fixed panels, items 9 and 15, one from the RHS, the other from the LHS, from corresponding locations on the two wing upper surfaces, and in a similar condition – further compelling evidence that the impact force must have been specifically directed on the airframe underside, as in a belly-flop following a deep stall. This is not what you would expect from a high speed, nose-first impact. The other trailing edge components, the nosewheel door plus the RH stabilizer part and the co-located fin piece (also RH) all point to the aircraft entering the water in a stalled condition, RH wing down. It is my belief that the debris suggests the aircraft hit the water in a deliberately controlled manner, designed to breach the hull sufficiently to promote rapidly sinking, but leave the minimum of debris on the surface.
@David
The two sample times at 00:19:29, and 00:19:37, a mere 8 seconds apart, were in effect “arbitrarily” determined by the SDU logon sequence. Are we really expected to accept the notion that during this brief, arbitrarily determined snapshot of the unpowered descent, the descent rate accelerated from 5000fps to 15,000fps?. Quite ridiculous. The BFO interpretation has to be wrong. For all the various members of the SSWG to be apparently on board with the consequent, uncontrolled high speed descent scenario, suggests a worrying degree of collusion between them, in order to present a united front, purely to keep the search area manageable, not to help find the plane.
@Mick Gilbert
Although I have my own opinion I agree the assumptions on the debris damage must be called inconclusive as long as the ATSB, Malaysia, France or any other official party are not able or willing to declare it must have been a high speed impact or not. None of them have declared something conclusive based on their forensics yet.
To me it’s hard to believe they don’t have the conclusive answer already. With the debris at hand and all their forensic techniques available this question must not be that hard to answer.
I rather think it’s an elephant in the room they try to keep invisible by staying vague and only providing ‘indications’ without conclusive proof and conclusions. Keeping everyone guessing.
To avoid suggesting secrecy and doubts about their independence they could easily provide some debris-pieces chosen by an independent party to be studied independently.
It seems they are not willing to do this. And I wonder why.
The only reason I can think of is they’ve got something important to hide.
@David
20 minutes duration of flight after second engine flame-out must be around the limit of a possible (controlled) gliding distance.
I would like to see those simulations (the report mentions several simulation 20 minute descents).
I imagine those can not be tight spirals or even one full spiral.
Rather a very slow turn with a very wide radius without a bank.
Then I also like to know in which attitude the plane reached the surface in those simulations.
Is there further information available on those simulations?
@DrB: Here is a chart for B777-300ER LRC, MRC & LRC/MRC.
@Mick Gilbert,
Re: “Not entirely well covered. As illustrated in the the CSIRO’s “The search for MH370 and ocean surface drift – Part II”, Fig. 4.2. Efficacy analysis of the surface search, the surface search coverage was not completely efficacious between 30°S – 33°S; there is a sizeable lacuna at about 30° 10’S 98°E and there is a smaller patch rated at less than 50% probability of detection at 32° 50’S 95°E. Moreover, on the basis that the surface search of the area between 30°S – 33°S didn’t commence until 20 days after the airplane went in, I think that assigning a 100% probability of detection for large tracts of that region is overly optimistic.”
I have analyzed non-detection issue in a similar manner, and came to the similar conclusions as CSIRO. Yes, there is a depression in the probability to find debris just at 30°S (my models show max coverage of approximately 20% just north of 30°S), but this gap is small.
With regard to 32° 50’S, note that the coverage exceeding 30-40% could occur during two or three days, a factor which would increase the probability of finding the debris field. In comparison, the coverage from approximately 35.5°S to 37°S is almost zero; there are also many gaps between 25°S and 30°S, where the coverage could drop to below 10%.
@Andrew,
Thank you for your answers. Is the source of your information publicly available? I just don’t want to bother you each time with my silly questions if I can find answers by myself.
@Dennis,
“Not sure where you are getting that nonsense. I stand by my statement.”
You earlier statement was absurd, so standing to it is a funny stubbornness. Can you name any credible reverse drift study produced up to date? I have explained you several times why. If you still do not understand, read Pattiaratchi’s papers. If you are not satisfied, think why most of the studies are forward and not reverse (e.g. CSIRO, UWA, Deltares, IPRC, Jansen et al, etc.). If you are still not satisfied, read Meteo Frances’ reverse drift report. Duh, I don’t know what to suggest you next if you still cant grasp.
Re: “So it takes an advanced degree in oceanography to toss stuff in the water and to watch where it goes? Sad, really.”
Just think why it takes a surgeon 10 years of university and internship study only to be qualified to hold a lancet in hands. I am sure you can also hold lancet in you hands. Same thing about the oceanography and tossing stuff in the water.
@ROB
There is considerable conjecture as to exactly how the the starboard flaperon separated from the wing. There is, however, no sign of impact or compression damage to the underside of the flaperon as might be expected if it were subjected to impact forces directed on the airframe underside, as in a belly-flop following a deep stall. AF447 was most assuredly a belly-flop impact following a deep stall and the damage to and separation of control surfaces is not even vaguely similar to what we see with either the flaperon or flap section.
It is also difficult to reconcile the symmetricalness evidenced by debris items 9 and 15 with the asymmetrical impact occasioned by the aircraft entering the water in a stalled condition, RH wing down.
Reconciling item 22 (part of vertical stabilizer panel) with anything other than a high energy impact is difficult if not impossible; the vertical stabiliser is one of the most robust assemblies on the airplane and is well removed from the point of impact in either a low angle of attack controlled ditching (look at Ethiopian Airlines flight 961 – the vertical stabiliser was completely intact) or a belly-flop impact (look at AF447 – once again the vertical stabiliser was intact). You simply could not deliver the energy required to shatter the vertical stabiliser in such a fashion as to liberate item 22 unless the airplane was either inverted on impact or entered the water almost vertically.
As I said earlier, the best that you can say about the debris is that it is inconclusive. Collectively it neither conclusively supports nor conclusively eliminates either a high speed, high energy impact, a low angle of attack ditching-like event or a belly-flop-style impact following a deep stall.
@Ge Rijn
Yes, I’m surprised that we haven’t seen a proper metallurgical analysis of the flaperon hinges with a view to determining how they failed or a more holistic approach to analysing all the debris items with a view to determining the type of impact. Something to watch for in Section 1.12 Wreckage and Impact Information of the forthcoming final report.
@Oleksandr
“You earlier statement was absurd, so standing to it is a funny stubbornness. Can you name any credible reverse drift study produced up to date?”
Geomar
As I recall we cannot even agree on what constitutes a coincidence, so there is little point in discussing this.
@Mick Gilbert
Yes, hopefully the final report will provide the conclusive answers.
Still I think it would be sensible/appropriate if independent investigation of some chosen debris-pieces would be allowed in the meantime. This will definitely back up their final report findings/conclusions removing doubt about objectivity on their half which is so deeply felt by many not the least the NoK.
Like to at to your comment that the Asiana 214 crash landing shows a simililar piece at the same spot seperated on the vertical stabiliser as in the MH370 stabiliser. Although in this crash the complete tail seperated I assume that stabiliser’s leading edge was hit by debris seperating from the wings and/or engines.
I agree a belly-flop (A447 kind of) impact would probably show extensive compression damage and not a pattern of mostly trailing edge wing/engine related debris. As would a high speed impact not show also (but a lot of compression damage).
Though during a low speed, low AoA, ~level ditch-like entry of the surface (like the Hudson-ditch) vertical impact forces wouldn’t be that high to cause severe compression damage I think.
Pieces would rather break off by torque forces and ripped off by the water passing with high speed horizontally causing mainly tension damage. This is what the pieces mainly show that are named in the MOT-report.
Concerning an assymetrical impact there’s also the left outboard flap trailing edge piece from an outer section of the left outboard flap.
This makes it hard to believe the plane entered the water right wing first under a considerable bank IMO.
But again; all still assumptions.
@DrB: I used my fuel model to calculate the fuel mileage for a range of weights at FL350. To make comparisons easier, I normalized the Mach number by the calculated Mmrc, and I normalized the mileage by the best fuel mileage, which occurs at Mmrc.
At the lowest weight (160 MT), the curve is fairly parabolic. However, as the weight increases, Mmrc occurs at a higher Mach number, and the curves fall off more rapidly to the right of the peak due to the increased drag with Mach number. There are many other effects, including the effect of parasitic losses which reduces the performance of the engine at lower thrust levels. (That’s the reason that for the same altitude, weight, and speed, the fuel flow is less for a single engine than two.)
The bottom line is I don’t think it will be easy to take a single fuel mileage curve and apply it to different conditions.
@Dennis,
Geomar… Are you kidding me? Wake up, Dennis. Geomar is the least credible among all the studies, and they should have taken it seriously long time ago.
If you compare it with Meteo France (report by Pierre Daniel, also reverse drift), you may find that under certain incorrect settings, Geomar produces results similar to Meteo France’s results. Therefore, throw away Geomar study to the garbage bin until they produce something better, in line with the updated information.
Also, I would be interested to know on the basis of what you mistrust Meteo Frances’ results, but trust Geomar’s study, both based on the reverse drift approach?
Finally, why do you think the cloud produced by Geomar is in-principle more informative than anything else based on the forward drift approach?
@Oleksandr
Placing things in the water and watching where they go is childish. It is not a drift study, and it is not good science. The very act of doing so implies that you have a preconception of where the aircraft terminated. As I said earlier, that approach can only validate what is possible not what is most probable.
I don’t have an opinion relative to the Meteo Frances’ study or the Geomar study. I am not an oceanographer, and I have no qualifications in that domain.
@Dennis,
“I don’t have an opinion relative to the Meteo Frances’ study or the Geomar study. I am not an oceanographer, and I have no qualifications in that domain.”
Geomar study, which showed a cloud near Indonesia, did not include even a basic thing – wind impact… Not talking about waves, Coriolis, Stokes drift, or other stuff, sometimes claimed to be also important. No other debris were considered by them – the flaperon only. No extensive statistical analysis. Geomar study is a childish exercise compared to MeteoFrance, not talking about CSIRO or UWA forward drift studies, for example. And you called it a credible study? How come?
@Oleksandr
You should be having this discussion with Geomar. As I said ocean drift analytics are outside my pedigree.
My guess is that you favor the forward method because it is something you can understand.
@Dennis,
“My guess is that you favor the forward method because it is something you can understand.”
It would be more correct to state that I have a better understanding of what can be achieved with a forward screening method compared to what can be achieved with a reverse drift approach. Apparently you have no idea at all. You have a batch of plots from MeteoFrance and Geomar: what can you conclude from these plots?
@Oleksandr
My comments have to do with the methodology not the calculation details.
@VictorI,
Thank you for producing that plot of normalized fuel mileage versus normalized Mach for different weights at a constant flight level. It’s very useful and answers my basic question.
For clarification of your plot, how did you determine Mmrc? Did you just find the Mach that maximized fuel mileage using your model? That seems like a lot of work to do at each weight/flight level. Did you figure out an easier method?
If you have a listing of the normalizing values of Mmrc and Best Mileage for the curves you plotted, it would be useful to have those numbers also, or, better yet, just the Mileage and Mach values (not normalized) for the points in your plot.
It would also be most interesting to see the same plot with a constant weight and variable flight level, if that is not too much trouble. If so, I understand. Time is limited for everybody.
@Gysbreght,
Thanks for making the 300ER plot. The MRC curve now appears consistent with my expectation relative to the LRC and Holding curves (they are well separated).
The next step is to figure out, if possible, how to scale the 300ER variation in Mach and FF with Cost Index to the 200ER configuration so we have a useful ECON model for 9M-MRO.
Unfortunately, the ATPL document only shows Mach and FF for two CIs (LRC and CI = 50) over a range of weights and Flight Levels. There is also a table that shows the Machs for a range of CIs and weights at FL350, but not the Fuel Flows. Still, I think that is enough information for a partial solution (at one Flight Level). We also know that Fuel Mileage is linear with CI, and we know that Boeing uses CI = 180 for LRC. Since we have relative speeds versus CI on page 1-12, and we have two Fuel Mileage values for CI = 50 and CI = 180, we can solve for the fuel mileage (and therefore the FF) at each value of CI. That gives us a complete ECON model, but only at FL350 (which is used for the CI table on page 1-12).
The next question is, how does this ECON model change with Flight Level? We can use the table for CI = 50 on page 1-14 which shows the Machs and FFs for various weights and Flight Levels (analogous to the usual LRC table on page 1-15). Taken together, the two tables allow us to calculate the (best) fuel mileage at MRC, but they don’t allow us to determine either the Mach or the Fuel Flow at MRC. In order to do this, it seems we need either more data from Boeing or a semi-empirical model like Victor’s that can accurately predict relative fuel mileage versus speed given weight and Flight Level.
One test that can be done to the ATPL 300ER data is to look at the relative fuel mileage of CI = 50 versus LRC (over the weight and Flight Level ranges of the tables on pages 1-14 and 1-15). If LRC is defined by Boeing as a XX% drop in fuel mileage relative to MRC fuel mileage, then the drop from CI = 50 fuel mileage to LRC fuel mileage should be (180-50)/(180 – 0) = 72% of XX%. The Aero Figure 1 plot is confusing because the label says XX% = 1%, but the actual numbers on the graph show XX% = 0.75%. Which is correct? Perhaps averaging the 300ER fuel mileage differences will give us a reliable answer. You can’t get great accuracy using only one FM difference because the speeds are rounded to three digits; you can do much better by averaging over the whole table.
@DrB: LRC speed is subject to a maximum speed limit of M.84, otherwise it is defined by the mileage of 99% of MRC.
Cost Index =0 is subject to maximum M.83, otherwise it is equal to MRC.
Just to be complete, LRC is limited to max M.84 and Vmo-5kt.
@DrB:
“For clarification of your plot, how did you determine Mmrc? Did you just find the Mach that maximized fuel mileage using your model? That seems like a lot of work to do at each weight/flight level. Did you figure out an easier method?”
It’s easy to “Solve” for Mmrc by finding the M that minimizes the fuel mileage. However for the plots, I calculated the values with a spacing of M=0.01, so I simply used the minimum value for the data set corresponding to a given weight. I don’t think the graph would change much if I used the exact value.
“If you have a listing of the normalizing values of Mmrc and Best Mileage for the curves you plotted, it would be useful to have those numbers also, or, better yet, just the Mileage and Mach values (not normalized) for the points in your plot.”
I only normalized so the curves could be more easily compared. Here is the “non-normalized” fuel mileage curves at various weights for FL350. Also note that I included in the plot the values from the LRC and Holding tables. The model output is normalized so that it exactly agrees at LRC speed.
It is interesting to note that because the LRC and Holding speeds lie to either side of the MRC speed, the fuel mileage values are fairly close.
@DrB: “We also know that Fuel Mileage is linear with CI”
I’m not sure about that. Can you remind me how you have determined that Fuel Mileage is linear with CI?
@Gysbreght,
Thanks for supplying the maximum speed limits for MRC and LRC. That helps.
I don’t think that Boeing defines LRC for a B777 as exactly 1.00% fuel mileage degradation. Aero Figure 1 indicates about 0.75%, and a few of the pairs of points in the CI50 and LRC tables for the 300ER in the vicinity of 240 MT and FL350 give values between 0.72% and 0.96%. In other words, none of the quantitative data we have show a fuel mileage degradation for LRC as large as 1.00%. I suspect the 1% is simply a rounded-up estimate of the true value, which is slightly less than 1%.
@VictorI,
You unnormalized plot is even more informative. I didn’t appreciate how much the fuel mileage varied (~100% increase in mileage for ~50% weight reduction).
@DrB: “we know that Boeing uses CI = 180 for LRC”
According to that AERO article, CI=180 is the “Approximate LRC Equivalent” for the B777. What that means is illustrated here.
“I don’t think that Boeing defines LRC for a B777 as exactly 1.00% fuel mileage degradation.”
I think that Boeing, just as all transport airplane manufacturers have always done, defines LRC as 99% of maximum mileage. In the AERO Figure 1 LRC is not 99% because it is overridden by the M.84 maximum speed limit. Likewise, in the upper left corner of the Long Range Cruise Control table, the speed is limited to 325 KIAS (Vmo-5).
@Victor Iannelo: Perhaps it would be helpful if you could indicate Mmrc in your fuel mileage curves for FL350, and also the speeds for 99% max mileage.
@DrB: ” I suspect the 1% is simply a rounded-up estimate of the true value, which is slightly less than 1%.”
How is the “true value” defined?
@Mick Gilbert said:
“In fact there is no logical reason for a malicious perpetrator taking the somewhat drastic action of depowering the Left AC Bus for any reason in general, leave alone doing so specifically just to disable the SATCOM.”
If the left bus (IDG and backup gen) and tie was disabled purposely (and there wasn’t an incident/accident) then the only logical reason might be the need to disable something that couldn’t be disabled from the cockpit by any other means. That narrows things down a lot. It certainly would not be to disable the SDU – ACARS and the IFE can be easily disabled in the cockpit, and any ping(s) – even if known about – would not matter when the aircraft was already being tracked by primary radar from IGARI up the Straits.
It would need someone with a copy of the 777 AMM (@Don? @Andrew?) to identify what systems and items would remain unpowered with the left bus (both gens) and tie off, ie. that would not be re-powered by any backup or standby system kicking in.
That same list of systems and items might also inform why the left bus and tie was re-energised later – disabling system/item ‘a’ might also have had the undesired (but necessary) consequence of also disabling system/item ‘b’. Later, in the Straits, having ‘a’ disabled was no longer necessary, but having ‘b’ back on was essential or preferable for the next step?
If the IFE was disabled not by switch but by disabling the left bus, it shows a detailed knowledge of 777 systems and perhaps planning. And it would explain why the IFE automatically come back on when the left bus was re-enabled.
@Oleksandr
“Is the source of your information publicly available?”
No, I don’t believe the AMM (Aircraft Maintenance Manual) is available to the public.
Latest from Higgins. SSWG deliberations sought, apparently with an emphasis on end of flight and the possibility that there was dissent within the SSWG as to there being no active pilot. That seems to be a hangover from the ATSB brief removal of a ‘nil SSWG dissenters’ statement from its website when it was realised that not all the Group had been given enough time as yet, and which it subsequently explained.
The Australian12:00AM May 19, 2017
EAN HIGGINS
ReporterSydney
@EanHiggins
Australian Transport Safety Bureau chief commissioner Greg Hood has declined to accede to pleas from families of Malaysia Airlines flight MH370 victims to release key documents about the bureau’s fruitless search for the aircraft.
Danica Weeks, who was left a single mother looking after two young boys when her husband Paul disappeared with MH370 three years ago, has claimed “Mr Hood has simply denied families throughout the world the information we so desperately need about what happened to our loved ones.”
The call came as a second relative of the six Australians lost on MH370, Teresa Liddle, and the association representing the families of the 153 Chinese victims, joined Ms Weeks in urging journalists to continue to probe for answers in the face of Mr Hood’s public attacks on The Australian for reporting critiques of the ATSB’s failed search by independent scientists and aviation experts.
After The Australian reported scientists in Europe and the University of Western Australia had long ago warned that drift modelling of MH370 debris showed early on in the two-year underwater search that the ATSB was looking too far south, Mr Hood, who took up his role in July after a career at Airservices Australia and the RAAF, criticised the reports in a letter to this newspaper and on the ATSB website.
Flight MH370 disappeared on March 8, 2014 on a scheduled trip from Kuala Lumpur to Beijing, with its radar transponder turned off and radio contact cut 40 minutes into the flight.
Primary radar and automatic electronic satellite tracking data showed the aircraft double backed over Malaysia to the Andaman Sea, then took a long track south to the southern Indian Ocean.
Ms Weeks and the group known as MH370 China Families called on Mr Hood to reverse his rejection of a freedom of information request from The Australian for international assessments of satellite tracking data.
An ATSB spokesman said this week that the data “has been painstakingly analysed by leading experts in their fields, who form the MH370 Search Strategy Working Group, to determine the aircraft’s most likely flight path”.
But Colin McNamara, the ATSB’s general manager, strategic capability, refused The Australian’s initial FOI request for the SSWG analyses, saying its public release “would, or could reasonably be expected to, cause damage to the international relations of the Commonwealth”.
In reviewing and rejecting the FOI request, Mr Hood invoked the Transport Safety Investigation Act, which makes it a crime for current or former ATSB staff to release material deemed restricted, punishable by two years in prison.
While independent experts generally agree that the satellite data gives a good indication of the track of the Boeing 777, they differ on whether it can accurately say how the aircraft finally came down.
The ATSB maintains that the satellite data shows MH370 went down in what’s become known as a “death dive”, or unpiloted crash.
Top US air crash investigator John Cox has said he does not believe the satellite data is good enough to conclusively support the ATSB’s rapid descent theory. Several senior airline pilots and air crash investigators maintain the evidence suggests Captain Zaharie Ahmad Shah hijacked his own aircraft and flew it to the end and outside the ATSB’s 120,000 square kilometre target search area, which it had defined based on the “death dive” theory.
In a letter to the The Australian published this week, Ms Weeks wrote that if Mr Hood continued to put “diplomatic niceties” ahead of “the sensitivities of the families and friends of 239 people”, he should stand down so a replacement could be appointed who would “release the FOI immediately”.
Speaking of the ATSB’s definition of the search area based on its interpretation of the satellite data, the Chinese families association noted that “the search based on this data has failed”.
“We would welcome the release of any additional information which highlights inconsistencies in the official explanation,” the association told The Australian.
“We are disappointed in (the ATSB’s) explicit and implicit endorsement and their acting as proxy communicators for the Malaysian authorities,” the Chinese families said.
The Chinese families spoke of their difficulties in campaigning in authoritarian China.
“There are more than 350 of us, who communicate by social media aliases,” the association said.
“We meet informally in small groups, or in larger approved groups attended by police.”
Mr Hood did not respond to questions from The Australian about whether he would seek permission from members of the SSWG to grant the FOI request, and whether Malaysian authorities had asked for this and other material to be suppressed.
An ATSB spokesman said the bureau was “very conscious of, and deeply saddened by, the prolonged and profound grief suffered by the families of those on board MH370” and remained willing to brief family members “on all aspects of the search”.
Concerning David’s quoted article…
“Top US air crash investigator John Cox”
Collector of approbations and serial opiner, but air crash investigator: no.
http://www.safeopsys.com/staff/captain-john-m-cox/
Yes, DennisW, I did intend to provide that link listing the approbations & the outlets that have carried his opinions.
If one has 80 minutes available, a video has been published at Vimeo:
IPAA Events – Australia’s role in MH370
@Don Thompson
Don, I think that you’re being harsh towards John Cox; he’s a Fellow of the Royal Aeronautical Society and Member of the International Society of Air Safety Investigators; he was part of the NTSB team that investigated USAir Flight 427 and he’s in pretty esteemed company as a winner of the Sir James Martin Award.
I disagree with John’s observations with regards to the final BFO data (I have a sneaking suspicion that he has revised his position on that matter) but I don’t think it’s appropriate to question his qualifications as an air crash investigator. I dare say he has more practical experience in the field than most, if not all, of us (I am more than happy to be corrected on this point).
@ALSM. The Search and debris examination update of Nov 2016, p10 notes, “The maximum OCXO drift value observed in the previous data of 9M-MRO was around 130 Hz and if the power interruption was sufficiently short, the OCXO drift could be negligible”.
In its page 11 Table 2, which implicitly is the ATSB gauge for assessing whether simulations meet the several descent rate criteria for BFO compatibility, the ATSB adds this 130Hz to Case B. This leads to the maximum descent rates applicable to the 2 BFOs being considerably higher than otherwise.
With the SDU still operating at the 6th arc, the maximum time it could have been off would extend from post 6th arc power loss to the left AC bus (in a configuration in which it was powered by the right engine, which ran down at that point) to power restoration by the APU, a maximum 7 ½ minutes later; that is supposing the right engine failed immediately after the 6th arc. The minimum time it would have been off would have been the nominal minute for APU autostart after all AC power was lost at left engine run down.
The ATSB has described electrical configurations which lead to the former as unlikely, leaving the 1 min unpowered as the most likely.
With a 1 minute power loss I assume the OCXO drift would be minimal. If in fact it would be near enough zero the maximum descent rates would be 6,500fpm (was 14,200) from 111Hz (at 1st BFO) and 17,350fpm (was 25,000) from 295Hz (at 2nd). The 10,800 odd fpm difference remains.
Can you say why they would have used the higher maxima?
@Gysbreght,
You said: “I think that Boeing, just as all transport airplane manufacturers have always done, defines LRC as 99% of maximum mileage. In the AERO Figure 1 LRC is not 99% because it is overridden by the M.84 maximum speed limit. Likewise, in the upper left corner of the Long Range Cruise Control table, the speed is limited to 325 KIAS (Vmo-5).”
Nice work! I think you have resolved that particular issue. Your explanation of the speed clipping makes perfect sense and explains why the Aero Figure 1 curve data show less than the nominal 1%.
Two issues previously raised remain unresolved. First, with regard to the relationship between CI and fuel mileage, I will look for my old notes or maybe just try to repeat that analysis to see if my memory is correct. I think there has to be a simple relationship between CI and some combination of airspeed and fuel flow. The Cost Index scale factor (why Boeing sets LRC = 180) may depend on actual relative operating hour (non-fuel) cost and fuel cost, but I don’t think the equation used to interpolate/extrapolate CI to speeds other than MRC/LRC depends on those costs. Without that equation, you don’t have an ECON mode, so it is necessary to get it right.
Second, I knew that LRC was not identical to CI180, but I thought this was only in the way the FMS handled headwinds (varying airspeed for ECON and not for LRC). I don’t understand your latest plot from the ATPL data that shows large airspeed differences (M0.06) at 180 MT and FL350. I’ll have a look tomorrow and see if I can figure out what is going on here. It’s certainly not what I expected to see.
@Mick
“Don, I think that you’re being harsh towards John Cox”
Don is in the not liking Higgins camp. Personally, I love the guy. He is the only journalist I am aware of that is critical of the ATSB which deserves a great deal of criticism.
@Rob. “For all the various members of the SSWG to be apparently on board with the consequent, uncontrolled high speed descent scenario, suggests a worrying degree of collusion between them, in order to present a united front…”
4th dot point p14:
https://www.atsb.gov.au/media/5771939/ae-2014-054_mh370-search-and-debris-update_2nov-2016_v2.pdf
Consider the possibility Rob that the SSWG members were actually able to agree, because the evidence before them was convincing. Since not much evidence has been published I do not think we can know either way. It then becomes gut feel on the one hand against faith on the other, the latter in the evidence actually being there though not releasable, at least as yet.
@Ge Rijn. 20 minute descent. Bottom of p13 of the above refers. Numbering them from the left, the 20 minute flight appears to be a left turner other than 3 with number 1, yellow, being likely; but then others have double loops.
Numbering them at the top from left; 1, 2, 4, 5 and 6 were incompatible with the BFOs’ descent rates. Numbers 7, 8, 10 (and Gysbreght thinks 9) were of novel electrical configuration, right engine failing first.That leaves 3 as the only one of this batch of left turners (ie standard electrical configuration, left IDG powering the SDU) reportedly consistent with the BFOs. From the general description there were varied speed and altitude starts and differing turbulence levels. The ‘novels’ right turned and had shorter distance descents. I assume that residual trim dominated, even with the left flaperon developing less lift than the right, its PCUs being in by-pass. The right turners might have been responding to trim but also the flaperon lift being out of balance. A lot of this information came from the ATSB to Gysbreght.
The 20 minute flight appears to be a left turner other than 3 with number 1, yellow, being likely but then others have double loops.
Ge Rijn. Please overlook the last two paras. Irrelevant/redundant
@DennisW
G’day Dennis, yes, I’d inferred that you were an Eanophile.
Like some of the other contributors to these comments, I have had some dealings with Ean and while he is most assuredly a personable chap, he regularly and, perhaps more annoyingly, repeatedly gets basic stuff wrong. For better but much more likely for worse, Ean and The Australian hitched their wagon to Captain Byron Bailey early on in the piece and Captain Bailey is vehemently, consistently and almost pathologically critical of the ATSB, even going so far as to fabricating a non-sensical statement that he attributed to the ATSB just so he could say how stupid they were for saying such a thing. It was one of the most appalling episodes I have witnessed in the entire coverage of MH370 (Ross Coulthard’s 60 Minutes effort is the runner up) and it passed almost completely without censure. Consequently, a lot of what Ean writes is stilted or at the very least filtered so as to offer as much protection for Captain Bailey as possible.
Now, I have been critical of the ATSB where I believe criticism is due; I would like to think I am fair in that regard, I certainly haven’t made anything up. If you want to see the ATSB copping it in the chops at every turn then Ean’s your man and The Australian’s your newspaper but so far as a considered and informed discussion goes, well their carry on is akin to a Punch and Judy show; you might be entertained but you’re not going to learn much. Frankly, I think that we can do much better than that and I firmly believe that the readers of that newspaper deserve much better than that.
@David
I did some searching. First it’s not clear to me if the 20 minute simulations are included in the graphic (do you know?).
But let’s assume the yellow trajectory is one of them for it went the most far from the 7th arc.
First I tried to find the location above which the trajectories have been projected in Google Earth and the ~point at which the turns of the simulations start. I found the location.
It’s at ~37.30S ~88.50E. Then I looked for the position at which the yellow trajectory ends. This is at ~89.50E.
So this is one degree longitude between the start of that yellow turn and the end of that trajectory.
On this latitude this makes a distance of ~55miles/90km.
As you know the more north you go the wider this distance becomes.
Concluding from this the yellow trajectory ends well outside the 40miles limits of the previous search area.
Curious about your thoughts on this.
Mick, DennisW
I do not regard Higgins’ commentary as a useful contribution.
:Don
@Don Thompson
Re: “I do not regard Higgins’ commentary as a useful contribution.”
I concur.
@DrB: ” I think there has to be a simple relationship between CI and some combination of airspeed and fuel flow.”
That relation is: Cost Index = a*(M/SAR)^2*d(SAR)/dM; where –
a = speed of sound
M = Mach number
SAR = Specific Air Range (fuel mileage)
For the AERO Figure 1 it looks like this.
@Don @Mick
“I do not regard Higgins’ commentary as a useful contribution.”
I can say that about all the contributions. The aircraft has not been found, and no one has a convincing argument for where to look. Who is making a useful contribution? The SSWG with their high priority search area? The IG with their cluster of pins at 38S? How in the hell can you make a “useful contribution” criticism when you and your friends have not accomplished jack shit?
The only useful contribution has been mine which is to assert that starting an underwater search based on the data we have is stupid. I have had this issue with all IG members virtually from the get-go. Get a sense of history.
The ATSB has more and better information now than it had when it decided to launch the initial underwater search. No comment as to what has changed, and why they will not resume. The obvious answer is that the people writing checks have lost confidence in the geeks waving spreadsheets.
@DennisW
Surely a “useful contribution” is anything that factually, logically and/or rationally informs the discussion. Doing nothing (which for any proposed initiative is always the default base case) would not have been an acceptable option for any associated party; the Malaysian government, the Chinese government, Malaysia Airlines, the next-of-kin, …
And if doing nothing was sage advice at what point might doing something move up to preferred option; what exactly was going to be your threshold for commencing an underwater search? Your “useful contribution” sounds like the position currently adopted by the three governments; we’re not going to look for it until it’s found.
As to why there has been no resumption of the search, that was made very clear by the Australian Minister for Infrastructure and Transport, Darren Chester on 20 December last year;
“As agreed at the Tripartite Ministers meeting in Malaysia in July we will be suspending the search unless credible evidence is available that identifies the specific location of the aircraft.”
@DennisW
Why not look at the bright side Dennis? The search might have been unsuccessful to date, but I dare say an awful lot has been learned along the way. Surely there has to be some value in that?
@Mick
Higgins contribution is to try and pry information from your government that they are withholding from the public domain. I regard that as a useful exercise. No one else is doing it. We simply talk about it. Higgins is actually hanging it out there. Granted, I do not regard his efforts as squeaky clean.
As it stands the information we have is incomplete. Whether the additional information would yield additional insights, I have no idea, What could possibly be the point of withholding information that is clearly not classified relative to national security?
Chester’s statement simply reinforces my point. The people writing checks do not believe that the analysis is credible information. Nor do I. I agree with the decision to suspend the search. I would have never started a search.
“As agreed at the Tripartite Ministers meeting in Malaysia in July we will be suspending the search unless credible evidence is available that identifies the specific location of the aircraft.”
@Andrew
I am looking at the bright side. I am not being critical of Higgins. He is a breath of fresh air in my view.
@DennisW
I agree that the release of all the information held would be valuable. I think that the pressure on the authorities would be all the more effective if it was coming from a credible, reputable source.
@Mick
“I think that the pressure on the authorities would be all the more effective if it was coming from a credible, reputable source.”
Who would that be? Your comment implies that Higgins is neither credible nor reputable. The man has balls. I can’t say that about anyone else involved on our side of the equation.
@DennisW
To credibility;
“Although she was not asked whether she thought police should have stormed the cafe earlier, Smith’s body language suggested she thought they should have.”
Ean Higgins, Sydney Lindt cafe inquest findings may be critical of NSW police, 18 August 2016
To reputation;
““The alternative is, frankly, that despite all the evidence, the possibility that someone was at the controls of that aircraft and gliding it,” he said.”
Ean Higgins, Darren Chester seeks briefing on hunt for MH370″, 20 February 2016
What “he” (Martin Dolan) actually said was;
“We’re not at the point yet, but sooner or later we will be — and we will have to explain to governments what the alternative is, and the alternative is, frankly, that despite all the evidence as we currently have the possibility that someone was at the controls of that aircraft on the flight and gliding it becomes a more significant possibility, if we eliminate all of the current search area. NIn a few months time, if we haven’t found it, then we’ll have to be contemplating that one of the much less likely scenarios ends up being more prominent. Which is that there were control inputs into that aircraft at the end of its flight.”
In my considered opinion passing acts of mentalism off as reportage is not credible and doctoring the bejesus out of quotes is not reputable. You may demur.
@DrB: The equation I gave at 1:46 am is for use with a fuel mileage chart such as AERO Figure 1. It can be rewritten more simply as:
Cost Index = F – M*dF/dM where F=fuel flow and M=Mach number, or
Cost Index = F – V*dF/dV, where V is airspeed or groundspeed, as required.
Simple but not linear.
@Ge Rijn. Nice. Yes the yellow travels the longest distance. From the scale at the bottom (I had missed it) and a piece of string, circa 120nm, which is about the glide range the ATSB had in mind (but manned) and of course about a glide ratio of 20:1 from 6 miles high. The 1-6 others are not really much less, the minimum, without 3, being a little less than 100nm in air miles.
Since that scale is offered I take it as a useful measure in all directions despite the Google Earth caption and the chart scaling you mention.
“First it’s not clear to me if the 20 minute simulations are included in the graphic (do you know?)”. From the captions of the diagram and dot points those simulations displayed are all they address and so that includes the 20min. The others than the yellow might be shorter in distance but similar in endurance, who knows? However none of those in that category fits the final BFOs’ descent criteria. The right turners do but they are of the ‘unlikely’ configuration.
Number 3, consistent with the BFOs, finishes 15nm distant from the log-on as do right turners and some of those of course do less. As it stands their assessment that, “Simulations that experienced a descent rate consistent with the ranges and timing from the BFO analysis generally impacted the water within 15 NM of the arc”, is borne out but by just one of the ‘likely’ configuration.
From use of that scale, as distinct from your method, the yellow finishes 36nm from the log-on but if they accept only those that meet the BFO criteria to assess maximum distance AND are of likely configuration, this and all others than 3 are ruled out so their distances of impact do not count.
Oops, same mistake again. Multiply the CI by -1.
@DennisW
If you want to get a sense of how credible or reputable The Australian’s handling of matters relating to the ATSB and MH370 are look no further than Ean’s 17 April article, “ATSB shuts down details on MH370 search”; in it Ean sensationally claimed,
“The Australian Transport Safety Bureau has invoked draconian legislation in refusing to release material about its search for missing Malaysia Airlines Flight MH370, warning that any bureau employee who provides such information to the public or a court could face two years in jail.”
The ATSB responded in Greg Hood’s letter to the Editor of 28 April in which Hood specifically addresses the issue;
“Mr Higgins’ articles have also consistently attempted to create the appearance of contention between the search strategy experts and members of the ATSB’s search team where none exists. He has suggested in his articles that some ATSB officers have had second thoughts about the ATSB’s position in relation to the end of flight scenario and further;
“(the) Australian Transport Safety Bureau has invoked draconian legislation in refusing to release material about its search for missing Malaysia Airlines Flight MH370, warning that any bureau employee who provides such information to the public or a court could face two years in jail.”
I find the use of this journalistic tactic particularly objectionable. No such warning has ever been issued, every member of the search strategy group and the ATSB’s search team understands, and is in agreement with, the science associated with the search and the implications on the search area of the analysis of the satellite communication data. Members of the ATSB’s search team are operating under the standard legislation that ATSB employees normally operate under, governing the disclosure and use of information.”
What was The Australian’s response? An immediate repudiation of this shameless attack on the credibility of one of their senior journalists? No. A staunch and rousing defence of the bona fides of the article? No. Publication of an ATSB email, memo or staff circular supporting the original claim? No. Reference to an “unnamed source” within the ATSB supporting the veracity of the original claim? No.
No, instead they fold like a two dollar TV dinner table leading to the not unreasonable conclusion that Ean had made it up. And to the extent that you think that might be an unreasonable conclusion on my part, let’s look at what follows; when the paper finally got around to publishing Hood’s letter 10 days later, the section dealing with the allegations of threatening employees with two years in jail is … pffft! Gone! Miraculously expunged from the record.
That’s the sort of nonsense I take exception to.
@David
Yes, according your string-messurement those ~20 minutes simulations 100+miles trajectories must be included.
I did see the 20km scale but I assume(d) they’ve been so correct to use actual ~7th arc lat/longs (which they did) and the correct Google Earth scale and area in their projection of the trajectories.
So I made an estimate based on geological features on the see floor.
Then I find ~55 miles for the yellow trajectory.
Anyway it’s a bit strange the high descent rate conclusion is only based on one likely simulation. That looks like cherry picking the one that fits the BFO’s which renders the other simulations as a kind of negative evidence saying this only likely trajectory is not that likely at all.
@Ge Rijn: “That looks like cherry picking the one that fits the BFO’s which renders the other simulations as a kind of negative evidence saying this only likely trajectory is not that likely at all.”
Particularly if they cannot reveal the particular conditions that were simulated in that one simulation.
@Gysbreght
Yes, why not one would think. What’s the problem in reveiling those particular conditions to credible investigators outside the official bunch.
Not only for this one particular trajectory but also for the other ones.
Especially the yellow trajectory as being on the other end of the spectrum.
Not that I count myself at all to one of them (that’s obvious I suppose) but I would be particulary interested in the plane’s attitude at the end of those long trajectories especially the yellow one.
This trajectory has a quite large turn radius (+30km) in it’s last turn. I assume regarding the long glide necessary to make this trajectory, the bank-angle, AoA and speed at the end of this trajectory’s final turn could not have been high.
They just have to reveil those simulation-data to credible outside-investigators to double check their findings and remove any doubt about them.
Now again, also with this subject, they load the suspicion upon them they’ve got something to hide.
@Victor,
Returning to the subject of the blog post: the Ocean Discovery XPrize.
Is there an accepted analysis for the balance of cost in today’s operational model: greater in the operation of the surface support vessel (eg performing only MBES surveys) or in the operation of the underwater vehicle (eg support vessel plus towed or autonomous UV)?
In my attempt to slice & dice ATSB’s contracts with Fugro, the surface support segment appears to be the much larger fraction of overall cost in a deep ocean survey and search operation.
That suggests to me a transformation might be more rewarding in the surface support segment rather than the deep ocean segment.
:Don
@Victor Iannello: I thought it would be interesting to compare your Mlrc/Mmrc to that in the ATPL(A) booklet. The value of Mlrc/Mmrc for your model is at Mileage/Best Mileage 0.99 from the normalized curves you posted yesterday at 8:30 am.
@Gysbreght: So the trend of my model looks correct, although the values are not precise. The question is whether the differences reflect inaccuracy of the model, or questions about the definition of LRC, or differences between a 200ER and a 300ER. I don’t know the answer.
@Don Thompson: Somewhere I saw the cost per day of the surface vessel, and it was quite significant. My recollection is that it was about $40,000 per day. Why I thought the drones were interesting was because of the opportunity to drastically reduce the cost of the surface vessel per scanning sensor since multiple drones (each with their own sensors) could be deployed from a single surface vessel.
@Ge Rijn: After reviewing the simulations at fuel exhaustion from Boeing, from Mike E., and from many FSX simulations I’ve run, I broadly conclude the following:
1. With no pilot input and shallow bank angles, it is not likely that the descent rates suggested by the BFO can be achieved.
2. With no pilot input, it is possible that large bank angles may develop that result in high descent rates, especially when the control law is either in secondary or direct modes, which would be expected after flameout. Contributors to the high bank angles are:
a) residual out-of-trim of the rudder, and/or
b) lateral asymmetry in the configuration or position of flight control surfaces
3. High descent rates consistent with the BFO values are also explained with “nose down” pilot input.
I think either (2a), (2b), or (3) is likely.
@Victor,
Phase 4 of the AF447 search took the multiple sensor approach, RV Alucia embarked with three Remus 6000. That phase was successful after 15 days (and after the Metron & other analyses). Its €4m cost isn’t broken down for shore segment vs surface vs deep ocean resources.
Of course AUV ops are limited by the need to launch and recover the AUV for servicing between every mission which implies operating limits due to sea state. If the recharge, data offload, and mission navigation update could be performed in an underwater cradle (contactless) the surface (or semisubmersible) support vessel capability might be redefined.
:Don
@Victor Iannello: I agree broadly with your simulation conclusion, but would add that in my opinion only a “nose-down” pilot input will produce rates-of-descent compatible with the BFO’s close to the 7th arc at normal cruise altitudes.
@Don: In the case of AF447, the scan rate was not as important, so an ROV approach made more sense. To improve the “duty factor” of the team of AUVs, one possibility is to have available more AUVs than is used by the “active team” at any time and stagger the time for between mission servicing. Developing and implementing the associated logistics is easier said than done, but if the cost per AUV is reasonable, this option enters the realm of reality.
@VictorI
Only considering the 10 Boeing simulations I generally conclude the same as you only with another view I guess:
1: “With no pilot input and shallow bank angles, it is not likely that the descent rates suggested by the BFO can be achieved”.
7 out of 10 Boeing simulations did not achieve the suggested BFO’s and the 2 other (right turns) are not likely. Only one was likely which only provides negative evidence in the context of the rest of the simulations.
2a: Unless the residual out of trim rudder is known for 9M-MRO. I mean also the specific overall inherant deviation in a straight line of the airframe that requires compensation.
b: lateral assymetry seems to me could be compensated by the horizontal stabiliser under RAT while only the left horizontal stabiliser will be actuated under RAT when only the right wing flaperon is actuated in neutral.
3: A nose down impact would not have left a nose gear door without compression damage and a clean break from its hinges.
@Ge Rijn: “Only one was likely which only provides negative evidence in the context of the rest of the simulations.”
The number of simulations that produced shallow descent rates is irrelevant. What we need to know is which configuration(s) produced high descent rates, and then determine the likelihood that the plane was configured in that way. We don’t have that information.
@Victor
Fugro Equator worked on the 7th arc bathymetry survey for 187 days covering ‘over 175,000km²’. The contract value was AU$13,200,000 = AU$70,588 per day.
In a separate contract, Fugro provided an AUV & services to another Aus gov entity for AU$9,868 daily rate.
From the overall cost of the Fugro deep ocean seafloor search, my estimate is AU$79,500 per day, per vessel incl underwater vehicle.
@Don Thompson: So AU$79,500 per day is about US$60,000 per day. My recollection of $40,000 might be a bit low.
@Victor
A cursory scan of charter rates for AHTS vessels, e.g. Go Phoenix, shows it’s a very volatile market with low demand. Research vessels no doubt a different proposition.
The operating cost of a Poseidon P-8 aircraft with a nine man crew is $4200/hr (consistent answer found by Googling). That would come out to ~$100k per 24 hours of operation.
Still, what is the point of the cost? The Aussies have been emphatic about cost not being a factor in canceling the search.
@Victor. “With no pilot input and shallow bank angles, it is not likely that the descent rates suggested by the BFO can be achieved.”
You might have seen a question I put to ALSM, 18th May, 0818 as to why the ATSB model assumed OCXO drift of 130Hz for a 1 minute SDU off time. That greatly increased the maximum descent rate permitted by BFO compatibility.
Dot point 4, p14 of:
https://www.atsb.gov.au/media/5771939/ae-2014-054_mh370-search-and-debris-update_2nov-2016_v2.pdfIn teh mos
says,”Some of the simulated scenarios recorded descent rates that equalled or exceeded values derived from the final SATCOM transmission. Similarly, the increase in descent rates across an 8 second period (as per the two final BFO values) equalled or exceeded those derived from the SATCOM transmissions. Some simulated scenarios also recorded descent rates that were
outside the aircraft’s certified flight envelope”.
To my mind exceeding the descent rates disqualifies compliance with BFO criteria as much as not reaching. That goes for the 10,800fpm between the two descents but also the maximum descent rate.
If the rate resulting from the 130Hz addition is too high, the upper limit of descent rate accepatability is too high.
Do you or other OCXO afficionados have comment please?
As a related issue you say,”The number of simulations that produced shallow descent rates is irrelevant”. If the number of shallows is proportionately significant when actual aircraft circumstances can extend beyond the constraints needed to replicate high rate simulations, then I think they are significant.
Judging by no ATSB comment being on offer to that effect, that the aircraft configuration etc would have led to BFO compliance, I suspect that remains the case. Quite possibly relevant, there is evidence from dot point 4 that the demonstration of BFO compliance in those simulations was hardly rigorous.
Re OCXO…
At 1825, there was no BFO “offset” or “drift” as those terms are normally used by engineers. There was a “warm-up transient”, caused by the PID thermal control system overshoot and decay. The response was obvious to the trained eye. It was normal and to be expected for ANY OCXO starting cold. I shared the observation with Bobby and he did an excellent job of fitting the observations to the expected critically damped control system response. So we can be reasonably sure the OCXO was off for some time prior to the 1825 logon. At 0019, the OCXO had only a 1 minute power off period, resulting in nearly no cooling and thus a rapid stabilization, probably preceding all the other SDU BITE checks. Thus, there was little or no BFO bias error at 0019.
Re ATSB analysis…
ATSB assumes, as I do, that the most likely BTO bias errors at 0019 were very small compared to the “vertical signal”, thus not a significant factor in the finding that the descent rate was high. The nominal 0019 BFO values are probably the actual values give or take a few Hz. The ATSB worst case analysis, using offsets at 1825, was in no way intended to suggest there was any similarity in the conditions at 1825 and 0019, or that those 1825 offset values were likely, or even possible, at 0019. Rather, it was an attempt to show that even if the error was as bad as it was at 1825 then you are still left with the undeniable conclusion that there was a rapid descent. I would not have used the 1825 values for this WC analysis because they are not really relevant to 0019, but I understand the ATSB logic. They were simply trying to explain the core point that no matter what you assume for a worst case bias error, you still get a strong vertical signal headed down fast.
@ALSM Thanks. If you were asessing whether simulations met the descent criteria including timing, of the 2 BFOs, what limits and tolerances up, down and in time would you apply? In other words how would you know that the actual BFO descent could be simulated and that that was likely?
David:
We were not “…assessing whether simulations met the descent criteria…”. We simulated what happens when the plane runs out of gas. It is that simple. It always develops a turn and usually phugoids are superimposed. The turns get steeper and steeper, and the rate of descent increases, decreases, increases again…. Very chaotic. We have been over this a hundred times.
Mike
@Victor,
Victor, you could add minor airframe damage occasioning laterally asymmetric drag and unconscious control inputs to your list of factors potentially giving rise to high rates of descent after the autopilot disengages.
With regards to the latter, it takes a reasonable amount of force to override the control column and control wheel autopilot backdrive actuators and force the autopilot to disengage, 50 pounds and 27 pounds respectively. Accordingly, if an unconscious or dead crew member’s arm was resting in the control wheel or a body was slumped against the control column and exerting less than the requisite disengagement forces, the autopilot would remain engaged until the loss of power. However, as soon as the autopilot disengaged the weight impinging upon the control wheel, control column or both would cause them to move. A malicious actor that did not intend to pilot the airplane all the way to end-of-flight but who wanted to ensure a high speed water entry might take advantage of the foregoing by simply hanging a bag with 15 – 20 pounds in it off a control wheel.
@Mick: Thanks for your input.
@ALSM. Thank you. I do understand what you were doing to progress the understanding at that time. However what am seeking now from you is your opinion on what would be reasonable criteria by which the ATSB could say that the simulations demonstrated that realisation of the 2 BFO criteria were likely.
I point to Table 2 and ask whether an appreciable proportion of simulations should meet fall within those descent rate criteria and also should observe the 10,800+fpm difference between, in 8 secs. Also, as to the last which has no band, what should be the limits? I have put a view (ie rounding only) though I think you could justify wider. Then when it comes to the timing during which the requisite descents occur, what should be the boundaries? If for example in the simulation the descent rate rises to match the first BFO’s at 3 mins after AC loss, the rate increases to match the second and within the 8 secs, does that give that simulation a tick when that sequence should have started a minute earlier?
I might appear pedantic here but the ATSB generality in taking it that the simulations demonstrate accord with the BFOs, when so far as I know all there have just been some examples of high descents and high accelerations, do not seem to match. Their apparent inclusion as conforming for anything which exceeds the predicted rate or rate increase is a clue, to me, of a qualitative measure only, in which case Table 2 is instructive and their judgement is the only guide.
Add to that lack of any comment on timing and connectivity with the satellite.
It is not that the outcome is unsatisfactory. It is that it is not as demonstrably so as it could be, non-disclosure nothwithstanding.
@David: To reach the high descent rates implied by the BFO values without pilot input requires bank. In FSX, I can simulate the flameout, the degradation of the flight controls to secondary law, and the ensuing flight dynamics. (I think the aerodynamics are reasonably accurate for M<0.87 and far from stall.) With the proper amount of rudder out-of-trim, for instance, the plane begins to roll slowly until it exceeds about 30 deg, and then the descent rate increases at about 1,000 fpm per minute, similar to what the BFO values imply. More out-of-trim, and it takes less time to reach the high descent rates. Less out-of-trim, and the plane can glide for 10 minutes or more before it crashes. By performing limiting cases, it is not necessary to closely match the descent rates to conclude that the BFO values are correctly interpreted and the plane was in a steep descent. Boeing certainly understands this, and felt confident in making the statements they did, even if the "BFO compliance" for any single simulation was not "rigorous".
@Victor. Thank you for that. If as you say the ATSB/Boeing can see what level of trim would be needed to replicate the BFOs and their timing; and that trim condition could well have been encountered by MH370 within vagaries of speed and altitude etc, then that would support their BFO compliance conclusion as you say. At 30deg I take it you mean a 10,000fpm rate increase and over 8secs? Otherwise, with an 1000fpm/minute increase starting at conclusion of a slow roll rate period it would not make the first BFO minimum of 3,800fpm after a total of just 2 mins.
You are suggesting there was a steep bank indeed though and unless slipping that would entail a high turn rate, which I suppose you encountered. That is not evident in their 3rd simulation, so there might be some other ingredient at play there. There may be earlier Boeing simulations (the ATSB did make a point of the difference between that and others) with that very high bank and turn though and I hope that establishing and holding satellite connection would have been seen as an ingredient.
Do you have a clue as to the ‘g’ you encountered?
@Victor. I disambiguate. Second para, “…(the ASTB did make a point of the difference between THEIR SIMULATOR and others)…”
@DennisW
My discussion of search & vessel costs relates to the Ocean Discovery Xprize: which element of current deep sea search capability would one target for optimisation: the undersea vehicles, the surface vessels, other…
If the UUVs could be air deployed & cheap enough to be deemed disposable, like a sonobuoy or SLDMB, then a P-8A or any MPA could factor in the solution.
:Don
@Don
Yes, I think I understand. I intended that the P-8 operating cost could be regarded as an upper bound which I should have stated explicitly. I think your numbers are in the ballpark.
@VictorI
Sorry but I just don’t get this:
“The number of simulations that produced shallow descent rates is irrelevant. What we need to know is which configuration(s) produced high descent rates, and then determine the likelihood that the plane was configured in that way. We don’t have that information”.
7 out of 10 Boeing simulations produced shallow descent rates and 2 were unlikely right turns. Only one likely trajectory produced the high descent rates that could fit the BFO’s.
And then those 7 shallow descents are not relevant??
Are you denying the the obvious statistic reality of those 10 simulations?
Regarding this remarkable majority of shallow descent rates I would rather opt to look for configurations that produced them also including the planes attitude when entering the water.
Not to just ignore them as you seem to suggest.
@Ge Rijn: I never suggested to ignore any simulation. What I said is we don’t know what the probability of each configuration is, so the number that produced a shallow descent rate is irrelevant. There is no “obvious statistic reality”. You are making an assumption that each configuration has about equal likelihood. That assumption could be very wrong. What we do know is the BFO values at 00:19 suggest a high descent rate.
@Ge Rijn:
Simulations 3, 7, 8, 9, 10 ended with high descent rates.
Simulations 7, 8, 10 were identified as being in an unlikely electrical configuration.
However, trajectories 9 and 10 are virtually identical. When I asked the ATSB to confirm or deny that #9 is in the same electrical configuration as 7, 8, 10 there was no reply.
I should have added that 7,8,9,10 have trajectories that are markedly different from 1,2,3,4,5,6. Trajectories 7,8 turn left and 9,10 turn right.
@Victor Iannello: “You are making an assumption that each configuration has about equal likelihood. That assumption could be very wrong.”
The ATSB writes:
“Reasonable values were selected for the aircraft’s speed, fuel, electrical configuration and altitude, along with the turbulence level.”
Equal likelihood would seem to be a reasonable assumption.
Since #3 is somewhat unique in the group 1,2,3,4,5,6 it is somewhat less likely than its neighbors.
@VictorI @Gysbreght
I can see the importance of getting information about simulations that could fit the 0:19 BFO’s.
What bothers me a bit is the seeming reluctance to look in another direction within the IG for a long time now. It seems all settled: high speed descent, high speed impact.
Focus seems to be on those final BFO’s with a tendency to ignore a lot of other possibilities shown by the simulations and other indications.
It looks like another case of confirmation-biased reasoning to me sometimes.
Discussion is about different points of view with the goal to come to solutions isn’t it.
I know everyone will agree on this but letting go of confirmation bias seems a long way due sometimes.
I’m not free of it also ofcourse but I won’t call facts irrelevant if they don’t suite my propositions.
@Ge Rijn
“Focus seems to be on those final BFO’s with a tendency to ignore a lot of other possibilities shown by the simulations and other indications.”
I am unable to attribute the final BFO values to anything but a high rate of descent. If there are other possibilities that you know of, please post. I suppose one could assume the AES was going haywire, but that seems like a stretch to me.
@DennisW: If there was a high rate of descent at 00:19 there must have been a pilot input to cause it. That opens a wide range of possibilities as to what happened after 00:19.
@Gysbreght
I do believe in a conscious pilot to the end, however, the high rate of descent seems like a deliberate action with the intention to continue into the ocean. As you say, maybe not.
@Ge Rijn: As I said, my comment about irrelevance referred to calculating probabilities of certain scenarios, not the investigative relevance of the simulations. Nor did I mean to imply your opinions were irrelevant.
If you truly believe the BFO data and the debris indicate there was a shallow descent, what does it matter what I, the IG, or anybody else thinks? I advise you to not get upset if others disagree.
As Dennis said, we look forward to hearing your theories about the BFO.
@DennisW: Or it could be an appropriate recovery from an inadvertant low speed condition to avoid stall. A pilot can increase the rate of descent 10,000 fpm in 8 seconds, and just as easily reduce it by the same amount in the same time.
@DennisW
I’m also unable to attribute the final BFO’s to anything else but a high rate of descent. But if those are not supported by most of the Boeing simulations I question their validity. And I won’t conclude anyway there must have been a high speed impact.
The BFO’s show a snap-shot of only 8 seconds. At which altitude this snap-shot occured is not known. It’s also not known at which altitude the plane was at the second engine flame out nore at which altitude after the 8 seconds.
So therefore it’s not known in which attitude the plane reached the ocean surface.
@Gysbreght’s interpretation appeals to me more then.
Maybe an ’emergency 10.000ft/m descent’ after second engine flame-out. Leveling out on lower altitude. Who knows..
Found a video also with instructions to cabin crew and passengers (confirmition on @Andrew’s comments by the way):
https://www.youtube.com/watch?v=4EaQLca9rJY&feature=youtu.be&t=1m21s
@all,
I have some new (confidential) information on the LRC Fuel Flow and Mach for B777-200ER with Trent 892B engines. This covers altitudes from 6,000 – 43,000 feet generally every 1,000 feet. The weights are in 10 MT increments rather than 20.
The most useful aspect is that now we can evaluate LRC (and ECON) fuel consumption at altitudes below FL250 with some accuracy.
In addition, the footnotes are included, notably the one regarding 3% FF increase per 10C increase in TAT. This confirms my original conclusions that this factor must apply to the Trent engine and that its value would be the same as used for the GE engine.
I have put these tables into EXCEL, and they can be downloaded here:
https://drive.google.com/file/d/0BzOIIFNlx2aUdVZMVmpKVE9kT2M/view?usp=sharing
For completeness, I also included the Holding FF and KIAS tables. The new data match the old open-source tables.
@Gysbreght
I did not read your comment before adding mine. Striking similarity that’s why I put this one on.
@VictorI
What you and the IG think matters a lot more than what I think regarding your influence on the case. You are the ones that go to the news-lines not me.
And don’t worry. I’ve been around for around 3 years now. When I get a bit upset I’m sure I can handle it now between the die-hard cracks around here.
@Ge Rijn
That video linked to some other interesting B777 videos on YouTube.
One video was Delta 95 B777 actual emergency landing due to apparent smokey tire upon take-off. You can see them jettison the fuel. I think it says they are landing (smoothly) with roughly 220,000-lbs of fuel and 308 PAX. If so, that would be approx. two(2) times MH370 take-off fuel load.
@ALSM
Correct me if wrong:
The GO PHOENIX search area map, and Richard Cole’s Furgo search area map, seem to show the area 32-34S “inside Arc7” had not been searched. Especially if we show the “Zero” altitude Arc7 position, the 32-34S “inside Arc7” area is (assuming at least small error bars on Arc7) is possibly not searched at all.
Slightly less definitive, but the debris drift studies probably suggest it is the (unsearched) area inside Arc7 at 32-34S that we would be most likely to find the MH370 crash site.
The bottom line is that the search to date has unfortunately skipped a key search area, as has been noted already by ATSB.
@Victor: re: sims from Boeing, Mike E., and your own FSX:
Have you posted all detailed data from all those three sources?
If not: would you be so kind as to post it for us?
Huge thanks in advance.
@Brock McEwen: I don’t have the detailed data from three sources. For the Boeing simulations, I know as much as has been included in the ATSB reports. For the simulations from Mike E., there was a compilation made of one of the simulation runs. As it would take me some time to find it, others might be able to access that easier than me. As for the FSX PMDG 777 runs, I was mostly interested in seeing what conditions would cause a steep descent based on the aerodynamics, which I did determine. However, I was only able to get reasonable results because I “tweaked” the simulation by forcing the PFCs to operate in “secondary law” and by introducing rudder and aileron trim. At this point, I don’t have any FSX PMDG 777 data that is worthy to share except for the qualitative observations I have already shared. I might share a representative video at some point.
@Victor Iannello: “For the simulations from Mike E., there was a compilation made of one of the simulation runs. As it would take me some time to find it, others might be able to access that easier than me. “
It is here.
@Victor. About those BFO descent rates and quantitative criteria, this is a summary and conclusion as I see it.
The ATSB at p11 Table 2 of:
.. https://www.atsb.gov.au/media/5771939/ae-2014-054_mh370-search-and-debris-update_2nov-2016_v2.pdf
has described maximum and minimum limits of descent rates which could result in the BFOs, the maximum including an OCXO drift possibility of 130Hz. I have queried the inclusion of that. Indeed at the earlier Figure 4, p10, where warm-up OCXO drift is depicted, the description adds, “if the power interruption was sufficiently short, the OCXO drift could be negligible”. ALSM has confirmed that for the 00:19 BFOs, so the 130Hz should not have been included in that Table 2, that is in the, “00:19:29 log-on request” and “00:19:37 log-on ACK” high descent rate listings.
That leaves the maximum limiting descent rate for the two BFOs at those in my post to ALSM, namely, “the maximum descent rates would be 6,500fpm (was 14,200) from 111Hz (at 00:19:29) and 17,350fpm (was 25,000) from 295Hz (at 00:19:37). The 10,800 odd fpm difference remains.”
I recall that DrB calculated a lesser descent rate, though I do not recall the details.
There are two consequences I mention. The first is to my quantitative estimation of the pitch down by which the aircraft could realise the BFO-compliant descent rates, by combination of acceleration and more vertical speed-vector direction. The extent of that nose down needed to meet the maximum descent rates boundary is now considerably less. Second, those reduced descent rates would alter matching of Boeing simulator descent rates, rates increase and timings with those calculated from the actual BFOs.
As to the latter, you and ALSM are of the view I think that the ATSB assessment has not been quantitative, nor need it be, but instead it has made a general judgement as to whether Boeing’s descents show that these BFOs could well have been realised. The basis for this approach is acceptance that the current ATSB interpretation of the final BFOs is valid. It follows from that that it is reasonable to discard simulations which are non-compliant with BFO estimated descents. The intent then is not to prove the BFOs valid but to establish that there are likely circumstances and configurations which can satisfy them.
However, taking drift analysis as an analogy, the purpose of that has developed from establishing that a flotsam arrival was “consistent” with a purported crash zone to one aimed at refining its origin. A like change to simulation analysis could establish more precisely what the above circumstances and configurations were and tighten search width more yet. That would entail a more quantitative approach and where the new limits I mention could be applied.
“In addition, the footnotes are included, notably the one regarding 3% FF increase per 10C increase in TAT. This confirms my original conclusions that this factor must apply to the Trent engine and that its value would be the same as used for the GE engine.”
I agree that the fuel correction factors for ambient temperature would be similar for Trent en GE engines.
However, firstly, the value rounded to 3% could be anywhere between 2.5% and 3.5%.
Secondly it depends on the theta exponent for fuel flow. That is an empirical value that can differ slightly between engine types. It is also affected by the amount of bleedair taken from the engine for aircraft services which varies with altitude.
Thirdly it varies with altitude and Mach number. For the PW2040 engine installed in the B757 that is used as an example in Jet Transport Performance Methods, theta exponent is 0.61 and the fuel flow correction factor at FL350 and M.84 is then 2.4% per 10°C increase in TAT.
P.S.
Jet Transport Performance Methods estimates the drag increase due to Reynolds number for the B757 airplane as 0.4% per 10°C increase in TAT.
Added to the increase of TSFC the fuel flow correction factor at FL350 and M.84 is then 2.8% per 10°C increase in TAT for the B757 with PW2040 engines.
Regarding the FMT, the pre/till 18:22 radar data, the 18:25 log-on sequence, the BFO’s and the 18:40 phone-call there are aspects that don’t fit the official conclusions/assumptions in my logic.
According a Duncan Steel’s article (which I show next) a rather steep descent has already set in after 18:13 till 18:22 according radar-data:
http://www.duncansteel.com/archives/2695
Then I argue the plane could have still been in a descent or at least at far lower altitude by 18:40 when the call was made.
I think it would be an extraordinary coïncidence when this call was made excactly at the time the plane performed a turn.
I regard this is highly unlikely.
I think while being descending since 18:13 it probably still did at 18:40. While progressing in a (maybe climbing afterwards) straight line to LAGOG waypoint and make the FMT around there.
@Ge Rijn
That is an excellent paper by Richard Godfrey. He is probably correct that Victor’s 180S CMH path from BEDAX is close to the truth (maybe it was constant speed at 180S CTH).
The (top secret) radar data out to 1822 is not really able to give us reliable altitude data, Victor even questions if there is any valid radar data after 1802.
I would probably say the 1941 BFO=111 is not consistent with an 1840 FMT. If there was a FMT at 1840 then 1941 BFO would be closer to 101, unless you invoke a big slow down, which Brain Anderson’s speed calculation suggests is not true (speed is fast). Therefore MH37O pilot was probably messing with headings, altitude and speed at least until a little after 1941.
@TBill: Brain Anderson’s speed calculation requires constant track and speed. It is not valid if the MH37O pilot was messing with headings and speed at least before and after 1941.
@TBill
I don’t know where Richard Godfrey got his radar numbers from till 18:22. They seem quite specific to me so I assume they have to come from an official/reliable source.
A ~18:40 FMT I would consider quite unlikely if it’s only based on the phone-call. The coincidence that call exactly interfered with a turn at that time is just too slim to have any value against more logical scenarios IMO.
@David, @Victor and others
Regarding the BFO values during rapid descent, have you considered the impact of the aircraft on the OCXO ? These oscillators are sensitive to lots of things, not just temperature. This paper from an employee of Vectron that makes such products goes into some depth on the subject. In particular vibration. Page 4 gives typical values for different vehicles. It also points out resonance can amplify the effects. Now I’ve no idea what sensitivity the real equipment in question has, but I can see G force and vibration during uncontrolled descent being very different to normal operating conditions. There is no mention of this in the ATSB paper. It could be a contributory factor in the BFO value.
https://www.vectron.com/products/g_sensitivity/Vig-tutorial%20on%20g-sensitivity.pdf
@All
Apologies the author of the above paper (John Vig) was at the time of publication (2004) in the US Army CECOM despite the paper being on the Vectron web site.
@JohnM: At 18:40, there were 49 recorded values over the course of about 1 minute. The measured data varied by +/- 2 Hz. If the frequency was significantly changing due to vibration, you would expect to see a larger variation.
@Victor
Understood, my fault for not being clear, but I was referring to the final two bursts – one of which is the last before the aircraft hit the sea. Does the raw data show the same consistency then ?
JohnM: The paper cites a frequency shift of about Δf/f = 1e-9 per g. For the AES frequency of 1.6e9 Hz, that’s a shift of 1.6 Hz/g. I doubt that would account for more than several Hz of shift.
@Victor. I agree with your maths, but what quality was the one in the SDU? Vectron are clearly in the business of selling oscillators for hi-vibe environments. Also the paper goes through all sorts of secondary effects, the result of which is not immediately clear from a very quick scan of the paper. I’m wondering if I can find some comparable kit to experiment on. I’m going to do some more research.
@JohnM. As to vibration/shock, specs for mounts and compensation followed by bench proving and in flight tests should give confidence of nothing undue within the flight envelope. Transients from landings would be of no import short of damage.
Beyond the envelope there is the potential for flutter and bits separating causing excess transients but that should be beyond log-on speeds; and any shock from RAT deployment should be before the log-on.
@Ge Rijn
I agree that 18:40 FMT appears unlikely. Probably there was pilot input or programming at least another hour or so before a turn south. A more correct aircraft-finding assumption probably would have been level flight/no turns after 20:41 (Arc3).
@TBill: The current recommendation for the search is an area deemed to be low probability by the DSTG Bayesian model. It appears that the recommended search area is only based on the 7th arc, drift models, end-of-flight dynamics, and the aerial search.
It would be interesting to know which of the assumptions that went into the DSTG model is now believed by the ATSB to be incorrect. After spending $226 million on a search, you’d like to know why it failed.
@VictorI
Your question above was directed at TBill, but I will provide my own take on it. As TBill reminded me on another thread, talk is cheap. So this is a freebie.
The biggest single factor leading to the search failure was how the question to conduct the search was posed. Two versions below. Version 1> was used.
1> The ATSB is going to start an underwater search. Where is the best place to look?
2> Does the data we have regarding the flight support the start of an underwater search?
The DSTG and IG answered 1> absolutely correctly. Unfortunately question 2> was the proper question, and the answer was/is NO.
There is no problem with the analytics. The problem is what the data is capable of telling us – radar data, ISAT data, and drift data (late in the game).
The failure is clearly in the administrative domain, and is plain and simple – asking the wrong question.
@VictorI,
You said: “@TBill: The current recommendation for the search is an area deemed to be low probability by the DSTG Bayesian model. It appears that the recommended search area is only based on the 7th arc, drift models, end-of-flight dynamics, and the aerial search. It would be interesting to know which of the assumptions that went into the DSTG model is now believed by the ATSB to be incorrect. After spending $226 million on a search, you’d like to know why it failed.”
The reason for this diminution in probability is primarily the assumption used by DSTG that the fuel was (equally) adequate for all arc locations. I doubt DSTG made that assumption on their own. More likely that’s what they were told by ATSB (and it made the analysis simpler). Other assumptions DSTG made may also turn out to be incorrect, but we already know that the aircraft could not possibly reach further south than ~35.5S. If you remove that part of the DSTG distribution due to the provably false fuel assumption, which as I recall amounts to about two-thirds of the integrated probability, the remaining probability distribution includes the proposed new search area at significant probability. Therefore I believe you could add the (revised) DSTG probability distribution to the list of factors on which the new search area was evaluated.
Is it possible that Boeing did not do a thorough fuel analysis, but instead simply provided best-case and worst-case range estimates?
You may recall that only ranges at high cruise altitudes were presented by ATSB. I think this may be a biased assumption that may also turn out to be incorrect. I have asked ATSB if Boeing’s range calculations took into account the elevated temperature. They would not answer the question, although they did confirm to me this year that 3%/10C TAT should be (or maybe “should have been”?) applied.
@Victor
@TBill
@DennisW
Re: “It would be interesting to know which of the assumptions that went into the DSTG model is now believed by the ATSB to be incorrect.”
Victor, I think that is answered, in part at least, in the First Principles Review (pp. 15-16);
“Search results and likely aircraft autopilot mode
The DST Group analysis which used a dynamic model of the aircraft, examined five different lateral aircraft autopilot control modes:
1. Constant magnetic heading (CMH).
2. Constant true heading (CTH).
3. Constant magnetic track (CMT).
4. Constant true track (CTT).
5. Lateral navigation (LNAV).
The initial analysis assumed equal likelihood (uniform probability) across the lateral aircraft modes throughout the southern portion of the accident flight. The results of the analysis suggested that based on the satellite data, the aircraft was more likely to be operating at the time in the CTT or LNAV lateral control modes. These lateral control modes generally resulted in the aircraft’s flight path being towards the southern end of the search area. The remaining three lateral control modes generally resulted in the aircraft’s flight path ending north of the CTT and LNAV results, up to 33°S in latitude along the 7th arc.
…
Flight crew advice
During the First Principles Review meeting, flight crew with extensive experience on the aircraft type indicated that the aircraft is usually flown in the LNAV or CMH lateral control modes.
This information, combined with the results from the areas already searched to increases the likelihood in the northern section of the probability map in Figure 13.”
In sum, they used the wrong navigation control mode!
Dennis,
This is where you can legitimately sink the slipper fair up the collective quoits of the ATSB, the SSWG and the DSTG. I was simply astounded when I read that section of the First Principles Review. It is beyond comprehension that they did not consult with at least one qualified B777 flight crew member about the most likely modes of navigation; hell, they could have googled the answer. It seems to be a case of scientists seeking scientifically elegant solutions without any consideration of “real world” factors. And it raises the question of who reviewed the DSTG’s working assumptions and how rigorously were they tested.
Regarding your two questions; enough data to act? where to look? As I’ve opined before I don’t think that doing nothing was ever going to be a palatable option for any of the major players. Hence, while looking at the threshold required for action was logically the correct matter to deal with first, I think that it was rendered moot by the reality of the politics. Besides, what was the threshold for action going to be? Waiting for debris to start coming ashore? For how long? How many pieces? How many different beaching locations? Something else?
My real gripe with the ATSB is that they were inflexible and/or demonstrated poor programme management. Something along the lines of the First Principles Review should have been held 12 months after the underwater search started – they had completed the high probability areas at that point – and repeated 6 monthly thereafter. By April 2016 they had 5-7 debris items found across a reasonable spread of locations; that should have been used to inform the question, why haven’t we found it yet?
I’m sort of reminded of Keynes’, “When my information changes, I alter my conclusions. What do you do, sir?”
All’s crystal clear in retrospect.
@Mick
Nice post.
@VictorI
re: assumptions
(1) I am thinking only the 20:41 to 22:41 period might be strictly constant speed, direction, and altitude. If it was a ghost flight, not so until after 19:41.
(2) Let’s face it, unintended ghost flight is counter-indicated by the fact that CTH setting buffeted by winds seems to be give the best fit to the data. So they are really struggling with the assumption that the flight south was not pre-programmed somehow.
@DennisW
Thank you, Dennis.
@Mick. “It is beyond comprehension that they did not consult with at least one qualified B777 flight crew member about the most likely modes of navigation.”
Isn’t there a likelihood they did? I would think that the assumption of no active pilot would have been part at least of the neutrality on mode selection. For all we know that approach might well have been seen as logical and even supported by appropriately experienced pilots at the time. Also, the weighting towards the south based on interpretation of satellite data again might well have been appropriate, based on what they knew at the time.
I much doubt that this was all done in some sort of academic isolation, noting the sources the SSWG was drawn from.
There were three things altering the weightings towards the north which included not finding the aircraft and the then new drift analysis results, though the way it reads is that it resulted from it suddenly occurring to them that pilots might have a view.
Where I find a gap in the logic is that ‘they’, including pilots, should now allocate more weight to the piloted preference when the presumption still is that there was none. The selection of mode might not be with the normal intent nor was whoever did it necessarily experienced with the system.
However I still reckon it better to concentrate on what next.
@David
David, I think that the ATSB’s choice of wording strongly suggests that the DSTG (as opposed to the SSWG) did not consult with B777 flight crew in determining the most likely navigation mode.
“During the First Principles Review meeting, flight crew with extensive experience on the aircraft type indicated that the aircraft is usually flown in the LNAV or CMH lateral control modes.
This information, combined with the results from the areas already searched to increases the likelihood in the northern section of the probability map in Figure 13.”
They seem to be treating the fact that “the aircraft is usually flown in the LNAV or CMH lateral control modes” as new information. And it’s probably worth looking at who wasn’t on the SSWG – the airline is not represented – and given the sort of analysis being undertaken you’d have to wonder whether there was a B777 pilot anywhere amongst the assembled crew from Boeing, the AAIB or the NTSB; I’m beginning to doubt it.
@Mick. It could be just as you say though the SSWG would have had to bless, if not actively participate in, the Review and the recommendation of the new search area, that being the centrepiece of their function. You will recall the brouhahha when the SSWG was reported as having concurred with the new search yet not all its members had yet.
If they did not have pilots in their party they sure would have been remiss and amateurish had they overlooked their major contribution.
@Mick Gilbert: “given the sort of analysis being undertaken you’d have to wonder whether there was a B777 pilot anywhere amongst the assembled crew from Boeing, the AAIB or the NTSB; I’m beginning to doubt it.”
B777 pilots are expensive. Pilots in general don’t like to sit in boffin committees, they rather go flying.
@David
Yes, so given the blessing/review process, I go back to “… the question of who reviewed the DSTG’s working assumptions and how rigorously were they tested.
@Mick. If the SSWG did not do so, their title is a misnomer. Implicitly they have at least a monitoring if not a supervisory and leadership role.
But yes that is just the theory of it.
However I think most likely the wording is the problem.
@Gysbreght
Re “B777 pilots are expensive. Pilots in general don’t like to sit in boffin committees, they rather go flying.”
Most experts are expensive and most don’t appreciate having their time wasted. None of that is an excuse for not consulting with appropriate people at the appropriate time in the appropriate fashion. Moreover, I doubt whether there would be too many pilots who would not see their professional contribution to resolving safety concerns or contributing to an accident investigation as anything other than time well spent.
@Mick Gilbert: Perhaps the ‘expensive’ experts were not sitting either in the SSWG.
@Mick Gilbert said, “In sum, they used the wrong navigation control mode!”
I don’t think it’s that simple. Suppose the path reconstruction model only allowed LNAV and CMH modes, both which the ATSB say is likely. There still is a better fit with the satellite data for LNAV than CMH paths, so the results would still be biased towards the south (ignoring fuel limitations, which the DSTG essentially did except in a very general way). It is only by considering the PDFs for each mode separately that you would see significant probabilities to the north of what was searched.
If a competent and coherent pilot was intentionally flying into the SIO, the only navigational mode that makes sense is LNAV.
I maintain that there is no good reason to exclude an ongoing descent at 18:40 combined with a loiter. That would also move the crash point to the north, regardless of navigational mode. I think that assumptions about how the plane was flown between 18:28 and 19:41 was one of the more fundamental flaws of the DSTG analysis.
So the recommended search area is now centered on 35S. Unfortunately, this area is contraindicated by the early arrival of “Roy” in South Africa, the lack of recovered debris in Western Australia, and the failure of the underwater search close to the 7th arc (+/- 20 NM) at this latitude. On the other hand, points further north, such as the 30S terminus estimated by Richard Godfrey, are contraindicated by the aerial search failing to find floating debris. (Some have begun to challenge whether floating debris near this latitude could have been missed by the aerial search.)
It’s no wonder that the ATSB has not received funding to resume the search near 35S, or anywhere else. Any claims of confidence based on the evidence at hand is unfounded, in my opinion. Just look at how many times in the past that confident predictions were proven wrong.
@Mick Gilbert
You said:”The remaining three lateral control modes generally resulted in the aircraft’s flight path ending north of the CTT and LNAV results, up to 33°S in latitude along the 7th arc.”
This is bottom line for me – I think the aircraft could be at ~33S +-1 degree. Makes perfect sense, as per ALSM above, inside Arc7 has not been searched there.
@VictorI
“…If a competent and coherent pilot was intentionally flying into the SIO, the only navigational mode that makes sense is LNAV.”
>>I have no problem with LNAV, after all, I can pick even a fractional lat/long target, but why limit to LNAV? I agree the PSS777 sim study may have used LNAV to get around Indonesian radar space for a Jeddah flight case, but I am open to straight south flight plan in any mode. If I could ever catch up to the physicists in satellite/earth geometry, I am thinking we might be able show the inflections of the BFO curve demonstrate a MH370 flight mode impacted by wind strength and direction, and that probably leaves CTH or CMH or manual flight holding to a certain meridian line.
“…I think that assumptions about how the plane was flown between 18:28 and 19:41 was one of the more fundamental flaws of the DSTG analysis.”
>> Yes! If one takes a look at your proposed “LAGOG switchback to BEDAX” route or @Nederland’s slightly more complex “Go around Sumatra” route: eg; IGOGU APASI TOPIN ISBIX” the timing is near perfect to hit Arc2 at 19:41. So this route is a very logical option, unless someone has secret intelligence info to rule that out.
Furthermore, I am not convinced that assuming no-loiter is actually a simplifying assumption. Maybe we need to define a period say 20:00-24:00 that might have been straight/level flight. I was much happier with the first ATSB report where they looked at all positions after 19:41, without assuming the 1840 telcon was the FMT.
“…So the recommended search area is now centered on 35S…It’s no wonder that the ATSB has not received funding to resume the search near 35S, or anywhere else.”
>>If ATSB had said the unsearched inside Arc7 area 32-34S, that would have made more sense. I wish we could come to that consensus and push that limited area. I get the sneaky feeling that ATSB wants avoid searching the Broken Ridge caverns at 32S, so when I say 32S, I am getting into that hard-to-search area where they may need to take partial shot at some coverage.
@DrB:
Thank you for posting the low altitude LRC data you obtained recently.
I have noted earlier that in the FCOM LRC Control table there is step-wise increase in fuel flow in the lower left corner (FL250-FL290, Weight 160 & 180 tons). Similar step changes can be observed in the new low-altitude data as shown here.
Another chart shows the same data in corrected form.
I have no explanation for those steps. They obviously defeat all FF modeling.
@TBill said, “I have no problem with LNAV, after all, I can pick even a fractional lat/long target, but why limit to LNAV?”
I just can’t imagine a B777 pilot navigating a long distance to any desired location using anything but great circle paths in LNAV mode. The CMH mode is probably the most difficult navigational mode due to the challenges of wind and magnetic variation. If the pilot really wanted to go due south, he (or she) could have just entered the South Pole as a waypoint (which would be equivalent to a true track of 180 deg). The only way I think that MH370 could have been in a CMH mode after 19:41 was if the plane was in this mode prior to 19:41 and the pilot became incapacitated, OR if the plane was in LNAV mode, the pilot became incapacitated, the plane reached a route discontinuity, and the plane continued in CMH mode. That’s why I said that a competent, coherent pilot would have navigated long distances in LNAV mode.
@Victor Iannello: Doesn’t that imply that tht pilot was not “competent, coherent”?
@Gysbreght,
You said: “I have no explanation for those steps. They obviously defeat all FF modeling.”
Maybe so, or maybe not, but there must be a reason for this. I will take a look at this again later on this week, along with numerous additional charts you have posted (combined with excellent questions).
As an aside, I don’t understand why you use “Weight/delta” often for the abscissa. Neither Fuel Flow nor airspeed seems to be constant against this variable. I understand the desire to combine weight and altitude into a single variable so you only have two, not three. Why not use weight/density or something similar combining pressure and temperature with the correct exponents.
@VictorI,
You said: “So the recommended search area is now centered on 35S. Unfortunately, this area is contraindicated by the early arrival of “Roy” in South Africa, the lack of recovered debris in Western Australia, and the failure of the underwater search close to the 7th arc (+/- 20 NM) at this latitude.”
I believe you are of the opinion that the drift data presented by Dr. Griffin is inconsistent with his conclusion that debris from 35S would not hit the WA shoreline. Is this correct? Were you able to get a response from Dr. Griffin on this point?
@DrB: ” I will take a look at this again later on this week”
You may want to change the FF/eng at FL190 150 MT from 2954 to 2594 😉
“I don’t understand why you use “Weight/delta” often for the abscissa.”
You could do a Google Search for “non-dimensional parameter”.
@DrB: While I’m at it, here’s another plot. It shows all weights, all altitudes up to FL250, except conditions where LRC speed is limited to Vmo-5kt.
@VictorI
“I think that assumptions about how the plane was flown between 18:28 and 19:41 was one of the more fundamental flaws of the DSTG analysis.”
I agree with you on this point. I would like to add that IMO you don’t need a late FMT and/or loiter to explain a crash location north of S36. An early FMT in combination with a low and slow passing of the tip of Sumatra with typical 19:41 latitude roughly around N1 degree could also result in a more northerly crash location. I have no idea about the fuel calculations for such a scenario, so perhaps one can eliminate it based on those.
@Tbill,
Re: “I think the aircraft could be at ~33S +-1 degree”
Forget about it. Respective area was well covered during the aerial search – hard to believe the AMSA (Australian Maritime Safety Authority) could miss the debris field. In addition some debris would likely be found in Australia and Kenya, where nothing was actually found.
@Victor,
“I just can’t imagine a B777 pilot navigating a long distance to any desired location using anything but great circle paths in LNAV mode. ”
This point is exactly the one, which conflicts with NW turn at Penang, the lateral offset, descent over the Malacca, manual piloting over the Gulf of Thailand, and so on… The two modes, before and after 18:40, are simply inconsistent with each other.
@Gysbreght asked, “Doesn’t that imply that the pilot was not ‘competent, coherent’?”
Not at all. In my opinion it is most likely the plane was following a great circle path in LNAV mode after 19:41. What we don’t know is how the plane was flown between 18:28 and 19:41. I believe there is a possibility of an ongoing descent at 18:40, which the DSTG analysis dismisses.
@Oleksandr: I think the plane was flown in HDG/TRK SEL mode after the turn back and before Penang. The scale of the MAP mode of the navigational display could easily allow navigating in the vicinity of Kota Bharu and Penang. Put a circle around the airport fixes and you could even do DME navigating. After the turn to the SIO, the navigational display could not be used because of the long distances. The maximum scale in the MAP mode is 640 NM.
@Victor
Re: “I don’t think it’s that simple.”
Maybe not, but it might be.
Re: “If a competent and coherent pilot was intentionally flying into the SIO, the only navigational mode that makes sense is LNAV.”
Yes, agreed, so maybe it’s time to consider some alternative scenarios; What circumstances might give rise to the airplane flying in modes other than LNAV such as CMH or CMT? Are there alternatives to a competent and coherent pilot intentionally flying into the SIO?
We’re very confident that the airplane was flying in LNAV from KL to IGARI, it may have been flown in LNAV from the turnback to Penang, there’s no good evidence that it flew in LNAV up the Strait (I’d argue that it was probably flying CMH at that time), you assume that it returned to LNAV for the SIO leg because the airplane couldn’t have been navigated to a desired end point well into the SIO any other way. What if you set that last assumption aside?
@Gysbreght
Re: “Perhaps the ‘expensive’ experts were not sitting either in the SSWG.”
Maybe, but it would be simply astounding if their was some sort of misplaced thrift associated with planning a two year, $200 million endeavour; pinching pennies to squander pounds.
@TBill
Re: “You said:”The remaining three lateral control modes generally resulted in the aircraft’s flight path ending north of the CTT and LNAV results, up to 33°S in latitude along the 7th arc.””
I’m quoting the ATSB’s First Principles Review.
Re: “I agree the PSS777 sim study may have used LNAV to get around Indonesian radar space for a Jeddah flight case …”
But it didn’t. Connecting the data points retrieved from the Captain’s flight simulator shows a flight path that blithely passes within 60 nm of Medan and 80 nm of Banda Aceh; there is no apparent attempt to avoid the coverage from either radar.
@Mick Gilbert:
“We’re very confident that the airplane was flying in LNAV from KL to IGARI”
My interpretation of the PSR civil radar data is a plane flying relatively straight segments connected by small turns. I think HDG or TRK SEL is the most likely A/P mode.
“there’s no good evidence that it flew in LNAV up the Strait (I’d argue that it was probably flying CMH at that time),”
I’d say without good radar data, we are only guessing about the A/P mode in the Malacca Strait. Considering that Malaysia only supplied the ATSB with one radar capture after 18:02, it is hard for me to assign credibility to the Lido Hotel image. Malaysia owes an explanation to the world as to what was shown the NOK so we don’t have to guess.
“Are there alternatives to a competent and coherent pilot intentionally flying into the SIO?”
None that I consider likely, considering the sequence of failures and non-failures required for a flight into the SIO with no communication. Not to mention the simulated flight found on ZS’s computer that leads from the Andamans to the SIO.
But then again, I’m just one person. Others feel differently.
Niels said, “I agree with you on this point. I would like to add that IMO you don’t need a late FMT and/or loiter to explain a crash location north of S36.”
Yes, you are right. I refer to “loiter” to generically include paths that are significantly slower than the flight before 18:28 and/or have multiple turns.
@DrB said, “I believe you are of the opinion that the drift data presented by Dr. Griffin is inconsistent with his conclusion that debris from 35S would not hit the WA shoreline. Is this correct? Were you able to get a response from Dr. Griffin on this point?”
I posted Dr Griffin’s email to me on the previous thread. In response to his email to me, I sent him an email, which I include below. I did not receive a response to my email.
David,
Thank you for taking the time to respond. I will post your comment on my blog so others can see your explanations.
I understand that the paths are representative and not to be taken literally. I also see how you injected particles over a range of locations in order to create the diversity, i.e., distribution of results. And I agree that trends are more important than any particular particle path.
My main concern is that researchers are using these results to proclaim with a high level of confidence that the crash site was along the 7th arc at 35S latitude. Your own results show that crash sites further to north would better explain the arrival of “Roy” in Mossel Bay as well as the lack of debris (both high and low windage) found along the coast of Western Australia. (I commend CSIRO for providing access to the drift model results in a form that others can easily use.)
As far as debris arriving in Africa too early for a 30S crash site, I would suggest that debris might have arrived but was not discovered. For instance, I don’t believe Blaine Gibson found “No Step” just after it arrived to the shores of Mozambique. Likely, it and many other parts arrived much [earlier] and were either not found or found and not reported. The fact that he found it soon after looking suggests that there were likely many more parts.
As for the aerial search, I question the 100% efficiency of those efforts based on the comments of others with more SAR knowledge than me. I also see that the stated probability of discovery is not 100% as you proceed north along the arc, including holes in the coverage around 30S and a fall-off of coverage north of 27S.
The seabed near 35S was also searched to a distance of 20 NM. The final BFO value indicates the plane was in a steep descent near the 7th arc. I believe that if the plane crashed near 35S, it likely would have been found.
I cannot say with absolute certainty that the plane did not crash at 35S. The purpose of my post was to question the confidence level expressed by CSIRO and the ATSB and to invite others to independently review and evaluate the results.
Regards,
Victor
@Gysbreght
“… if their was … ”
Good grief, pardon my homonymic error.
@Victor
Re: ““We’re very confident that the airplane was flying in LNAV from KL to IGARI”
My interpretation of the PSR civil radar data is a plane flying relatively straight segments connected by small turns. I think HDG or TRK SEL is the most likely A/P mode.”
I suspect that you’re taking about the IGARI to Penang rather than KL to IGARI. If so, I don’t disagree that MCP HDG SEL or TRK SEL are possible; if not, I do disagree, it’s difficult to conceive of a circumstance where a flight operating normally would try to track to direct to IGARI using MCP HDG SEL or TRK SEL.
@Victor Iannello: “We’re very confident that the airplane was flying in LNAV from KL to IGARI”
I’m not so confident. If the “Direct to” had been used in LNAV the airplane would not have overflown IGARI before turning towards BITOD.
@Gysbreght
Re: “If the “Direct to” had been used in LNAV the airplane would not have overflown IGARI before turning towards BITOD.”
What may have occured was the usual early turn towards BITOD (given the 30-odd° change of heading it wouldn’t have been much more than 2 nm before IGARI) with the subsequently commanded/flown left turn bringing the airplane back to or very near to overflying IGARI.
@Mick Gilbert: I’m sorry, I don’t understand your reply. If the active flight plan had been changed by executing “Direct To” IGARI, that waypoint would have been passed in the normal way with a turn onto the track IGARI-BITOD. Instead the airplane overflew IGARI and turned to a heading south of BITOD to intercept the track IGARI-BITOD.
BTW I’ve updated the chart I posted yesterday.
@Mick Gilbert said, “I suspect that you’re taking about the IGARI to Penang rather than KL to IGARI”
Yes, that’s what I meant to say.
@Gysbreght: I don’t believe the plane was in LNAV between IGARI and Penang. As I said, my interpretation of the PSR civil radar data is a plane flying relatively straight segments connected by small turns. I think HDG or TRK SEL is the most likely A/P mode.
@Victor Iannello: “As I said, my interpretation of the PSR civil radar data is a plane flying relatively straight segments connected by small turns. I think HDG or TRK SEL is the most likely A/P mode.”
There are no facts that support your interpretation, which is no more than a guess. You assume that you know who was flying but you can’t be sure of that.
My guess that the plane was flown manually is equally valid. Do not forget that groundspeed and possibly altitude are also variable.
@Gysbreght said, “There are no facts that support your interpretation, which is no more than a guess.”
I constructed this plot as part of my study of the radar data. The civil PSR data shows a plane flying relatively straight segments connected by small turns, as I said.
“Do not forget that groundspeed and possibly altitude are also variable.”
The evidence for changes in groundspeed is the DSTG plot, which is based on the Malaysian military radar data. However, the plot depicts an unrealistically low speed at the turn back, multiple overspeed conditions over the Malay peninsula, and the groundspeed inexplicably does not match the ADS-B data before IGARI. It is hard for me to believe this data. I suspect that Malaysia supplied inaccurate radar data to the ATSB, which the DSTG filtered and presented as a plot. That might also explain why the Malaysians are so reluctant to release the radar data despite requests from independent investigators and from the ATSB.
When I posed questions about the inconsistencies of the radar data to the Malaysian officials, I was assured those questions would be answered in the Interim Report in March 2016. The Malaysians lied, as no additional radar information was included in the report.
@Victor Iannello: Yes, of course I’m well aware of all that. But a guess is a guess.
@Andrew,
I think I will be bothering you time to time, if you don’t mind.
Re SDU you wrote: “Both channels of IRS data are sourced from the ADIRU. One channel is from the L AIMS cabinet and the other is from the R AIMS cabinet.”
I am looking at the ADIRU diagram. The ARINC629-C FCA (Fault Containment Area) only receives data from both the processor FCA and SAARU, but its output is sent back to the ADIRU’s processor FCA. Apparently this is where comparison of the ADIRU and SAARU data is performed. Respectively, what happens when the original ADIRU’s output does not match the SAARU’s data, while both of them do not detect any internal error? What output is sent to the ARINC629-L&R in this case?
I’ve also updated the TSFC chart I posted earlier.
The update incorporates DrB’s LRC Low Altitude FF and Mach data without the FF increments of 2.5% and 5% at low weights and altitudes.
Sorry, wrong chart. Here is the TSFC chart.
@Victor
Maybe Andrew can chime in on navigating (to a general region in the ocean) via CTH or CMH vs. waypoints. I see no problem, but then again with my vision prescription, nobody ever mentioned “pilot career” and my name in the same sentence.
I am just trying to make sure you are not wrong to push waypoints, because you personally like that. If I can shake you off of waypoints, then you can possibly agree that the unsearched inside Arc7 32-34S is the most valid search area. Or you can use 3394S for a waypoint.
@Mick Gilbert
I was not following when you said: “Connecting the data points retrieved from the Captain’s flight simulator shows a flight path that blithely passes within 60 nm of Medan and 80 nm of Banda Aceh; there is no apparent attempt to avoid the coverage from either radar.”
I am saying the portion of the sim flight path 1090E NOBEY in SkyVector suggests to me flying around Indonesian air space from the MAS Jeddah flight path. Of course the outbound path to Jeddah is normal thru the straights.
In the end, the Jeddah flight was not chosen for the diversion, thus I am not yet convinced that the McMurdo waypoint has any significance for the Beijing flight path actually used.
I feel the actual MH370 Beijing flight may have gone something philosophically similar to the sim studies, heading toward LAGOG but cutting corners to save fuel to NISOK then due south down around 94E to 33-34S.
@TBill: Perhaps I am missing something. If a pilot wanted to navigate to a waypoint that is 1000 NM away, for instance, how would he or she do it without entering the waypoint and selecting LNAV?
@TBill
“If I can shake you off of waypoints, then you can possibly agree that the unsearched inside Arc7 32-34S is the most valid search area. Or you can use 3394S for a waypoint.”
Most valid? The only thing lacking is any motive or causality for terminating there. Suicide is the only explanation, and that seems very unlikely to me.
@Victor
To clarify my position, I see no need for waypoints to fly South.
The Inmarsat arcs are telling us (quite obviously) MH370 may well have gone straight South of ISBIX (93.67E) or BEDAX (93.76E) or BULVA (94.0E). That’s an obvious flight path, and the end point is unsearched inside of Arc7.
I offer you a close by end-of-flight waypoint: and that would be 3394S or similar oceanic position. That (waypoint) was my first approach but I feel 180S True Heading probably gives best match to BFO/BTO. I see the flight plan as getting to deep water just past 31-South, but I do not really need a waypoint for that general flight path.
@DennisW
I do not know if MH370 was a suicide or a negotiation gone bad.
A heading 180S from ISBIX can divert to Cocos, CI, Jakarta, and probably Learmonth as an option.
@Gysbreght,
It would be helpful if you would include a legend on your Corrected FF plot so we can understand what is being plotted.
Thanks for catching the typo in the FF table – dyslexia strikes again.
You said: “You could do a Google Search for “non-dimensional parameter”.
I already did that. In this case, it is a waste of time because you still have 3 variables (not 2) so there is no benefit. The function needed to combine weight with properties that vary with flight level is unlikely to be non-dimensional. I’ll see if I can come up with something better.
@TBill
Re: “I am saying the portion of the sim flight path 1090E NOBEY in SkyVector suggests to me flying around Indonesian air space from the MAS Jeddah flight path. Of course the outbound path to Jeddah is normal thru the straights.”
That portion of the sim flight path makes no sense whatsoever from virtually any perspective.
IF (capital I, capital F) we are looking at a recreation of MH150 then R467 GUNIP MEKAR N571 keeps you in Malaysian airspace until just before the Andaman and Nicobar Islands, you then pass into Indian airspace. If you wanted to divert south and avoid Indonesian radar there is no requirement at all to divert north – that makes no sense, all you are doing is wasting time and fuel – you simply stick to your flight plan till you reach 91°E and then hang a left. There is a radar blackspot in the Bay of Bengal that doesn’t require any fancy manoeuvring to take advantage of, MH150 flies straight through it (as does MH16 and a handful of other westbound MAS flights out of KL).
@Victor
Returning to the notion that the Left AC Bus might have been deliberately depowered in order to (temporarily) shut down the CVR, having spent some time looking at that matter and possible alternatives, I now think that it is unlikely that someone familiar with B777 systems would take that action (however, as I am often reminded, never say never).
For starters, the seminal event, either a flight crew member saying “kapal terbang saya. adakah anda keberatan mendapatkan saya kopi?” or the dull thud of the flight deck fire extinguisher impacting the back of someone’s head, would have already been captured by the CVR. All shutting the CVR down subsequently achieves is preserving that record; it prevents it being overwritten on the two hour loop.
Second, if the perpetrator’s intent was to prevent the CVR from capturing his actions subsequent to assuming command of the airplane, then defeating the CVR turns out to be a relatively straightforward exercise that can be accomplished very simply with little more than two sticks of chewing gum; remove wrappers, chew until pliable, place wad of gum over CVR flight deck area microphone on the overhead panel – et voilà. If you wanted to adopt a belt and braces approach you might also physically disconnect each of the headsets, hand-held microphones and the pilot handset.
We have already discussed some of the collateral effects of depowering the Left AC and Transfer Buses; among them is the loss of the Captain’s forward and First Officer’s side Window Heat and Backup Heat. The initial effect of the loss of heat would be fogging of both inner panes but the near term effect on the Captain’s windshield would be subjecting the outer pane to increasing stress as it cooled and contracted within its frame. That contraction, of course, may lead to cracking which may in turn lead to pressurisation issues. That is probably not something that a perpetrator would want to risk if there was 6 hours of flying ahead.
Additionally, as Don pointed out to me some time back, depowering the Left AC and Transfer buses would also affect a number of air data sensor heaters, specifically the Left Pitot Probe Heat, the Left AOA Vane Heat and the TAT Probe Heat. All of those heater failures would be sensed and their respective data feeds would be ignored. While that collective loss of data would not force the PFCs to revert to secondary mode (thus causing the autopilot to disengage), the loss of redundancy places the airplane that much closer to reverting to secondary mode (and the concomitant requirement for it to be flown manually) in the event of failures on the right hand side. Not a high risk but is it one that a perpetrator would be willing to take?
Given the alternatives I think that someone who was ostensibly very familiar with B777 systems and who had supposedly dedicated at least 6 weeks planning to the venture would have been more likely to have opted for the gum solution.
@Victor
I show the simulator route (1090E NZPG) hitting Arc3 at 2111 (vs. 2041) so going out that far could have cost about 30-minutes, and the wreck would probably be located in the 30S/93E area in shallower water before going over the Broken Ridge cliff (assuming a turn to the south in the 93S area).
@DrB: “It would be helpful if you would include a legend on your Corrected FF plot so we can understand what is being plotted.”
I understand your difficulty, but what your asking is not quite simple. Let’s start with a chart that shows all conditions in your LRC table for FL250 and below, plotted indiscriminately.
In those conditions four groups can be distinguished with distinctly different characteristics. The points with the lowest FFc/W values represent conditions of weight and altitude where the LRC speed is limited to Vmo-5 kts (325 kIAS). These are not shown in the chart you’re referring to, and the conditions are indicated by yellow triangles in the next chart.
The next group is what I would call the baseline. They are the solid symbols in the chart you’re referring to. These points form a line that is a continuation of a similar line for points from the FCOM LRC Control table for FL250 to FL430.
Then there are two groups above the ‘baseline’ with FFc/W values of baseline +2.5% and baseline +5%. The weight/altitude conditions are indicated in the second chart.
Finally the red line in the chart you referred to is obtained by subtracting 2.5% from the values in the ‘2.5%’ group, and 5% from the values in the ‘5%’ group.
@Mick Gilbert: If isolating the left AC bus was meant to minimize the evidence on the CVR, it is possible that some of that incriminating evidence was created when both pilots were in the cockpit, such as the captain granting permission to the first officer to leave the cockpit. When the captain was alone in the cockpit, he could have isolated the bus at that time, stopping further evidence from being recorded, such as the sounds from the first officer trying to get back to the cockpit. When the plane was at a location where interception was no longer probable, the pilot could have been repowered the bus and the CVR to erase whatever evidence was recorded prior to the isolation. As you say, it also put the plane in a safer electrical configuration. There might have also been a desire to monitor SATCOM calls.
Your bubble gum trick doesn’t seem nearly as easy, effective, or fast as flipping two switches to isolate the left bus.
Tampering of the CVR by a suicidal pilot was part of the SilkAir 185 and Federal Express 705 crashes. It is a possibility that we should consider for MH370.
@Mick Gilbert: “All shutting the CVR down subsequently achieves is preserving that record; it prevents it being overwritten on the two hour loop. “
A long time ago I described a hypothetical scenario where that could have been the objective.
@Victor
Re: “Your bubble gum trick doesn’t seem nearly as easy, effective, or fast as flipping two switches to isolate the left bus.”
Seriously? Placing something like a piece of gum (or plasticine if you wanted to speed the process up by obviating the need for chewing) over the flight deck area microphone neutralises the CVR for cockpit sounds, including that of someone belting on the flight deck door. It is easier, as effective and probably as fast as taking down half the electricals with all the collateral losses. And it has the bonus of bringing evidence of that seminal event an hour closer to being erased.
A minor point but it’s three actions to depower the bus – selecting the Left Bus Tie Breaker to ISLN, selecting the Left Backup Generator to OFF, and selecting the Left Generator to OFF.
@Oleksandr
RE: “I am looking at the ADIRU diagram. The ARINC629-C FCA (Fault Containment Area) only receives data from both the processor FCA and SAARU, but its output is sent back to the ADIRU’s processor FCA. Apparently this is where comparison of the ADIRU and SAARU data is performed. Respectively, what happens when the original ADIRU’s output does not match the SAARU’s data, while both of them do not detect any internal error? What output is sent to the ARINC629-L&R in this case?”
The SAARU transmits data to the C Flight Controls ARINC 629 Bus and the ADIRU receives data from the same bus, but what makes you think the ADIRU uses SAARU data? According to the AMM:
“The SAARU sends inertial data on the center flight controls ARINC 629 bus. The SAARU data goes to the:
– Left and right AIMS cabinets
– Left, right, and center primary flight computers (PFCs)
– Left, right, and center autopilot flight director computers.”
There is no mention of the ADIRU using SAARU data or performing any comparison function. The ADIRU monitors itself and if the output data is invalid, the PFCs and AFDCs automatically use the SAARU data in lieu.
@Mick Gilbert: I don’t think you can dismiss the possibility of isolating the left bus to disable the CVR because you believe it’s easier to do with bubble gum. I don’t believe that’s true, but even it is, that possibility might not have been considered by the pilot. And although isolating the left bus is not an ideal configuration, the pilot would know that any critical systems are protected by redundancy.
As for the actions required to isolate the bus, I don’t think you are correct. The left backup generator would supply power to the left transfer bus, but the left transfer bus cannot supply power to the left main bus. Therefore, the only two switches required are the left bus tie and the left generator switches.
It seems to me, with the increasing indications and evidence gathered till now that contradict a (partial) ghost flight and the lack of indications, let alone evidence, the plane was not actively piloted till the end, it gets about time to seriously consider a completely planned flight and disappearance.
I mean a flight planned with also a specific chosen end-point.
An end-point which suites the goal of minimizing risk of ever being found and recovered so no evidence of who was responsible would ever get found.
An end-point deep enough, reachable from the 7th arc after flame out of both engines.
There’s one option near 32S on the 7th arc that suites this criteria.
This is the Ob’trench with depths of ~5300m starting at 32.27S/97E till 97.40E.
If it would be possible to scan this area (which I estimate around 5000km/2) with the possibilities named in the current topic with a resolution of 5m I think it would be worth it.
The resolution would probably be high enough to find the wreckage.
@VictorI
As I also suggested as a scenario before, I find your answers to @Mick Gilbert adding more credibility.
Isolating the left IDG would be very efficient with only two swithes to handle. The only one left would be the transponder-knop won’t it.
This could all be done in seconds rendering the plane completely dark (including main cabin-lights).
And I think this it not your regular all experienced captain with +10.000 flight hours. This captain was also an instructor, a pilot with a sophisticated home-computer, an expirienced (but broken back) para-glider, a man who liked to sort out things in detail (according his videos on other subjects).
It seems to me a pilot who lived his profession rather extreme.
A pilot who knew all the ins and outs of a B777. Knowing exactly where it’s limits and possibilities were.
Depowering the CVR right after the co-pilot left the cabin (by isolating the left IDG) would indeed only reveil the co-pilot left the cabin.
If the plan failed within two hours this would still leave not any proof on what happened afterwards. Suspicion but no proof.
@Mick’s gum-solution is creative but ofcourse very unlikely as a chosen option by such a well trained pilot when another simple ~10 seconds option was available which not only would depower the CVR but also the ACARS, SDU, most radios, main cabin lights, IFE etc. Everything you want if you want to disappear and block communications in an instant.
And it happened also at exactly the right time.
The whole flight just shows deliberate actions till 18:40. There’s no reason/evidence at all to assume this changed after 18:40.
Imo this remains the biggest ‘mistake’ the DSTG/ATSB made. Although not quite: there biggest mistake happened after the flaperon was found and they refused for more then one and half year to adjust this assumption till it was to late for the search ended.
And they still seem to refuse their mistake by holding on to 35S with the support of CSIRO.
I hope some official there gets his act together.
@Ge Rijn: The ATSB is in a difficult position. My guess (and it’s only a guess) is they believe the diversion was deliberate. But because the criminal investigation is led by Malaysia, and because of other political sensitivities, the ATSB had to analyze the data without considering the criminal activity leading up to the last radar point.
We also know that certain information was withheld from Australia by Malaysia, such as the Lido Hotel slide of the radar captures in the Malacca Strait, and the cell phone registration with a tower on Penang. I suspect there is other important information that is being withheld.
@DrB: RE (quasi-)”non-dimensional” parameters: “In this case, it is a waste of time because you still have 3 variables (not 2) so there is no benefit. The function needed to combine weight with properties that vary with flight level is unlikely to be non-dimensional.”
If you treat Weight/delta as the independent variable, the dependent variables for LRC are FF (corrected to standard conditions) and Mach. Without using non-dimensionals, FF and TAS depend on weight, altitude, and temperature.
The same applies for holding speeds.
If you treat speed as an independent variable, then the non-dimensional fuel flow depends on two (2) independent non-dimensional variables, Weight/delta and Mach. Without non-dimensionals the fuel flow depends on four (4) independent variables weight, altitude, temperature, and speed.
“I’ll see if I can come up with something better.”
I’m not holding my breath.
@Victor
From the moment I hypothesised that depowering the Left AC Bus might have been a response to a rapidly escalating inflight emergency I have had pilots and systems engineers telling me that, while it’s possible, it represents extremely drastic action. I don’t profess to understand the mind of the hypothesised perpetrator but I think that a pilot would be reluctant to pursue such drastic action when a far simpler but as effective countermeasure was available to him. Add it to the column of matters we disagree on.
@Ge Rijn
Re: “Mick’s gum-solution is creative but ofcourse very unlikely as a chosen option by such a well trained pilot when another simple ~10 seconds option was available which not only would depower the CVR but also the ACARS, SDU, most radios, main cabin lights, IFE etc. Everything you want if you want to disappear and block communications in an instant.
And it happened also at exactly the right time.”
The good people at Boeing went to the trouble of providing flight deck switches designed to specifically depower the cabin lights, the IFE, the radios (all of them, not just a few) and the transponders (both of them, not just one). For fligth crew used to following checklists it would not be beyond the realm of possibilities that a hypothesised perpetrator would either create a checklist for turning the airplane “dark”.
Moreover, you’d have our highly organised and proficient perpetrator repower the Left AC Bus only an hour later while leaving the IFE on; that in of itself makes no sense. If the perpetrator knew that depowering the Left AC Bus was his one stop shop for, inter alia, depowering the IFE then he must have realised that repowering the bus would bring the IFE back on line and he could have not been confident that everyone in the passenger cabin was unconscious at that point.
As to it happening at exactly the right time, the only exact time we have for a failure is for the transponder; the SDU could have been depowered any time in the 55 minute period between 1708 UTC and 1803 UTC.
@Gysbreght,
You said: “If you treat Weight/delta as the independent variable, the dependent variables for LRC are FF (corrected to standard conditions) and Mach. Without using non-dimensionals, FF and TAS depend on weight, altitude, and temperature.”
The temperature dependence of TAS exists whether you use non-dimensionals or not. You cannot avoid this.
In ECON mode, it is obviously simpler to predict Mach than doing so using TAS. You can let the TAT/SAT sensor in the aircraft tell you what is needed to convert between Mach and TAS when flying and controlling the airspeed.
The same is true for Fuel Flow when considering temperature deviations from ISAT. The corrections to TAS and to FF are given as footnotes in the Boeing LRC tables. I presume these same corrections are also done by the FMC in flight based on real-time temperature measurement.
So both Mach-to-TAS corrections and FF actually depend on temperature, and this does not depend on the dimensionality (or non-dimensionality) of your plots.
This is also true for Holding in that the KIAS-to-TAS conversion depends on temperature (as does Holding FF). There is no “avoidance” of this temperature dependence no matter what speed variable you start with.
You said: “If you treat speed as an independent variable, then the non-dimensional fuel flow depends on two (2) independent non-dimensional variables, Weight/delta and Mach.”
I disagree. It also depends on a third variable: temperature. This is true no matter which fuel flow variable you use (FF or FF/W) or which weight you use (W or W/delta).
You are claiming that FF/W only depends only on M and W/delta. I note that temperature is conspicuously absent. That can only be true if the temperature is assumed to be the standard ISAT. When SAT is not equal to ISAT, FF/W also depends on (SAT-ISAT). So in actuality you have FF/W depending on 3 (not 2) variables: M, W/delta, and (SAT-ISAT). Still, that is better than four variables (M, FL or delta, SAT, and M). By using FF/W you can combine W and FL into a single variable (W/delta) that allows you to have only 3 variables instead of 4, and that is worthwhile. My previous comment was based on the idea of further incorporating the temperature, as well as the pressure (i.e., delta), into a new variable that would reduce the number of variables from 3 to 2. I don’t know if such a function exists. If not, 3 is the best one can do for FF.
For airspeed, using Mach (or KIAS) is advantageous over TAS for prediction because its prediction is simpler as it does not depend on temperature. Still, the temperature must be known/measured and used in the airspeed control loop. So temperature is still necessary, although it is easier to incorporate it into the airspeed sensor/controls than in the airspeed prediction equations as an additional variable.
@Victor
Re: “As for the actions required to isolate the bus, I don’t think you are correct. The left backup generator would supply power to the left transfer bus, but the left transfer bus cannot supply power to the left main bus. Therefore, the only two switches required are the left bus tie and the left generator switches.”
The CVR has one power source, 115V AC from the Left Transfer Bus. You would have to select the Left Backup Generator to OFF as well as selecting the Left Bus Tie Breaker to ISLN and selecting the Left Generator to OFF in order to depower it.
And just to be clear, I’m not discounting the possibility that a malicious perpetrator did depower the Left AC and Transfer Buses, I’m just saying that it is unlikely.
@all,
Does anyone have a guess as to why Boeing would add 2.5% (for W = 140 MT) and 5.0% (for W = 190 MT) to the Fuel Flows for the LRC entries below FL250? Could it be an allowance for the effect of extra bleed air at lower altitudes (just a wild guess on my part)?
@Gysbreght,
For instance, suppose instead of FF/W the fuel flow variable was FF/W/[theta^0.68], where theta is SAT/ISAT (both in K). That combines temperature and FF into a single variable, which depends only on two things: M and W/delta (ignoring for the moment the “extra” FF of 2.5/5.0% at low altitudes and weights seen in the LRC tables).
@Mick Gilbert: Yes, you are correct that the CVR is powered from the left transfer bus. That means that 2 switches are required to isolate the left main bus and power down the SATCOM, but 4 switches are required to isolate the left transfer bus and power down the CVR, as the right backup generator can supply power to the left transfer bus. Nonetheless, all the switches are in close proximity, and it would be easy to accomplish.
When the left main bus is isolated, the following messages appear on the EICAS, among others:
SATCOM DATALINK
SATCOM
SATVOICE LOST
When both the left main bus and the left transfer busses are isolated, the additional messages are:
NO LAND 3
HEAT PITOT L
WINDOW HEAT R SIDE
You said, “From the moment I hypothesised that depowering the Left AC Bus might have been a response to a rapidly escalating inflight emergency I have had pilots and systems engineers telling me that, while it’s possible, it represents extremely drastic action.”
If there was an escalating emergency, powering down the left main bus may be a drastic action. If there is no emergency and systems are operating with full redundancy, perhaps it’s not so drastic as all critical systems will continue to function.
@Victor
“… the right backup generator can supply power to the left transfer bus.”
That’s a good point, Victor. Are the Left and Right Transfer Buses isolated from one another in normal operation or are they tied via the respective Converter Circuit Breakers?
@Mick Gilbert: Under normal operation the left main bus supplies the left transfer bus and the right main bus supplies the right transfer bus. The left and right transfer busses must therefore be electrically isolated.
@Ge Rijn
For the sake of this brief exercise, let’s suppose that one (out of many) of Zaharie’s main objectives/goals was to disappear the aircraft as ‘completely’ as humanly possible. Setting aside impact and debris considerations, it’s a SURE bet that inherent in such a pre-meditated act would be the consideration of location and underwater topography.
This is not something that would be overlooked wiily-nilly, because, for example, the location was by default so difficult to search and access that it would cease to be a mandatory consideration.
A pilot intent on ‘disappearing’ would: 1) Control the aircraft through EOF.
2) In conjunction with a pre-determined expectation of potential debris field , the location, depth and contour (underwater features) would also be factored in. I would think that trenches and other daunting contour features would be sought out and prioritized.
If this was a pre-determined disappearance, the chances of a ghost flight/hypoxic southern leg are all but a big fat 0…unless one cares to make allowance for a thwarted plan. And evidence does not support this remote possibility whatsoever.
I think those that are using the 15,000 fpm BFO to conclude that a high speed impact occurred should perhaps be spending more energy on EOF scenarios that would have an alive pilot in a high speed dive and subsequent pullout. Just a suggestion.
@DrB
In this slide presentation, at around Slide 104-106 there some discussion about Boeing bleed air source switching:
https://www.slideshare.net/sanjuvinaykumar/jet-engines-basics
@Donald
I came into this search borrowing an oceangrapher’s suggestion that Broken Ridge was an possible intended ditching zone because the aircraft would be hard to find there. The problem with that theory for me is that I have no idea what specific feature or area an oceanographer or pilot would pick as a “good” spot. The obvious choices such as Dordrecht hole are outside of Arc7. Right now I am thinking a simple attempt to get over the Broken Ridge wall at 31S into deeper water. I got some trenches too. But I would like to see some discussion or annotated map of difficult areas. I also have a sneaky suspicion that the search zone areas being chosen for ease of search (no way they want to search in the 31-32S BR badlands).
@DrB: “You are claiming that FF/W only depends only on M and W/delta. I note that temperature is conspicuously absent.”
Temperature is represented in the corrected FF by theta and its exponent. If you are unable to see the benefit of non-dimensional parameters then just do not use the concept. I’m tired of explaining it to you, perhaps someone else can do that better.
@TBill: Thank you for the “Jet Engine Basics”.
@DrB: Corrected or referenced (non-dimensional) engine parameters are mentioned on on many slides, e.g. slide 38 (N1), 55 (thrust), 102-104 (FF).
@Victor
@Mick
RE: “That means that 2 switches are required to isolate the left main bus and power down the SATCOM, but 4 switches are required to isolate the left transfer bus and power down the CVR, as the right backup generator can supply power to the left transfer bus.”
Gents,
The right backup generator can indeed supply power to the left transfer bus via the converter, but if the left backup generator switch has been selected OFF, then the backup generator converter will open the left converter circuit breaker, isolating the left transfer bus from the backup power system. If the left generator control switch has also been selected OFF and the left bus tie switch selected to ISLN, the left transfer bus will be unpowered.
@DrB: If you go to TBill’s reference and scroll down, you’ll find interesting footnotes to the slides. For example:
“92. P3, Page 6 “Generalized” Thrust Chart • This PEM chart presents “corrected” installed net thrust/δ as a function of Mach number and %CN1. • The chart is valid for any phase of flight. But, it represents an average engine thrust level, and as such should not be used for takeoff calculations. • If we plotted net thrust versus speed and %N1, it would be necessary to prepare a different graph for each altitude (because thrust is a function of air density) • Instead, we plot “corrected” thrust versus Mach and %CN1. This allows the presentation of the data for all altitudes in one single chart. Refer to 7X7 PEM, Section 3, Page 3.16 δ = ΝF thrustCorrected Χ Τθ Ν=Ν1 1%C%
P3, Page 16 “Corrected” Fuel Flow • Many fuel flow charts or tables present “generalized” fuel flow, so that one chart or table can be used for all temperatures. For example, the data in the 7X7 fuel flow tables is generalized. • To generalize fuel flow data, it is necessary to present the fuel flow as “corrected fuel flow” (sometimes called ‘referred’ fuel flow). Refer to 7X7 PEM, Section 7, Pages 7.13 – 7.15 Χ Τ θδ = FW flowfuelCorrected Τ
103. P3, Page 17 “Corrected” Fuel Flow Refer to 7X7 PEM, Section 7, Pages 7.13 – 7.15 Χ Τ θδ = FW flowfuelCorrected WF = the fuel flow rate p p ο Τ =Τδ the total pressure ratio, ο Τ Τ Τ =Χ Τθ the total temperature ratio, raised to the power ‘x’ Τ
104. P3, Page 18 “Corrected” Fuel Flow • Notice that corrected fuel flow is expressed in terms of total pressure ratio and total temperature ratio raised to the ‘x’ power. • The fuel flow data for different engines will not all generalize in exactly the same way, due to design differences. In order to make the data generalize, different engines use different values of ‘x’, the exponent for the total temperature ratio. • For the 7X7, the value of ‘x’ is .60, so that the equation for corrected fuel flow is: Refer to 7X7 PEM, Section 7, Pages 7.13 – 7.15 (7X7) 60. FW Corrected fuel flow ΤΤ θδ = “
@TBill: “In this slide presentation, at around Slide 104-106 there some discussion about Boeing bleed air source switching”.
Those slides discuss the operation of “surge bleed” valves, which are engine features, not Boeing.
Surge bleed valves operate at low thrust, starting, engine accel/decel. They are normally closed in stable cruise flight.
@DrB: “Does anyone have a guess …”
It’s certainly not yours.
@TBill
@Donald
@Ge Rijn
While not discounting the possibility that a hypothetical perpetrator may have planned the flight right down to a terminus over a deep ocean trench, doesn’t it strike you as odd that there’s no apparent evidence of any research along those lines by the purported perpetrator? I mean this is a fellow who lacked the wherewithal to employ even the most basic forensic countermeasure with regards to his flight simulator; he didn’t delete the flight of interest, he simply didn’t save it which allowed it to be overwritten. Moreover, there’s no evidence that he had deleted his internet browser history; the leaked RMP report references internet searches dating back to 20 December 2013.
Then there’s the yawning dichotomy with regards to his alleged desire to render the airplane electronically dark so quickly that he’s disconnected two of the airplane’s four operating generators from and isolated their respective buses yet he’s elected to blithely pilot the airplane back across the Malay Peninsula by choosing a route straight past two radars, one civilian and one shared, that he knew were operational, past another two military surveillance radars, that he had no reason to believe weren’t operational, and within 10 miles of Malaysia’s third busiest airport, all at high level. If our hypothetical perpetrator was concerned about stealth and avoiding interception that is not evidenced by the choice of route; he took the airplane within 5 minutes flying time of two RMAF fighter bases.
@Andrew
Thank you for the clarification regarding depowering the Left AC and Transfer Buses.
@Mick. The vertical stabiliser panel (debris item 22) most probably holds principal clues as to the speed of the crash. I believe you have postulated previously an air surge at a high speed impact causing what looks to be compression damage to its inside skin and honeycomb.
This piece was attached at its front to an auxiliary spar. You will see part of the web of that still attached in the 7th photo down here:
http://www.dailymail.co.uk/news/article-4297982/How-three-year-hunt-yielded-fragments-MH370.html
Attached to the other side of the auxiliary spar flange is the fin leading edge, the chord of it and the front part of the torn off section being of course curved, though that curvature is no longer evident at slide 1 here:
http://www.msn.com/en-au/news/other/mh370-tour-operator-claims-to-have-found-debris-from-missing-plane/ar-BBw5RpE?fullscreen=true#image=1
The inner and outer CFRP skins are separated by aluminium honeycomb and you will see there is a finish over the inner skin which gives it the shiny appearance. The inner skin tapers towards the outer to their join at the auxiliary spar. At the rear the inner skin tapers to join the outer again at the fin front spar. The combination has apparently been levered off that spar.
Some other observations of the damage, to extend discussion, are:
• There are two creases evident on the outside skin, one concave the other convex, which you will see at the 11th photo down of the 1st URL above, and there are corresponding breaks on the inside skin, respectively in tension and buckling/shear: see the 2nd URL. I have the impression that the honeycomb can fail in compression when the whole section is subjected to bending, that then leading to compression failure of the skin inside the bend but having a more complex appearance than one from straight compression. A weaker stress resistance in compression that in tension would explain why the inner surface is more damaged than the outer but so would a thinner skin there. I suspect both, particularly as the outer skin, at its concave crease (see above) has not buckled like the inner. Instead the inner has failed in tension.
• Most of the MAS logo red paint looks to be stripped off. If this is confirmed, as distinct from separating later by immersion, that could give a clue as to impact force.
• The top end of the auxiliary spar remnant is bent in. If this was not subsequent damage it suggests there was side force sufficient to buckle it there.
• The bottom end of that spar section is also bent in (where incidentally much of the auxiliary spar web has been retained). That bend can be seen clearly by returning to the 7th photo of the 1st URL and accessing the video there. Should you look for a repeat of that, refresh the site.
• The abrupt top auxiliary spar buckle, which you can see then extending smoothly in fact along the whole spar to its bottom, that is along what should be a straight spar, is important for it does suggest a large lateral force on the fin, at the front of its chord at least, enough that is to both bend and buckle it That could happen if the whole fin or rear fuselage broke off, the tip of the fin then striking the sea, or it could be that the fin hit the sea asymmetrically with the aircraft upside down. The former seems more likely since a large lateral force is less likely with the fin still attached. In that case there were most assuredly high impact forces.
@Ge Rijn. I remember you had an interest in the above too, with ASIANA flight 224 in mind.
@David: Do you know if there is fuselage skin between the space inside the vertical stabilizer and the space in the rear fuselage below it?
@Gysbreght. No, though to maintain fuselage pressurisation integrity I would bet a bob on it, ie the fin is not pressurised.
@Gysbreght. I take that back. Pressurisation would stop at the pressurisation bulkhead. Did my bob.
@David
Thank you for that extraordinarily detailed set of observations. My speculation as to a possible cause for that sort of damage was really just based on similarities between the internal damage and that seen with JAL 123. I think that it is safe to say that the amount of energy required to create a piece of wreckage like Item 22 was very high.
Two accidents are probably worth considering in contrast; Ethiopian Airlines flight 961 and (as has been raised by Ge Rijn previously) Asiana flight 214. Although only the former was an on-water impact, in both cases the tail section separated from the fuselage (in OK214’s case, such was the violence of the impact, after the tail section separated, both horizontal stabilisers and the vertical stabiliser subsequently separated from the tail cone). Ethiopian 961’s vertical stabiliser was essentially undamaged while Asiana 214’s vertical stabiliser suffered strike damage that is in no way consistent with what we see with Item 22.
@Mick Gilbert:
“Moreover, there’s no evidence that he had deleted his internet browser history; the leaked RMP report references internet searches dating back to 20 December 2013.”
I’m not a big fan of the pilot trying to hide the plane in a deep trench in the SIO. However, the absence of references to browser activity presented in the RMP report proves nothing. He might have searched on another device, or searched on his computer but the RMP did not include the evidence, or not searched at all, preferring to use Google Earth or another method.
“Then there’s the yawning dichotomy with regards to his alleged desire to render the airplane electronically dark so quickly that he’s disconnected two of the airplane’s four operating generators from and isolated their respective buses yet he’s elected to blithely pilot the airplane back across the Malay Peninsula by choosing a route straight past two radars, one civilian and one shared, that he knew were operational, past another two military surveillance radars, that he had no reason to believe weren’t operational, and within 10 miles of Malaysia’s third busiest airport, all at high level.”
He may have intended the airplane to appear on radar as an unidentified airplane flying first towards Kota Bharu and later Penang International Airports. Whatever he did, he succeeded in preventing an intercept over Malaysia.
@Andrew: Thank you for supplying the hard facts, as always. Just to be clear, can you confirm that in this image, the electrical synoptic is not correct for the switch configuration that is shown?
@Andrew,
Re: “The SAARU transmits data to the C Flight Controls ARINC 629 Bus and the ADIRU receives data from the same bus, but what makes you think the ADIRU uses SAARU data?”
According to the following diagram:
https://www.dropbox.com/s/yhnvnvehcy0gz4f/adiru_scheme.png?dl=0
the central ARINC629 FCM receives some data from the ARINC629 bus and ADIRU’s processor’s FCA, and sends some data back. What data are sent back to the processor? If the central FCM is needed only for input, then why does it receive output data from the processor if not for comparative purpose?
Re: SAARU data goes to the Left and right AIMS cabinets
Can the SAARU’s data reach the SDU through ARINC 429?
@Victor
RE: “Just to be clear, can you confirm that in this image, the electrical synoptic is not correct for the switch configuration that is shown?”
The synoptic in the image you posted seems to be incorrect for the switch configuration. According to the AMM:
“AC GENERATION -BACKUP POWER -BREAKER CONTROL
TBB and CCB Control
The backup generator converter controls the automatic operation of the TBB and CCB. The converter gets a switch position signal from the backup generator (BACKUP GEN) switch.
When the switch is in, the converter controls the TBB and CCB. The TBB and CCB are normally not energized. A spring holds the TBB closed when it is not energized. A spring holds the CCB open when it is not energized.
The converter energizes the TBB and CCB to permit the backup system to supply power to the transfer buses. The TBB opens when it energizes. The CCB closes when it energizes.
When the switch is out, the TBB and CCB are relaxed.”
In other words, when the BACKUP GEN switch is out (ie OFF), the CCB and TBB are both de-energized. In that state, the CCB is open and the TBB is closed. The following diagram shows how the system is configured when the CCB and TBB are de-energized: B777 Backup Generator Power System
@Donald said, “I think those that are using the 15,000 fpm BFO to conclude that a high speed impact occurred should perhaps be spending more energy on EOF scenarios that would have an alive pilot in a high speed dive and subsequent pullout. Just a suggestion.”
I don’t think this is likely, but it’s certainly physically possible. A pilot could increase the descent rate to 15,000 fpm and then level off and lose only about 4,000 ft of altitude without overloading the wings. If the plane was initially at 35,000 ft, a 3 deg glide after the initial plunge would put the plane out 120 NM. Even if the debris could be definitively linked to a high speed impact, a high rate of a descent could have occurred after some distance of glide. That’s not a scenario that prospective searchers want to think about because of the implications on the size of the search area.
@Andrew: Yes, that makes it even more clear. The behavior in the PMDG 777 model is incorrect. You are an incredible wealth of information. Thank you again for participating here.
@Victor: You’re welcome!
@Oleksandr
I’m a bit pressed for time at the moment. I’ll have a look at it tomorrow and get back to you.
@Victor
Another “victor” in this accident, besides yourself and ocean science, should be PMDG if they are interested in an accurate fuel model and other fixes. I am up to 3 victors. Hoping for more.
https://www.wired.com/2017/05/victor-search-mh370-ocean-science/
@Victor
Re: Assumptions- The BFO trend should be curved for level fight
Measured BFO is as follows (almost perfect straight line):
Time 1941 BFO=111? (Apparent maneuver in progress)
Time 2041 BFO=141 (Level flight)
Time 2142 BFO=168 (Level flight)
Expected Level Flight BFO’s at M.84
Time 1941 BFO=103 (This would be straight level flight BFO)
Time 2041 BFO=141 (Level flight)
Time 2142 BFO=168 (Level flight)
There should be a curve in BFO since the satellite is closest between Arc2 and Arc3. Therefore straight line BFO trend implies the 1941 BFO is impacted by a maneuver (it’s the FMT off L896 in my 180S path).
Maneuver at 1941 implies the 1840 mark is the start of a period of maneuvers (possibly flying around behind Great Nicobar Island and heading south to ISBIX via NISOK/L896).
@David @Mick Gilbert
On your views of the H.stabiliser piece (no.22) I first like to add another series of photos:
http://www.dailymail.co.uk/news/article-3775809/The-red-white-aircraft-pieces-match-missing-MH370-jet-prove-Malaysian-airliner-exploded-
mid-air.html
and:
http://www.airliners.net/photo/Air-India/Boeing-777-222-ER/734621
I think there is an important detail you don’t mention. Piece 22 was located behind the non-metalic leading edge cover of the HF-antenna.
I think this can be considered as being a weaker spot in the H.stabiliser’s leading edge.
In the latter url-picture you can see this leading edge antenna cover clearly.
Maybe this cover seperated or got damaged first exposing the area behind it to air or water forces ripping the piece out.
The piece seems to be ripped out starting from the front tearing off to the back bending outwards which caused the tension cracks in the backside of the piece. Also the bends in the auxilarry spar you mention show this ‘from the inside blown out’ sequence imo. As does the wide inwards bend in the outer skin. As far as I can see the alu-honeycomb shows no obvious compression damage.
Regarding Asiana 214 I like to mention again that in this H.stabiliser a similar piece at the same spot was ripped out.
Indicating to me no high speed impact is necessary to cause this kind of damage/seperation.
Found some good hi-res photos of the Asiana 214 H.stabiliser. They show this damage was caused right after impact:
http://www.bianoti.com/asiana-airlines-flight-214-tail.html
@TBill
I fail to see the logic in your conclusion that the 111Hz BFO value at 19:41 is somehow indicative of a maneuver. The 111Hz BFO can be achieved in straight and level flight over a wide range of aircraft speeds. From 410 knots to the maximum speed to the aircraft.
Take a look at Figure 5.7 in “Bayesian Methods…”
@DennisW
Sure I take a look, but show me a single flight path with 111 BFO that is also straight thru Arc5. For example, CCYap (IG?) spreadsheet adds 300 ft/sec ascent at 1941 to get 111 BFO, otherwise he is at 104 BFO. You and Freddie do not want to say BFO’s for Freddie’s path, but you are taking a huge slow down just as you approach 1941 Arc2.
@Mick
Got you on this one.
The whole reason I am using Microsoft Flight Sim 9 with the PSS 777 model is to see the ocean the way Z might have seen it, with the same tools he might have used. One thing I was wondering if MH370 could be directed to deep ocean features like we always see on Google Earth depictions of the the MH370 flight path.
Ureka moment for me was, yes, it is easy to monitor FS9 flight path in real time on Google Earth. The common shareware flight path tool “Flight Sim Commander” does this, and that’s fairly old software. Most of my paths feature the idea of shooting for a deep trench or deep water.
The only question I have is, could the perpetrator have had similar software on his laptop during the final flight itself? Can the lat/long position info be read out to Google Earth on a real life flight?
In regard to this Vertical stabiliser piece, (i.e, this piece);
https://d31fjbthwxlyse.cloudfront.net/blogs.dir/10/files/2016/08/Mike-Exner-overlay-of-MH370-fragment-e1472245605940.jpg
it is as Ge Rijn opines this piece “can be considered as being a weaker spot” within the stabiliser.
We can also see, from this picture (MH17, of course);
https://lh4.googleusercontent.com/proxy/27gm9GrjvzuXtkx9d4ryWImxvsneQVy97aDx1BUK2SBfwxaXgabrE3PkbutWc-fTrYOkpVPGSWHSyZ2dDubtCWgQ7oZAc40vU2EsRbtb_BObQzYXpPA0r-NOsGTbcLVs_vgWgXrwpHMp7cw6ozfXSLArZy4ZEMWenqGZXGVdRkevjamxWxSFcyqoV0EVd5YtZ-9DzxTf2zw=s0-d
that the vertical stabiliser (except the missing piece) is a robust assembly, as Mick asserted.
(The vertical stabiliser also remained attached to the {just visible partially remaining} hull).
Readers will form their own view as to whether the ‘shattering’ of the MH370 piece may have been caused
by impact with the water (e.g. either airframe {including tail} concertinaed as the airframe impacted
the water or inverted tail impacting against the water), or whether aerodynamic forces in a high speed
dive could have stressed and ripped the MH370 piece from the stabiliser.
(Note; the actual ‘MH17 piece’ was not recovered,
https://shinestore.blob.core.windows.net/shine-lawyers-media/2014/09/Shine-Lawyers-MH17-Lawyers.jpg
so it seems to me the MH17 piece was probably not detached when the tail struck the ground, otherwise
the MH17 piece would have been close enough to the tail to have been recovered – so I view it as more
probable that the MH17 piece is likely to have detached in the air. (I’ll allow that possibly the MH17
piece could have been detached as a result of the tail passing through the decreasing over-pressure
of the warhead blast that initiated about above the cockpit – more likely, however, that the MH17 piece,
like the missing MH17 rudder, detached from the tail as a result of aerodynamic forces the tail
experienced as it tumbled down from ~33000 feet.)
_____________________________
@DennisW
(To reply to your question),
I would hasten to add that the accuracy in this case is substantially less than tens of metres, due
to only a single base station registering the mobile (rather than several base stations).
The method used to determine the distance is not plainly stated in the RMP report. A probable method
used may have been the ‘Round Trip Time’ (probably UTMS) – this reference suggests to an accuracy of
upwards to 440m (‘Cell ID + RTT’ method);
http://www2.ucy.ac.cy/~laoudias/pages/penek/deliverables/D3.pdf
The above document, you may notice, cites as the source of the Cell ID + RTT method, this reference;
http://research.ac.upc.edu/EW2004/papers/63.pdf
wherein may be seen a similar figure of e.g. 436m, or e.g.
“Theoretical analysis of geometry shows that the accuracy of the hybrid Cell ID + RTT
positioning method depends heavily on the network topology together with mobile location,
and varies from 16m to almost 440m as a function of these parameters.”
TBill says:
…”you are taking a huge slow down just as you approach 1941 Arc2.”
Couldn’t it be possible that the ‘slow down‘ occured as MH370 actioned an ~1841
FMT, due to MH370 shedding altitude (for instance, as a result of the turn) – the time
MH370 took to get to 1941 Arc 2 could thereby be accounted as time taken to transit
‘a horizontal distance + a vertical vector’ (rather than merely only a
horizontal distance). (‘Shedding altitude’ occuring say, as a result of certain C-L & L
hydraulic actuated aircraft control surfaces being non-operational.)
EDIT
@DennisW
‘the accuracy in this case is substantially more than tens of metres’
@buyerninety
Thanks for those pictures. The Mike Exner overlay picture shows the HF-antenna leading edge cover very clear and where the piece ripped out.
And the MH17 picture shows the upper part of the leading edge is gone but broke away on its lower end at the ~same position as the MH370 piece and also as in the Asiana 214 H.stabiliser. All at the HF-antenna cover.
Indeed what caused it to seperate in MH370 remains to be seen.
I only argue a high speed impact is not necessary considering Asiana 214.
And it seems to me the kind of damage doen’t show obvious high speed impact damage either.
But ofcourse only a detailed forensic study and report can give conclusive answers in the end.
@TBill
“show me a single flight path with 111 BFO that is also straight thru Arc5”
I think you are hung up on a track of 180 degrees. Is that what you mean by straight?
“You and Freddie do not want to say BFO’s for Freddie’s path”
I have not computed any BFO’s for Freddie’s path. I always post BFO values for the paths I put in the public domain.
@TBill
The incorporation of Google Earth into the flight sim is an interesting possibility. However, if that was the case, why doesn’t the only piece of evidence that supports any notion of pre-planning, the flight of interest recovered from the Captain’s flight simulator, terminate over a trench or a section of difficult sub-sea terrain?
@Ge Rijn
The damage to the vertical stabiliser of OK214 is very clearly a gouge caused by something striking the tail; as I said previously, it is not consistent with the sort of damage we are seeing with Item 22.
@Victor
Re: “… the absence of references to browser activity presented in the RMP report proves nothing.”
I wasn’t suggesting that the lack of references to the Southern Indian Ocean or oceanographic matters was in any way probative, I was simply suggesting that it might be seen as odd.
With regards to the your hypothetical perpetrator’s intentions and actions, there seems to be quite the mixed bag – he wants the airplane to “disappear” electronically but is happy to place it in a position where it can be tracked easily; he wants to avoid interception but flies straight past the two air force bases that could launch interceptors; he wants to fly to a destination unknown but fires up the only device on the airplane that would allow for it to be tracked, albeit rudimentarily, into such a remote terminus.
The whole event could have unfolded exactly as you believe it did, I have never discounted that, however the scenario is not short on inconsistencies.
@TBill,
Thanks for posting the link to the jet engine primer. I’ll have a look at it.
@Oleksandr,
This might help, again AMM Chapter 34 sourced. The schematic depicts how the SAARU tracks the ADIRU, received on the L & R A629 Flight Control busses, and outputs the data on the C A629 Flight Control bus.
“The SAARU backup navigation outputs go to the autopilot flight
director computers (AFDCs) and the primary flight computers
(PFCs).
@Mick
“flies straight past the two air force bases that could launch interceptors”
In this instance, consider that ‘could’ is a very long way from ‘have operational capability to’.
@Don Thompson
Re: “In this instance, consider that ‘could’ is a very long way from ‘have operational capability to’.”
Absolutely, we’ve discussed Quick Response Alert capabilities previously and I agree wholeheartedly. Did our hypothetical perpetrator know that, though?
If so, why the purported concern about shutting the CVR down while there was the threat of interception? If not, why choose such a vulnerable route? Even if you accept that passing by RMAF Gong Kedak was inevitable why would you proceed all the way to Penang, potentially exposing yourself to tracking by RMAFs Butterworth and Western Hill, interception from Butterworth and late traffic around Penang, before turning up the Strait? Why not track direct to NILAM or IGOGU once you’ve cleared the Thai FIR boundary near 5°39’N 101°E?
As I’ve opined previously, the malicious deliberate action scenario is not short on inconsistencies.
@Mick
I agree that the limited simulator data does not give many clues as to possible exact SIO targets. Others have observed and I agree that the pilot could have branched off from the McMurdo path to go to Cosos, CI or other target. My current thinking is the McMurdo path avoids Indonesian space for a hypothetical flight time earlier in the day. The Beijing flight having less fuel and being after weekend off time for the radar crews, may have had less need to go quite as far clear of Indonesian airspace boundaries.
Another piece by An Higgins in today’s The Australian;
“MH370: Expert tells of data limitations in hunt for plane
The mathematician who led the complex analysis of satellite data in the hunt for Malaysia Airlines flight MH370 has told a select audience of its limitations, saying it only provides “some vague hints about the speed and direction that the aircraft was doing”.
Neil Gordon, who leads the Data and Information Fusion Group in the Defence Science and Technology Organisation, also said in an unreported address last week: “You’re never going to end up with an ‘X marks the spot’.”
Dr Gordon’s remarks to a small gathering of the Institute of Public Administration in Canberra are likely to fuel the international debate over whether the satellite data is good enough to support the Australian Transport Safety Bureau’s theory that MH370 went down in a rapid, unpiloted crash.
The IPA seminar also heard some revealing disclosures from the senior federal public servant in charge of co-ordinating the government agencies involved in the search effort, in which she spoke of “a few bumps along the way” in dealing with representatives of the other two governments involved, China and Malaysia. Judith Zielke, the head of the Joint Agency Coordination Centre, told the seminar: “The biggest thing that I have learnt from being involved in the search is actually the huge cultural differences between all the countries involved.”
The ATSB’s head of the underwater search, Peter Foley, also addressed the event, telling of the fears he had that someone on the search vessels would be seriously injured or killed in the treacherous seas of the southern Indian Ocean where the agency had defined a 120,000sq km search area.
MH370 disappeared with 239 people on board on March 8, 2014 on a scheduled trip from Kuala Lumpur to Beijing, with its radar transponder turned off 40 minutes into the flight and radio contact cut.
Analysis of hourly automatic electronic “handshakes” from the aircraft to ground stations via an Inmarsat satellite indicated the Boeing 777 came down somewhere along a band in the southern Indian Ocean.
That assessment, Dr Gordon said, came from one type of data measuring the time the signals took to get to and from the aircraft to the satellite and the ground station.
“There’s no data attached to those communications, there is no ‘I am here’ type information, but there is some metadata attached to those which you can use in a way it’s not, in a sense, intended for, to produce some predictions,” Dr Gordon said.
In addition to that, Dr Gordon said, “It turns out there’s another piece of metadata attached to this which you can use to give some vague hints about the speed and direction that the aircraft was doing.”
That data measuring a Doppler effect was used by the ATSB to conclude MH370 went down in what’s become known as a “death dive”, the theory the bureau relied on to define its search area.
In its public statements, the ATSB, under pressure over the failure of the search to find the aircraft, has expressed confidence in this conclusion.
In a recent statement criticising The Australian for reporting critiques of the search strategy by international pilots, aviation experts, air crash investigators and scientists and the counter view of many of them that a rogue pilot hijacked MH370 and flew it outside the search area, ATSB communications manager Carl Fellows wrote:
“The metadata from the aircraft’s satellite communications system has been painstakingly analysed by leading experts … this analysis concludes that the aircraft was in a high and increasing rate of descent at the end of flight.”
ATSB chief commissioner Greg Hood has refused, despite calls from families of the MH370 victims in Australia and around the world, to grant a freedom of information request from The Australian seeking what the bureau claims to be those supporting analyses.
Many international experts disagree with the ATSB’s conclusions, with former airline pilot and top air crash investigator John Cox telling The Australian: “I do not believe there is sufficient data in the Inmarsat data to draw any conclusion on the rate of descent.”
Against the ATSB’s bold statements for public consumption, Dr Gordon’s remarks to the IPA audience were more cautious.
“You have got these things which in a sense have never been used for this purpose before, so you put a big effort into checking that you understand them correctly,” he said.
“The key thing you have to remember is there’s lots of uncertainty that’s feeding into this.”
ATSB spokesman Dan O’Malley denied the bureau exaggerated, saying: “There are a range of possible rates of descent, all of which are high and increasing.””
Mmm … Not just “An Higgins”, a very particular Higgins, Ean.
@TBill
@Victor
Having just re-read Victor’s and Yves’ “Further Analysis of Simulator Data Related to MH370” paper, I note that the airplane heading at or just after fuel exhaustion was 178.2° (presumably Magnetic). I’m wondering whether a simple turn onto a CMH of 180° off N571 near 89°14′ might produce a similar terminus?
@Mick
“As I’ve opined previously, the malicious deliberate action scenario is not short on inconsistencies.”
As I have opined previously and often, the route selected was done for two purposes.
1> Maximize the time for negotiations by minimizing the time between take off and diversion.
2> Fly a route where the diversion can be obviously confirmed.
If Shah wanted to sneak into the SIO, a flight to the West (Europe or Jeddah) would have been much preferred. He obviously intended that the diversion be known and known early on.
@Mick Gilbert said, “The whole event could have unfolded exactly as you believe it did, I have never discounted that, however the scenario is not short on inconsistencies.”
I don’t know what you mean by “unfolded exactly as [I] believe it did” because I don’t have a specific sequence of events that I “believe”. Here is what I do believe is likely:
1. The disappearance is very unlikely due to one or more failures. The sequence of failures and non-failures required for a flight to follow the path back over Malaysia, up the Malacca Strait, and into the SIO, without attempting a landing at Penang, and without any communications, is very improbable, in my opinion.
2. The captain likely planned the diversion, as the simulator data suggests. I think it is very improbable that the deleted files found on his computer, which were snapshots of a simulated flight from KLIA, up the Malacca Strait, and ending with fuel exhaustion in the SIO, were not related to the disappearance. I also don’t think it is normal for a captain of a B777 to be using WeChat from the cockpit while lined up on the runway one minute before takeoff.
Other than that, I am open to lots of possibilities, including negotiations during the flight that went sideways as DennisW suggests (in which case the CVR doesn’t matter), or a pilot suicide accompanied by a desire to hide the crime (in which case the CVR does matter). The reason I am open to different possibilities is because I acknowledge that I have not seen a scenario that is not without its inconsistencies. I think part of the reason that things don’t add up is Malaysia has additional information that has been withheld from the public, and probably also the ATSB.
@Mick Gilbert said, “Having just re-read Victor’s and Yves’ ‘Further Analysis of Simulator Data Related to MH370’ paper, I note that the airplane heading at or just after fuel exhaustion was 178.2° (presumably Magnetic). I’m wondering whether a simple turn onto a CMH of 180° off N571 near 89°14′ might produce a similar terminus?”
I believe you are referring to the simulator data for data set 45S1.
First, the heading is true, not magnetic.
If you look at the data in Table 2, you’ll see that the plane was banked at 10.9° to the right and turning. As such, you can’t look at the heading (or even the track), and draw conclusions about the path prior to this data set.
There is also the remarkable coincidence that points 10N and 45S1 align almost exactly with McMurdo Station, Antarctica. I’ve yet to hear an explanation how this could have occurred other than McMurdo Station was chosen as a distant, unreachable waypoint.
@Mick
I had never said a critical word re: ATSB but I am not sure I agree now with Gordon above. I am felling like a better analysis of BTO/BFO might have pinpointed a better spot to look than we ended up getting.
@Mick Gilbert: Thanks for posting Ean Higgin’s new article.
When John Cox says, “I do not believe there is sufficient data in the Inmarsat data to draw any conclusion on the rate of descent,” I don’t think he understands that the vertical speed is not compensated in the Doppler correction algorithm in the AES, and therefore the BFO is very sensitive to vertical speed. To make the claims he does, he should explain why he believes the BFO is not sensitive to vertical speed.
@TBill
“I am felling like a better analysis of BTO/BFO might have pinpointed a better spot to look than we ended up getting.”
Good grief, Bill. A lot of really really smart people have beat the BTO/BFO data to death. In fact, they have gone beyond what is prudent IMO. The ISAT data cannot tell us where the aircraft terminated. You seem fixated on the fact that we are missing something. We are not. The ISAT data is under constrained plain and simple.
@DennisW
I’m not familiar with your negotiations hypothesis, might I please trouble you to expand on negotiations with whom? for what? and by what means? Thank you.
@Mick
My thesis is simple. Shah diverted the aircraft in order for people on the ground in KL to negotiate with Najib to return the money he embezzled from the the Malay Sovereign Wealth Fund (over a billion USD). This deed (the embezzlement) is well documented. The return would be by electronic fund transfer (which is not readily reversible), and is quite easy to do (and verifiable) with online access to the proper accounts.
Negotiating for concessions relative to Anwar is ridiculous. The Malay’s could agree to anything and then do nothing when the plane and passengers were secure.
@Mick
BTW, my theory is consistent with everything we know. I do not take Victor’s stance of all theories having inconsistencies.
@DennisW
Okey doke, thanks Dennis. Do you think the timing (dark o’clock on a Saturday morning) might have presented some problems with regards to (a) making contact with Najib Razak and (b) him being able to effect a money transfer of that magnitude?
@Victor
Thank you for the clarifications regarding the sim data, Victor. The coincidence of the McMurdo track notwithstanding I’d be interested to know whether a simple turn onto a CMH of 180° off N571 near 89°14′ might produce a similar terminus? I do not have the tools to run that analysis.
@Mick
It is easy to run that case on MSFS but the hard part is winds. You have to say if you are looking for wind effects. Spell out the case.
As far as Dennis’s theory, there are others who have that theory based on political negotiations. I believe there was an email claim early on. One person over on JW blog actually claims to have second hand knowledge from his contacts in MY. Dennis has other believer’s on the theory, except Dennis has concluded only cash payout would make sense. The overall theory is that there was loiter and plan to land on Christmas Island or Jakarta, and those advocates are thinking the crash is way up north by Java. Actually some recent rehash on JW site in recent days.
@TBill
Thanks for that, if you could run it as “Clear Skies” that would create a good base case.
Thank you also for the expansion on the negotiation hypothesis; for better or for worse, I gave up reading comments on JW’s blog a long time ago.
@Ge Rijn. I agree with Mick that the ASIANA 214 fin damage most likely is from hard object impact. There would be inboard flaps, engine cowls and other wing and engine debris flying up well short of the threshold.
It is hard to see how the leading edge was attached though thank you for the Daily Mail photo 4. However for it to be a weak link suggests it is under designed. It should cope with impact damage for example at least as well as the skin, being struck by hail and birds more directly.
I think any account as to what happened with MH370 has to explain the auxiliary spar bend. Mick’s internal air compression theory would do that except that I think the curvature is too great to be forced by a pressure pulse. Also from your ASIANA pics it looks like there is fuselage skin sealing the fin/vertical stabiliser (Gysbreght’s question) and there is no obvious reason why a pulse would cause that in particular to yield.
@Buyerninety. MH17’s rear fuselage seems to have detached from the wing/centre section before they hit the ground. It would have tumbled violently surely on so doing, with air loads quite possibly separating the rudder and the leading edge. I have my doubts that before this there was a high speed, the combination being very draggy and not heading straight downwards by any means after the Bukh impact.
That said I agree with Ge Rijn that it is curious that the bottom of the MH17 piece comes from a similar position to that from MH370.
What might support high speed aerodynamic loading as the cause is the No Step piece which separated from the MH370 right stabiliser.
Still, as above I think the bend in the auxiliary spar needs explanation.
@Victor. “Even if the debris could be definitively linked to a high speed impact, a high rate of a descent could have occurred after some distance of glide. That’s not a scenario that prospective searchers want to think about because of the implications on the size of the search area”.
That should have influenced the conclusion of the Review, about the further search, which was, “The experts concluded that, if this area were to be searched, prospective areas for locating the aircraft wreckage, based on all the analysis to date, would be exhausted”.
That conclusion looks misleading.
@Dennis
>BTW, my theory is consistent with everything we know. I do not take Victor’s stance of all theories having inconsistencies.
Not withstanding (and I’m being generous as the inconsistencies are numerous) the aircraft all but disappearing and 239 human beings losing their lives because (according to you) the captain ran out of fuel…or whatever your reason du jour for the eventual outcome currently is?
There are so many inconsistencies with the negotiation theory that it’s difficult to know where to begin.
Did everyone here hear about the tragic suicide of Chris Cornell, the 52 year old frontman for the band Soundgarden? Came out of nowhere. In fact, he gave a wonderful live performance the night before he hung himself.
By all accounts, he was ‘thriving’. Left behind children and a wife. The fifties can be terribly difficult years from which to emerge alive and intact from, for certain people. Just food for thought.
@David
@Gysbreght
Re: “… from your ASIANA pics it looks like there is fuselage skin sealing the fin/vertical stabiliser (Gysbreght’s question) …”
I don’t think that it is the case that the vertical stabiliser is sealed off from the unpressurised rear fuselage section although publicly available drawings and schematics that I have been able to access are inconclusive in that regard. From a practical perspective there would have
to be runs for hydraulics and electrical cabling and there is no logical reason to suspect that they would be air tight.
Being unpressurised, the vertical stabiliser would have to be able to “breathe” and in order to prevent a pressure differential developing between the rear fuselage section and the vertical stabiliser it would make sense for there to be no obstruction to air flow between the two.
@ALSM
Mike, you’ve been the go-to guy on airframe structures; can you assist?
@David
My argument is mostly about the damage that occured on the Asiana 214 H.stabiliser happened during a relatively low speed crash landing.
It well could have been caused by debris from wings and engines hitting that leading edge during the crash as you mention.
I argue the same could have caused the seperation of piece no.22 during a ditch-like impact.
I don’t think the HF-antenna cover is an underdesigned weak spot in the leading edge but a ‘weaker spot’ for it’s different non-metalic material and it’s fitted into the leading edge making sides and edges along which a breakage is more prone to happen under stresses it’s not designed for.
And I like to mention the whole piece along it’s auxilairy spar has this distinqtive bend/curve indicating to me the piece was forced outwards by air or water forces, first bending in that curve and than breaking at the ends of this auxilairy spar tearing away backwards and ripping from the main leading edge spar from its fasteners. I think this sequence also caused the tension cracks in the backside of the piece leaving the probably more flexible outer skin relatively undamaged.
@Oleksandr
RE: “According to the following diagram [ ] the central ARINC629 FCM receives some data from the ARINC629 bus and ADIRU’s processor’s FCA, and sends some data back. What data are sent back to the processor? If the central FCM is needed only for input, then why does it receive output data from the processor if not for comparative purpose?”
The ADIRU receives data from the following components via the Centre ARINC 629 FCA:
Centre Pitot ADM
Centre Static ADM
Centre PFC
That data is sent to the Processor FCA, where it is processed with data from the Gyro FCA, the Accelerometer FCA, and the Left and Right ARINC 629 FCAs. The processed ADIRU data is then sent back to the Left, Centre and Right ARINC 629 FCAs. The Left and Right ARINC 629 FCAs send the ADIRU data to user systems via the Left and Right Flight Controls ARINC 629 buses.
It’s not stated in the manuals, but I suspect the ADIRU data is sent back to the Centre ARINC 629 FCA to increase redundancy and improve the ADIRU’s fault tolerance. The ARINC 629 FCMs all vote on the data they receive from the four processors to ensure that consistent data is output on the ARINC 629 buses. The aircraft can dispatch with only one of the two ARINC 629 FCMs working in each ARINC 629 FCA (ie three FCMs out of six), so perhaps the Centre ARINC 629 FCA provides an extra layer of redundancy for data voting in case there is a subsequent failure of the remaining Left or Right ARINC 629 FCM.
As Don Thompson said earlier, the SAARU receives ADIRU data from the Left & Right Flight Controls ARINC 629 buses. If the ADIRU data is valid, the SAARU tracks the ADIRU and the SAARU output data is the same as the ADIRU. If the ADIRU data goes invalid, the SAARU calculates data independently of the ADIRU and outputs it on the Centre Flight Controls ARINC 629 bus.
RE: “Can the SAARU’s data reach the SDU through ARINC 429?”
No, the SDU gets aircraft position information from the left and right AIMS cabinets. That data comes from the ADIRU and is converted to ARINC 429 by the DCGF within each AIMS cabinet.
@Mick Gilbert, all
Concerning the IPAA ACT seminar. I found some time to watch the entire 80 minute seminar presentation (It’s posted on IPAA ACT’s channel on Vimeo).
I do wonder whether Ean Higgins watch the entire 80 minutes? I am much more concerned over the comments made by Alan Lloyd of AMSA (and a few by Peter Foley).
@TBill, @Mick Gilbert: If you are going to reproduce the run in FS9, you should choose “Fair Weather”, which is what Zaharie Shah chose when he ran the simulation.
@David: My comment regarding the end-of-flight with a steep descent, then a glide, then a steep descent, referred to a scenario with active pilot input. I don’t believe that the ATSB considered that possibility, but I don’t know.
@Victor
Regarding glide scenarios, while I very much doubt that the ATSB considered a specific pilot-controlled dive-glide-dive-glide scenario, they most assuredly did consider and define a 60 nm low priority band either side of the 20 nm high priority and 20 nm medium priority bands to encompass “… the possible but less probable controlled glide scenario.” Search area width, p.14, MH370 – Definition of Underwater Search Areas, 3 December 2015
@Mick Gilbert: I was referring to the more recent Boeing simulations that were summarized in the ATSB report dated Nov 2, 2016, entitled “MH370 – Search and Debris Examination Update”, which made recommendations based on uncontrolled descents.
@Don Thompson
Don, I agree, Alan Lloyd from AMSA was talking through his hat on some matters, his references to AF447 and floating debris recovery were wrong in almost every regard.
Apart from his choice of tie, I thought Peter Foley was okay; which of his comments concerned you?
Some of the information on Alan Lloyd’s slide about the surface search was instructive – 345 sorties, 3177 search hours, 4.7 million sq km. 9.2 search hours per sortie makes no sense particularly as he himself describes a typical sortie as 4 hours transit, 3 hours searching, 4 hours back. It may have been 9.2 flying hours per sortie but actual search hours couldn’t have been much over 3.5 hours on average (the number would obviously vary across the mixed fleet of P-8 Poseidons, P-3 Orions and C-130 and the task areas). But if you accept 3.5 search hours per sortie as there or thereabouts that gives 1207.5 search hours for 4.7 million sq km – 3392 sq km/hour. What’s the fastest you can reasonably cruise while conducting a surface search?
@Victor
Ah, my apologies.
@VictorI
I think your suggestion of a dive-glide-dive scenario with an active pilot is a kind of break-through/change in the common thinking of the IG over the years.
Am I right about this?
If so, I’m glad about it for it finally opens up a serious look at this kind of scenarios on another level.
I would support a dive-glide-impact scenario but this doesn’t matter in this regard. Only to consider what an active pilot could have done not violating the data matters in this regard I think.
As @Mick states a 60Nm range was considered. They never searched beyond 40Nm. And those max 60Nm were still based on an inactive pilot scenario.
Probably according the ~20 minutes yellow trajectory of the Boeing simulations that reached (imo) over 50Nm beyond the 7th arc to the East.
I completely agree an active pilot scenario including a glide would be not appealing and considered unmanagable by the officials but ofcourse no reason to deny this (imo obvious) possibility.
To look at this in other way there’s only a need to change perspective which could open-up other opportunities.
For this to happen seriously by the way, it’s crucial to get conclusive evidence on how the debris separated and got damaged imo.
And I surely like to hear @ALSM’s opinion on the current debris-discussion too.
@Ge Rijn
Re: “And those max 60Nm were still based on an inactive pilot scenario.”
No, that’s not correct; the glide scenarios contemplated pilot control.
@all
I think Jeff Wise put a quite interesting topic on his site about debris damage and specifically the nose gear door for all to read.
With thanks to Jeff:
https://jeffwise.net/2017/05/23/reading-the-secrets-of-mh370-debris/comment-page-1/#comment-206505
@Mick Gilbert
If you can provide an official statement which imply pilot controlled scenarios I will believe you.
@Ge Rijn
Re: “If you can provide an official statement which imply pilot controlled scenarios I will believe you.”
It’s spelled out in the section of the ATSB’s MH370 – Definition of Underwater Search Areas, 3 December 2015, I referenced earlier; specifically the section dealing with Search area width, p.14. The following excerpt describes how the 100 nm wide search area was developed;
“Glide area
A simulation was performed to determine the glide distance of the aircraft under active control to maintain wings-level attitude. The simulation (from FL330) resulted in the aircraft gliding for a total distance of approximately 125 NM from the point of the second engine flame-out.
In order to make this distance the aircraft would travel approximately 15 NM in the first 2 minutes of the descent (approximate time required to start the APU and initiate the log-on sequence).
Therefore, from the 7th arc, the aircraft has the potential to glide around 110 NM. Due to the initial direction of travel and the wind conditions on the day, around 100 NM is a more realistic value.”
“… the aircraft under active control … ” – it can’t be any clearer than that.
The assignment of a low priority to the bands associated with a pilot-controlled glide are discussed a number of times in that section and summarised in the caption to Figure 7 on p.15;
“… Note that due to assumptions made by the analysis, Priority areas 3 and 4 are significantly lower probability as they rely entirely on a controlled glide flight scenario.”
It is a common misconception that the ATSB never considered pilot control at end of flight; they most assuredly did, going so far as to factoring it into their search area width calculations. However, they assigned the 40-100 NM bands either side of the 7th arc a low priority because there was no evidence to suggest that the aircraft was under controlled flight after the FMT.
@Donald
“There are so many inconsistencies with the negotiation theory that it’s difficult to know where to begin.”
So you decided not to begin? 🙂
@Mick Gilbert
Thanks for the response. They were obviously aware of the possibility of an active pilot and glide.
I did not read/understand it before so clearly.
@DennisW
Call me a dreamer, but in my career I worked a lot with small experimental data sets – testing experimental catalysts in the lab at a couple temps, couple pressures.
The aircraft BTO/BFO data looks “accurate” to me (not yet sure about the sat calls). The 1941 BFO looks to me like the data “sample” was taken in the middle of a condition change (maneuvers). The next 3 pings 20:41, 21:41, 22:41 potentially point unambiguously to the MH370 crash site.
Mostly I am hearing a rhetoric about how bad BFO data is, followed by proposals of flight paths to do not honor the key BFO data above. DrB’s path does a pretty good job honoring the BTO/BFO data, but I think there were maneuvers after 1840 up to at least around 1941 (almost self-evident in the data).
Somebody kindly give me a path that meets BTO/BFO above and later we can worry about whether or not to throw out the predicted “x marks the spot”.
@DennisW: What’s your current thinking about why the plane passed the airport at Christmas Island and ran out of fuel?
Mick Gilbert:
WRT the vertical stab…I don’t have any Boeing documentation indicating that the vertical stab is sealed or not. However, I can’t think of a single good reason to attempt to seal it, while I can think of many, many reasons not to (attempt to) seal it.
First, any attempt to seal something as large as the vertical stab is almost certain to fail, if not immediately off the production line, then surely withing a few flights. There are simply too many holes to plug. Over my 40+ yr career, I designed many products that had to live outdoors in a harsh environment. I have a lot of scar tissue from all the attempts to achieve IP67 enclosures that last…and that was for products operating at nominal ambient pressure. Probably the single most important lesson learned from all that brain damage was that if you cannot make an enclosure absolutely 100% hermetic (forever), it will accumulate liquid water inside over time. A small leak in a “nearly sealed enclosure” will act like a “diode”, a one way valve, to moisture in the air. The water vapour goes in, condenses inside, and never comes out. (How would you like to open up a fin and find 100 gallons of H2O?) Eventually, whatever is inside the enclosure fails due to the moisture and corrosion. Not a good idea.
Second, the structure would need to be much heaver to withstand the pressure differences. I can’t imaging the added weight would be justified whatever purpose one might have in mind for sealing.
Third: Maintenance headache. I can’t imagine how inspections of the internal structure could take place without constantly breaking seals. Makes no sense.
@Andrew,
Thank you for your comments.
Re: “I suspect the ADIRU data is sent back to the Centre ARINC 629 FCA to increase redundancy and improve the ADIRU’s fault tolerance.”
How can this help to increase redundancy if the Centre ARINC 629 FCA sends data only back to the Processor FCA?
Re: “No, the SDU gets aircraft position information from the left and right AIMS cabinets. That data comes from the ADIRU and is converted to ARINC 429 by the DCGF within each AIMS cabinet.”
What I meant is not direct connection, but via DCGF, similarly to the ADIRU. I.e.: SAARU->ARINC629->AIMS(DCGF)->ARINC429->SDU.
In your earlier post you mentioned that data from the SAARU are sent to both the AIMS. Hence my question. Where next the data from the SAARU are sent to?
@Mick Gilbert
@David
@Gysbreght
@ALSM
RE: “Do you know if there is fuselage skin between the space inside the vertical stabilizer and the space in the rear fuselage below it?”
The following photos of aircraft under construction on the Boeing factory floor might help answer the question:
B777F Rear Fuselage
B737MAX Rear Fuselage
(Not a B777, but you get the general idea)
@Victor
“@DennisW: What’s your current thinking about why the plane passed the airport at Christmas Island and ran out of fuel?”
My opinion is based on long range “opposition party” thinking. To land the plane anywhere after negotiations failed would be giving in to Najib’s “calling the bluff”. The diversion was simply a small step in a long process of restoring Malay democracy. At that point Shah was a martyr to the cause. While he did not initiate the action with that intended outcome, he knew full well that it was a possibility. Giving in would send the wrong message to the establishment – that is, the opposition party ultimatums did not need to be taken seriously.
A parallel is the employee who walks into your office with a “I need a raise or I quit” message. You cannot comply (painful as it sometimes is) because of the precedent it would set. That is the reason why many Silicon Valley companies have transitioned to a focal review process where performance reviews and merit increases are done synchronously for all employees. In that framework managers cannot be blackmailed because increases are only allowed at a fixed time each year. It also defuses the employee asking for the raise since he/she is well aware that you are working with a merit budget, and their exceptional raise is at the expense of lower raises for their colleagues. Seems like a silly process, but it is surprisingly effective.
@Ge Rijn
Re: “Thanks for the response. …”
You’re most welcome.
@Ge Rijn
Re: “Thanks for the response. …”
You’re most welcome.
@ALSM
@Andrew
@David
@Gysbreght
Thanks for the detailed response, Mike, very much appreciated.
Thanks Andrew, those photos are quite instructive. I’m wondering where the cable runs for the electricals, antennae and hydraulicals penetrate the rear fuselage?
@DennisW: OK, thank you. And what’s your explanation for the SATCOM log-on at 18:25?
@Andrew: Thank you for those pictures.
I wonder how the vertical tail is attached to the fuselage structure. They may need to cut some holes in that shiny skin.
@VictorI
“@DennisW: OK, thank you. And what’s your explanation for the SATCOM log-on at 18:25?”
I disagree with the ALSM/DrB explanation that the SDU was powered on, and the transient response miraculously happened to be at the frequency bias which exactly met the expected BFO value. My own explanation is that the SDU was never powered off, and that communications missing before 18:25 were the result of misalignment of the antenna which could occur for a number of reasons. Is it not suspicious that none of the plots in Holland’s paper show a transient response that mirrors what seems to be the accepted view here.
@DennisW: The SATCOM was inoperable from no later than 18:03 to 18:25. What could have caused the misalignment of the antenna over such a long time period? The low-gain antenna (LGA) was also available if the high-gain antenna (HGA) had a problem. I doubt that a drop-out of SATCOM coverage for a duration of at least 22 minutes is a common occurrence.
@Victor, why did you disable the avatar icons ? 🙁
@DennisW. Extortion. I toss in some questions to add to Mick’s and Victor’s:
• Do you think he got the government’s attention by leaving that to the military radar? Would he not have got some feedback if unsuccessful, which seems apparent, and had a plan B like another overfly? He would be aware surely that there was no ready intercept capability?
• As I understood you earlier he would have set course for the SIO when clear there would be no government agreement. Would he do so if the professed reason was that there was insufficient time: agreement, action and then getting confirmation to him – lots of uncertainties there? Yet he seems to have given up and set course for the SIO after just a couple of hours. What would explain that?
• I have asked before what would be the benefit of killing all on board if clear his case had been lost, suicide and slaughter not being his aim? I think your answer was to the effect that it was not so simple. Yet in negotiations how would his accomplice convince anyone that he would go actually go ahead if his case was lost when apparent there would be no gain from such sacrifice? Likewise his accomplice would need to convince them he would not go ahead even if they did comply; and that no-one, including the accomplice, would spill the beans and wreck the government in so doing, even after compliance?
• How would he stop them arresting his family and friends then holding them hostage, as purported accomplices?
• What could be his end game if successful do you think? Surely no hope to land back in KL and return to his family? Land somewhere else, hope to skedaddle and abandon his family?
@VictorI
“@DennisW: The SATCOM was inoperable from no later than 18:03 to 18:25. What could have caused the misalignment of the antenna over such a long time period? The low-gain antenna (LGA) was also available if the high-gain antenna (HGA) had a problem. I doubt that a drop-out of SATCOM coverage for a duration of at least 22 minutes is a common occurrence.”
It is not common. There were no other examples in Holland’s paper. I don’t know specifically what caused the misalignment (or the possibility of the unavailability of aircraft position/velocity info).
@David
The intercept capability is not an issue. Who is going to shoot down a commercial airliner?
The course he chose was still still rife with landing possibilities. That is why the conventional pins in a map at 38S make no sense at all. He was in it until the very end, and then did what was necessary.
@DennisW. The above having not seen your earleir response to Victor.
“The diversion was simply a small step in a long process of restoring Malay democracy.”
Not did it help that?
@Mick. I did not have it in mind that ‘sealing’ of the fin from the fuselage interior would be other than against a surge of air pressure on crashing. That to me would entail a large gush, big lightening holes for example, as distinct from a dribble. There may be sealing against condensation intrusion coming down pipes and looms but not hermetic in purpose. To hermetically seal the fin would be to increase the in flight loadings on it, like the cabin, for no gain that I can imagine.
@Gysbreght. There are holes for the a fin spar to pick up on one fuselage frame in the 737 photo and I imagine the white patch forward would be something to do with picking up the frame there.
@DennisW. In a hurry. ‘How’ in place of ‘not’ please
@DennisW: The drop-out of the SATCOM for 22+ minutes is a rare occurrence. If the diversion was deliberate, then very likely the drop-out of SATCOM coverage was deliberate, or at least a side consequence of some other deliberate action. Yet, in your scenario, there is no need to deliberately power down the left main bus. I consider this to be an inconsistency.
@Mick Gilbert
@Gysbreght
“I’m wondering where the cable runs for the electricals, antennae and hydraulicals penetrate the rear fuselage?”
I don’t know, but there must be a hole somewhere!
“I wonder how the vertical tail is attached to the fuselage structure. “
Again, I don’t know; it’s not obvious from the photo. Most of the vertical stabiliser is made of carbon fibre reinforced plastic.
@Ge Rijn. I understand you better about debris flying up on ditching though the ASIANA example did not result in a missing chunk.
There are two lugs on that bent auxiliary spar piece where the full web still remains (my Daily Mail video) but their purpose is unclear, eg whether they are for bracing. However the web retention nearby indicates a relationship between them and nearby web stress. Most likely that is due to the lugs’ effect on stress distribution. That in turn suggests the web failure was in bending (ie shear) rather than tension from an outwards force.
This piece really should receive a full examination and analysis by a stress man, if it hasn’t.
Dennis:
Re “…and the transient response miraculously happened to be at the frequency bias which exactly met the expected BFO value.” I think you must not understand the relationship between the cold-start transient response and the steady state bias.
First, there is no miracle. In fact, it is quite the opposite. The transient response (as suggested by me and carefully modeled by Bobby) is exactly what is known to exist in virtually all OCXO designs.
Second, the 18:25 observations were not at all consistent with the aircraft state and the steady state BFO bias (~150 Hz). It would take physically impossible maneuvers to explain away the “excess bias”. However, the observations are consistent if the expected warm up transient is added. Knowing that a transient must exist, and fitting the observations to the “excess bias” caused by the thermal transient, is a perfectly valid way to derive the plots Bobby published.
@David
“This piece really should receive a full examination and analysis by a stress man, if it hasn’t.”
I fully agree. It’s one of the key-pieces. I would like to see this piece, among other pieces, being send to independent specialists outside of Malaysia to undergo an independent second opinion/investigation.
Gysbreght. Vertical stabiliser attachment. The method below may well be typical, there being a need for removal in case of severe damage.
From ICAO Doc 9756, 9.8.31, “The composite vertical stabilizer of the A300-600 was designed to retrofit aircraft metal stabilizers, and therefore is attached to the fuselage through the same method as that used for the metal structure. Three pairs of composite lugs (forward, middle, and aft) along the union between the stabilizer and the fuselage, transfer bending and vertical loads through large diameter bolts. Between each pair of lugs is a composite transverse load fitting that transfers lateral and torsional loads from the stabilizer to the fuselage.”
The above is from an account of AA587s vertical stabiliser loss, it breaking off in bending overload, the right aft composite lug breaking in tension, the rest following.
@Ge Rijn. Thanks. Subject change. About that nose wheel door part, debris item 18, found to have failed in tension (see Malaysian Debris Examination), supposing that finding applied to both skins, there is no attachment at each end by which a tensile load could be applied other than the hinges.
Both are missing but it is hard to see how they could apply tension unless there was a separating force between them, ie there was a fracture of the fuselage between them. Doesn’t seem that likely.
Water force from any direction would put the door in bending, only one skin then being in tension, the other in compression.
Forcing up or down against its actuating arm (ALSM’s diagram, JW site) would account for the arm’s attachment being missing but again would not result in tension.
The nosewheel crashing down on breaking its fuselage attachments under high ‘g’ would apply bending and would account for the missing hinges and actuating attachment but again no tension that I can see.
Any thoughts please, or from others?
@David
Imo there is one simple solution that could explain the gear-door seperation, it’s shape and (tension)damage.
I suggested it shortly after the piece was found but this solution was deemed to be too controversial to be taken seriously.
I read in JW’s topic this has changed in the minds of some experts.
When the landing gear was lowered before a low AoA impact the gear door would be deployed creating the situation in which it could break away from its hinges with only tension damage the moment the piece struck the water.
Or the other solution mentioned on JW’s blog: when the landing gear was lowered to serve as an extra speed brake during a high speed descent.
The gear-door could have been ripped out by aerodynamic forces, also only leaving tension-damage.
Any other solution (f.i. high speed nose down impact) has to explain the lack of compression damage but also the undamaged clean side of the piece. Which is hard to do when this gear-door was closed on impact imo.
@David: Thanks for your information>
In AF447 the vertical stabilizer separated at impact and was recovered between floating debris. Its mode of failure offered information as to how the airplane hit the surface, which was confirmed when the FDR was recovered. In that context I am familiar with the AA587 accident. One of these six attachments was tested and failed at a load of 904 KiloNewtons (kN). I believe the B777 vertical tail is attached at four points, at the four corners of the main torsionbox.
@Oleksandr
“How can this help to increase redundancy if the Centre ARINC 629 FCA sends data only back to the Processor FCA?”
First, the Centre ARINC 629 FCA doesn’t send data back to the Processor FCA. It sends data from the Centre Flight Controls ARINC 629 bus to the Processor FCA, where it is processed.
The four Processor FCMs within the Processor FCA send their output data to the Left, Centre and Right ARINC 629 FCAs. The two ARINC 629 FCMs within each ARINC 629 FCA then vote on the data. Any processor FCM whose data does not match the data from the other processor FCMs is faulted and its data is not used.
As I said previously, the aircraft can dispatch with only one ARINC 629 FCM working in each ARINC 629 FCA. Let’s say the aircraft dispatches in that condition and the remaining FCM in the Left ARINC 629 FCA subsequently fails. The AMM differs on this point, but according to Honeywell, the ADIRU remains functional as long as it is able to transmit on any ARINC 629 bus. In this example, the ADIRU can continue to transmit through the Right ARINC 629 FCA. Its remaining FCM will vote on the data it receives from the processor FCMs before transmitting it to the Right Flight Controls ARINC 629 bus. I’m only speculating, because I don’t know exactly how the voting process works, but I suggest that the Centre ARINC 629 FCA is included in that process to provide a ‘second opinion’ before the data is transmitted to the bus.
“In your earlier post you mentioned that data from the SAARU are sent to both the AIMS. Hence my question. Where next the data from the SAARU are sent to?”
The AIMS always sends SAARU data to the standby attitude indicator (where fitted) and to the digital flight data recorder system. It also sends SAARU data to the following systems whenever ADIRU data is not available:
Engine data interface units
Primary display system
Warning electronic system
@Victor
“@DennisW: The drop-out of the SATCOM for 22+ minutes is a rare occurrence. If the diversion was deliberate, then very likely the drop-out of SATCOM coverage was deliberate, or at least a side consequence of some other deliberate action. Yet, in your scenario, there is no need to deliberately power down the left main bus. I consider this to be an inconsistency.”
I agree Shah had no need to power down the left main bus. That is why I don’t think he did power down the left main bus. I also do not believe Shah, or any other pilot would have known about the continued ISAT pings. There is no imaginable reason that would be a part of their training or knowledge base.
@ALSM
Second, the 18:25 observations were not at all consistent with the aircraft state and the steady state BFO bias (~150 Hz).
At 18:25:27 using:
bias = 150.6 Hz
speed = 510 knots (DSTG ground speed estimate)
track = 296 @6.8N 95.9E
yields a predicted BFO of 142Hz (virtually no error).
@DennisW: Any scenario that assumes the captain diverted the plane, yet was not responsible directly or indirectly for the SATCOM drop-out for 22+ minutes, has a major inconsistency. To have an “unforced” drop-out of coverage that coincided with timing of the diversion is very improbable.
I agree that the captain was likely not aware of the underlying signaling of the SATCOM communication, nor the associated BTO and BFO measurements that were recorded. However, the captain would be very aware that isolating the left main bus would end any SATCOM communication, as an EICAS message explicitly states this.
@DennisW: Let me be more specific. It is possible that the captain’s intention was to disable ACARS communications. His “quick and dirty” method was to take down the left bus and power down the SATCOM. Later, when he had more time, he fumbled through the ACARS pages of the CDU and disabled ACARS by unchecking SATCOM communications. He could then re-power the left bus, which put the plane in a safer electrical configuration and also allowed him to monitor and answer calls, if he chose.
@VictorI
Think about the positives for a moment.
1> The choice of the Beijing flight.
If Shah wanted to dump the plane in the SIO as sneakily as possible, the Beijing flight was the almost the worst possible choice. The diversion is obvious early on the radar tracking is good. A flight to Jeddah or Europe would allow the diversion to go virtually unnoticed. The Beijing flight was selected to minimize the time to the diversion and maximize the time available for negotiation.
2> The Malay response.
The Malay response after the turn back at IGARI and for the remainder of the time the flight was in the air is very telling. There was little attempt to communicate with the aircraft. No alerting of authorities in other countries. No attempt to intercept. The Malays (at the top) knew exactly what was going on. There is no other rational explanation.
3> The flow of information
Information flow to the ATSB and to the public has been choked off by the Malays. If they really wanted to find the aircraft, why not make everything available to all parties especially to the ATSB.
It is hard for me to believe that anyone can possibly support a mechanical scenario or a suicide scenario given the above. It is equally hard for me to believe that the DTSG and IG can create paths to the SIO without giving any consideration to motive or causality. What is up with that? I found the attitude bizarre on Duncan’s site and continue to find it bizarre relative to respected contributors here – ALSM, DrB, Richard Cole,..
DennisW is probably correct that MH370 stayed within reach of airports versus heading out to a far West oblivion, probably there was also a loiter
@Andrew,
“First, the Centre ARINC 629 FCA doesn’t send data back to the Processor FCA. It sends data from the Centre Flight Controls ARINC 629 bus to the Processor FCA, where it is processed. ”
If this was true, then there would be no reason for the Processor FCA to send data to the central ARINC 629 FCA.
I think data validity check is performed within the Processor FCA, but not ARINC 629 FCAs. If you notice, all the three ARINC FCAs send the data back to the processor FCA. I think this is done to ensure that each of them receives valid data, i.e. that no data corruption occurs in the ARINC FCAs or on the way from the processor FCA. I.e. the processor makes sure that ARINC FCAs receive exactly what it sent.
Re: “The AIMS always sends SAARU data to the standby attitude indicator (where fitted) and to the digital flight data recorder system. It also sends SAARU data to the following systems whenever ADIRU data is not available…”
There is no mentioning about the SDU, however, from my earlier experience with the ATSB reports and various manuals, any statement containing “including” or its equivalent does not mean the complete list. If I am not mistaken, ACARS may optionally use data from the SAARU for reporting, which is not mentioned, right?
@Victor
“I agree that the captain was likely not aware of the underlying signaling of the SATCOM communication, nor the associated BTO and BFO measurements that were recorded. However, the captain would be very aware that isolating the left main bus would end any SATCOM communication, as an EICAS message explicitly states this.”
So what? The captain did not have to respond to any communication be it SATCOM, VHF, or UHF. Why would he be concerned about shutting these systems down? Makes no sense at all to me. If I were diverting an aircraft I would want to be in the active listening mode not in a shut down mode.
@DennisW: I am simply showing you that the SATCOM drop-out is a big inconsistency in your scenario. You keep presenting the inconsistencies you see with other scenarios without explaining why in your scenario the SATCOM was not available from 18:03 to 18:25, and why when it did finally log-on and become available, it did so without ACARS capability.
If you recall, this conversation thread began with my claiming that all scenarios that I have seen have inconsistencies. You said that yours does not. An explanation of the SATCOM drop-out does not require that you negatively comment on the IG, the DSTG, and others that are wrestling to understand the disappearance.
If your stance is antenna misalignment due to system malfunction or other random events caused the 22+ minute drop-out, I would say that is extremely improbable if it just by chance coincided with an intentional diversion. But so far, that’s the only explanation that I’ve seen you propose.
@DennisW: “The captain did not have to respond to any communication be it SATCOM, VHF, or UHF. Why would he be concerned about shutting these systems down? Makes no sense at all to me. If I were diverting an aircraft I would want to be in the active listening mode not in a shut down mode.”
I think you agree that in a diversion scenario, the pilot turned off the transponder. The same logic that prompted him to turn off the transponder prompted him to turn off ACARS. Both would identify the aircraft and the position very accurately.
@Victor
@Victor
“I think you agree that in a diversion scenario, the pilot turned off the transponder. The same logic that prompted him to turn off the transponder prompted him to turn off ACARS. Both would identify the aircraft and the position very accurately.”
You don’t have to take down the left bus to turn off ACARS.
“An explanation of the SATCOM drop-out does not require that you negatively comment on the IG, the DSTG, and others that are wrestling to understand the disappearance.”
My negative comments on the IG and DSTG have nothing to do with the SATCOM dropout. It is has everything to do with ignoring any possible motive or causality and sticking pins in the map at 38S. There is absolutely no logic that would support the flight path proposed by the DSTG and initially by the IG. I was asked to leave Duncan’s site when I pointed this out early on. It has come home to roost. Live with it.
@DennisW: No, the SATCOM doesn’t have to be powered down to disable ACARS. However, it might be the fastest way to do it until the situation becomes more manageable.
I still haven’t heard an explanation for why in your scenario the SATCOM was not available for 22+ minutes. If you recall, the purpose of this thread was related to your claim that there are no inconsistencies in your scenario, not to discuss the sins committed against you by Duncan Steel.
It’s perfectly acceptable to say you don’t have an answer. It just means there is a hole in your scenario. As I said, all the scenarios I’ve seen have one or more holes.
@ALSM,
Re: “the 18:25 observations were not at all consistent with the aircraft state and the steady state BFO bias (~150 Hz).”
Firstly, I am not sure what you mean. I am aware only of the two BFO samples during this interval: 142 Hz (18:25:27) and 273 Hz (18:25:34). The BFO of 142 Hz is perfectly consistent with the radar-derived data. So, what do you mean?
Secondly, I think the critical point about your “warm up” theory is what Victor explained later with regard to the correctness of the first BFO and corruption of the second BFO. Unfortunately his explanation invalidates your statement with regard to 00:19 BFO of -2 Hz. The correct BFO would need to be somewhere in the range -30 to -200 Hz (based on Holland’s paper), which is nearly impossible. Alternatively Victor’s explanation is invalid, or your “warm up” theory is invalid/inaccurate.
@Victor
“It’s perfectly acceptable to say you don’t have an answer. It just means there is a hole in your scenario. As I said, all the scenarios I’ve seen have one or more holes.”
A hole is not an inconsistency. It could have happened for reasons I stated. You are objecting to that on the basis of probability. 7.0 earthquakes are not probable, but they happen all the time. Yes, I do not KNOW why it happened, just like you don’t KNOW the left bus was powered off.
@Dennis
>It is hard for me to believe that anyone can possibly support a mechanical scenario or a suicide scenario given the above
Yet you have in your present scenario a malevolent, suicidal, mass murdering pilot. You’ve gone from a benevolent Zaharie (in your original scenario) who somehow ran out fuel and was a gentle soul who would never bring harm to his pax intentionally, to a Zaharie who was fully capable of killing himself and slaughtering 238 human beings because he ‘meant business’ (in so many words).
Wow.
@Donald
I have no reference to malevolent or suicidal – those are your words. Why attribute them to me? Shah was a patriot. Mass murder was the result of a failed negotiation, not a preferred outcome.
Look at the facts, and stop being argumentative.
1> Why divert a flight to Beijing?
2> Why was the Malay response what it was?
3> Why are the Malays withholding information?
Why not provide your insights here instead of posting drivel?
Oleksandr:
There were 7 BFO values circa 18:25, not 2. You have to consider all 7 as part of a temperature (and frequency) transient pattern resulting from the cold start at ~18:23-18:24. BFO 1a happens to be close to the steady state value reached 3-4 minutes later because the OCXO frequency was increasing at that time, starting from a much lower value when the OCXO power came on. Unlike the warm start at 00:19, the SDU BITE tests at 18:24 were probably all complete (satisfied) except for the OCXO temperature test. That being the last test to pass the BITE criteria means the first transmission (18:25:27) occurred while the frequency was still settling in on the equilibrium set point temp. We don’t see this transient at 00:19 because the OCXO temperature test passed before some or all of the other BITE tests.
ARC UTC BFO
1-a 18:25:27 142
1-b 18:27:03 176
1-c 18:27:04 175
1-d 18:27:08 172
1-e 18:28:06 144
1-f 18:28:10 148
1 18:28:15 143
You can call it a coincidence that 1a is close to equilibrium, or you can recognize it as part of the expected transient response.
See Figure 2 and related text for a more complete understanding of the 18:25 OCXO transient.
https://goo.gl/CUGl0O
Mike
Question – suppose one wanted to incorporate fuel model predictions along with BTO/BFO data (and the sketchy drift modeling predictions, if one were so inclined) in deriving a revised estimate of the end-point latitude for MH370? Let us say that the fuel model has been beat upon to match the HOLD and LRC tables for the Trent 892B engine, a sketchy procedure has been implemented to make it work as well with the limit MRC information from the ATPL document and thus predicts fuel consumption between HOLD and LRC speeds, and had been outfitted with at PDA to match the predictions of the Fuel Analysis from RMP Folder 5. What error model would one use for the fuel prediction model? In particular, what is the rms error (say, in tons) of a fuel model prediction? We are not Boeing, so don’t have access to all that proprietary information, but we are not working in a vacuum either.
@Oleksandr
“I think data validity check is performed within the Processor FCA, but not ARINC 629 FCAs. If you notice, all the three ARINC FCAs send the data back to the processor FCA. I think this is done to ensure that each of them receives valid data, i.e. that no data corruption occurs in the ARINC FCAs or on the way from the processor FCA. I.e. the processor makes sure that ARINC FCAs receive exactly what it sent.”
The ADIRU block diagram that you posted previously is featured in a paper by Mike Sheffels, a Honeywell systems engineer who worked on the design and development of the ADIRU (Sheffels, 1993). Sheffels states that “The ARINC 629 FCM votes all of the data coming from the four processors in order to produce consistent data to be output on the ARINC 629 bus.” He further states the following:
“The processors perform redundancy management on the gyros and accelerometers by using Fault Detection and Isolation (FDI) software. The data received from the sensors is checked by the sensor FDI routines before it is used to calculate outputs. This assures that only data from healthy sensors will be used to compute the ADIRU outputs.
Processor redundancy management is accomplished by having the ARINC 629 FCMs vote the data that they receive from the four processors. Any processor FCM whose data does not match the data from the other processor FCMs is faulted.”
Reference:
Sheffels, M.L. (1993). A Fault-Tolerant Air Data/Inertial Reference Unit. IEEE Aerospace and Electronic Systems Magazine, 8(3), 48-52.
“There is no mentioning about the SDU, however, from my earlier experience with the ATSB reports and various manuals, any statement containing “including” or its equivalent does not mean the complete list. If I am not mistaken, ACARS may optionally use data from the SAARU for reporting, which is not mentioned, right?”
I’m not sure about ACARS, but in general SAARU data is substituted for ADIRU data if the ADIRU data is not available or invalid. However, the SAARU only provides attitude, heading and air data; it does not provide position information. The AES needs aircraft position information for antenna steering, so there is no point sending SAARU data to the SDU.
@Victor:
sorry for asking again, but why did you disable the avatar icons ? 🙁
thank you.
@Victor
@TBill
Gents, a question about MS FSX/FS9 please. Given the “Clear Skies” setting referenced in Victor’s and Yves’paper what, if any, winds are modelled?
@ALSM,
“There were 7 BFO values circa 18:25, not 2.”
I don’t know how you count, but there were 2 BFOs, not 7. In total there were 12 BFO samples during 18:25-18:28 interval.
Re: “BFO 1a happens to be close to the steady state value reached 3-4 minutes later because the OCXO frequency was increasing at that time, starting from a much lower value when the OCXO power came on.”
Take a look at Fig 8 and 9 of Holland’s paper. Logon #5 suggests that the equilibrium reached in somewhat around 40 seconds. For the Logon#7 the ‘settling’ time is less evident, but I would guess it is around 100 seconds (to reach 5 Hz error band). There is no other data to judge on the settling interval, so your and Holland’s statement that the steady state value is reached in 3-4 minutes is not supported by any data, and furthermore, it contradicts to the data presented in Holland’s paper.
@Andrew,
I think I took that diagram from Krings, 2007 presentation. It really cites Sheffels M.L., 1993, so I think you are right with regard to the origin.
If you notice, there 4 incoming arrows to each of the ARINC FCA; I guess each of them indicates individual processor. In other words 4-channels. I think this explains your citation “The ARINC 629 FCM votes all of the data coming from the four processors in order to produce consistent data to be output on the ARINC 629 bus.”. However, how to control what final data the ARINC FCA produces? In my opinion, the most logical thing is to send final result back to the processor FCA. In case if the ARINC FCA fails, the processors would know about it before the data are transmitted outside of the ADIRU.
But this explanation does not solve the mystery of the central ARINC FCA: why does it need data from the processors? The processors can compare their results internally – the use of the ARINC FCA for this purpose would be weird.
Re: “However, the SAARU only provides attitude, heading and air data; it does not provide position information. The AES needs aircraft position information for antenna steering, so there is no point sending SAARU data to the SDU.”
Do GPS send data to the AIMS(DCGF)? Also, I think antenna steering is much less sensitive to a geographical location than attitude, especially near the equator.
@DrB you wrote May 20 at 1:29 pm: “In addition, the footnotes are included, notably the one regarding 3% FF increase per 10C increase in TAT. This confirms my original conclusions that this factor must apply to the Trent engine and that its value would be the same as used for the GE engine.”
I commented earlier that fuel flow correction for temperature could be similar for GE and Trent engines, but not necessarily equal. Also the 3% is probably rounded up. For example, based on the data in Jet Transport Performance Methods it would be 2.8% per 10°C TAT for the PW2040 engine in the B757.
For the Trent engine in the B777 I looked at the fuel flow values at LRC speed. I varied the theta exponent and found that the best fit was obtained for theta-exponent 0.625 as illustrated here.
Using theta-exponent=0.625 results in 2.5% FF increase per 10°C increase in TAT, and would be more accurate than 3% for the Trent engine.
@Mick Gilbert: For “Clear Skies”, there is no wind. However, from looking at the velocity information from data set 5N, we find that Zaharie Shah had chosen “Fair Weather”, which defaults to wind of 25 knots coming from 270°M.
@Perfect Storm: I turned off the option to display avatars to help reduce the latency, as few commenters here have avatars. Eliminating the avatars helped some. I’m looking into other options to increase the speed.
@Victor
Thank you for that clarification regarding FSX/FS9.
@Victor
Victor, I have been continuing to look at your and Yves’ analysis of the Captain’s sim data and I note that your postulate that there is “… little doubt that a user created a simulation in which a B777 is successively positioned in the Malacca Strait, the Andaman Sea, and the SIO.” is largely predicated on the assumption that an “FMT” of sorts occurs just subsequent to the airplane reaching 10N (data point 3). Although the fuel consumption on the final leg suggests that 10N and 45S1 (data point 4) are not linked directly (ie by flying directly from one to another by the shortest route) your posited solution for the discrepancy is the Captain “offloading” fuel so as to terminate the flight around 45S; a perhaps not unreasonable Deus Ex Machina (or at the very least, Malus Homo Ex Machina) intervention.
Might I offer an alternate scenario that I believe provides a superior fit to the data.
We start with 10N. I say we start here because 10N cannot be unequivocally linked to 2N, 3N or 5N as neither the fuel load nor, as one might expect with a load difference, maximum engine rpm parameters are consistent. While we can assume that 10N is part of a simulated flight from Kuala Lumpur that must have started with a fuel load in excess of that recorded for 2N what we know is that the airplane is at 10°10’59″N 90°13’28″E at FL400 with over 10 hours of fuel on board (62,700 kg). The data shows the airplane to be in a banked turn to the left passing through 255.5°.
The Captain may, as you have suggested, created his then upcoming flight to Jeddah (MH150) or he may have recreated his recent 20 January 2014 flight to Amsterdam (MH16); the latter would possibly be a better fit to the fuel load. The normal route for both MH150 and MH16 from the Andaman Sea and across the Bay of Bengal is N571. The airplane’s heading at 10N is taking it back towards N571 at a convergence angle of only 40° so, rather than assuming that this is the start of a departure from flight plan towards the Southern Indian Ocean, let’s instead assume that the airplane rejoins N571 and continues its flight west. It dutifully tracks all the way to the east coast of India, crossing it some 85 nm north of Chennai, and then tracks all the way across the Indian sub-continent until it reaches the west coast at waypoint BEKUT, some 85 nm south of Mumbai.
Having reached BEKUT if an LNAV track to waypoint SPOLE (the South Magnetic Pole) is executed the great circle track will take the airplane almost directly through 45S1; the precision here is really quite extraordinary, less than 2 nm or about 4 times better than the McMurdo track. While this alignment in of itself might be coincidental my estimates suggest that at or near 45S1 is where fuel exhaustion will actually take place for this simulated flight; there should be no requirement to invoke a manual adjustment.
The distance from 10N via N571 to BEKUT and then to 45S1 via a great circle route to SPOLE is about 5525 nm; at cruise speed that would take about 10.5 hours. So, with 67,200 kg of fuel on board at 10N and assuming fuel exhaustion at 45S1, that yields fuel consumption rates of;
– about 5970 kg/hr, which closely approximates the rule-of-thumb planning number for the B777 of 6,000 kg/hr and where you’d expect it to be if compared to MH370’s 6,100 kg/hr from its last ACARS transmission to fuel exhaustion with pax and cargo of 31,086kg; and
– about 12.2 kg/nm, which approximates the 13.8 kg/nm that you and Yves achieved for your much shorter final leg into the SIO on your recreated flight and approximately 13.5-ishkg/nm MH370’s from its last ACARS transmission to fuel exhaustion.
I don’t have FSX so I haven’t been able to model this in any detail. The flight path I am suggesting looks like this:
101059N0901328E 095356N0891351E BIKEN N571 BEKUT 450507S1040844E 641648S1363524E
(450507S1040844E is 45S1; 641648S1363524E is the South Magnetic Pole as at 2015; it’ll be within 15 kilometres of where it was in 2014. As I understand it NPOLE (North Magnetic Pole) is a waypoint in most FSX FMC AIRAC databases and SPOLE is in some but not all AIRACs – it would be interesting to find out if SPOLE was included in the Captain’s sim’s database.)
Now, I could be completely off piste with this and would very much welcome someone running this scenario and providing feedback.
@Mick
My B777 is grounded for the Memorial Day holiday weekend
@Gysbreght
So the question then is would a 2.5% FF increase per 10deg increase in TAT allow the plane to reach S38 at fuel exhaustion?
@ROB: I don’t know. You still have to add 1.5% PDA. Finally searching the seafloor at S38 didn’t turn up anything.
@Victor: thank you for the explanation
@Ge Rijn. The photo of the nose doors opened, presumably for access, is a good one thank you and the video is on the money also.
Returning to my previous post, to be in tension the remnant needs to be stretched by being pulled at its ends, which requires a force separating those two hinges. Even a flat impact leading to fuselage separation above most likely would break the door by bowing it inwards. In any bending like that from a force outside (or indeed within) will place one skin in tension but the other in compression. Yet the Malaysian finding ostensibly was tension in both.
As to longitudinal inertia forces on the aircraft hitting the sea, anything sizable will be compression from the structure in front. Leveraged inertia forces applying tension to rear wing and stabiliser root attachments would not act like that here.
What might apply an instantaneous tension would be a shock wave from structural failure elsewhere, transmitted through hinges and surrounding structure, though that would be associated with a high speed impact I would have thought.
@Mick Gilbert: An aircraft following a great circle path to the South Pole is also following a rhumb line at a track angle of 180°T. This is a line of constant longitude. Essentially, you have found a navigational waypoint (BEKUT) that shares the same longitude as 45S1. Or equivalently, you have found two waypoints (BEKUT and South Pole) that align with a single simulation coordinate (45S1). I don’t find this nearly as persuasive as two coordinates (10N and 45S1) from the simulation that align with McMurdo Station.
(According to this Boeing reference, in a B777, the way to enter a waypoint of the South Pole is 99SP, S90EXXXXX, or S90WXXXXX. The waypoint corresponds to the true pole, not the magnetic pole.)
@David said, “What might apply an instantaneous tension would be a shock wave from structural failure elsewhere, transmitted through hinges and surrounding structure, though that would be associated with a high speed impact I would have thought.”
Yes. This is the type of analysis that was performed by Wierzbicki and Yue when they studied the damage to the crew compartment of the Space Shuttle if the nose impacted water at high speed. They found that the pressure pulse transmitted through the structure would lead to tearing and fracture of the fuselage.
Here’s a short report I did running a grid of models in altitude and speed and seeing how fuel consumption varies as a function of final latitude. Bottom line – fuel consumption constraints can set a maximum Southern latitude, but not much else without more information.
https://docs.google.com/document/d/14hleZyx1pUPL44yaeHKt6jnSQ3DbgRq2zibbKkFLq2c/edit?pref=2&pli=1
@Victor
Victor, thank you for the Boeing references on polar flight.
Re: “Essentially, you have found a navigational waypoint (BEKUT) that shares the same longitude as 45S1. Or equivalently, you have found two waypoints (BEKUT and South Pole) that align with a single simulation coordinate (45S1).”
Waypoint BEKUT is on N571 at 17°45’21″N 73°12’55.8″E; it is most assuredly not on the same longitude as 45S1 (45°5’7″S 104°8’44″E). At first blush there is no apparent relationship whatsoever between the two.
What I have suggested as an alternative to the direct track to McMurdo is;
a. entirely consistent with the Captain either creating a simulation of MH150, his then upcoming flight to Jeddah, or recreating MH16, his then recently completed flight to Amsterdam; and
b. most likely consistent with the fuel data for 10N and 45S1 (ie it does not require a manual intervention to create fuel exhaustion at an arbitrary point).
If we attempt to view the simulations of 24 January 2014 contemporaneously rather than retrospectively we would at least consider that the Captain was seeking to test the operation of the PSS Boeing 777-200LR No VC program given the documented software crashes he experienced when he previously used that software in December 2013. With that in mind, given the amount of discussion in AVSIM forums and similar about the vagaries of FS9/FSX when modelling polar flights, surely it is not beyond the realm of reasonable possibilities that, having satisfied himself with the general stability of the B777 program, that is what he then looked at, a diversion to the south magnetic pole?
McMurdo requires the invocation of a manual intervention to create fuel exhaustion; my alternative requires the invocation of the continuation of the flight that we assume the Captain was simulating. Tomarto, tomayto.
I’m surprised that you don’t find it even mildly curious that a deviation from a waypoint on an airway used by a flight that the Captain was most likely (re)creating just happens to precisely align with another data point ostensibly from that same flight soms 4120 nautical miles away? It’s not like I’ve just conjured any of this out of thin air.
@sk999
So here is a dumb question for you, but I have to ask. I have the same issue with “Bayesian Methods…” book.
What do you mean by RMS BFO error? Normally RMS is defined by subtracting each value of a data set from the data set mean, squaring, adding, and taking the square root. Are your values computed this way or is the RMS value computed using an assumed mean error of zero. Obviously the mean value of the BFO errors in your data set is not zero unless you have a program that selects the flight paths to produce a mean BFO error of zero.
If you did structure your flight paths to have a zero mean BFO error, I would be suspicious of these paths since I assert all BFO logs will resemble figure 5.4 of “Bayesian Methods…” just not as extreme. It is the way oscillator error evolves over time.
Thanks in advance.
DennisW,
If I allow the mean (a.k.a. “bias”) to float to a new value after the 18:25 logon (as Bayesian Methods says you should), I get almost no change from the pre-IGARI bias value for a wide range of flight paths, so I long ago kept it fixed at the value determined at the gate. So no, I do not subtract off the sample mean first. In fact, for this exercise, only the BTOs are used to constrain the flight paths – the BFOs are what they are. It does seems odd that the BFO should be so well-behaved, but that’s what I find.
@sk999
Thanks. That was my assumption of how you computed the RMS BFO values. Likewise how your paths were constructed.
@Mick Gilbert
Mick, did you mean magnetic pole or geographic pole in your scenario? Inputting a waypoint for the South Pole (99SP, S90EXXX, etc) will take the aircraft to the geographic pole.
@Andrew
Andrew, I was refering to the south magnetic pole not the south geographic pole. My understanding is that the south magnetic pole is included in the AIRAC database as waypoint SPOLE but that it is not always included in FMC navigational databases.
@Mick Gilbert:
a) You refer to waypoints NPOLE and SPOLE, which refers to the location of the true poles, as the reference I provided shows. That’s why I referred to a flight towards the South Pole as having constant longitude. If you mean a flight towards a magnetic pole, please don’t use the symbols NPOLE and SPOLE to avoid confusion.
b) You said, “The Captain may, as you have suggested, created his then upcoming flight to Jeddah (MH150) or he may have recreated his recent 20 January 2014 flight to Amsterdam (MH16); the latter would possibly be a better fit to the fuel load.”
In the recreated flight, we measured a fuel rate of 13.8 kg/NM for the segment of 10N until fuel exhaustion. The great circle distance from KLIA to Jeddah is 3,818 NM and the great circle distance from KLIA to Amsterdam is 5,529 NM. The fuel load for the simulation was 68,424 kg, which is a range of about 4,958 NM, or about 571 NM short of the direct distance between KLIA to Amsterdam, and even more if a realistic path following airways is chosen. Based on these calculations, the fuel load looks appropriate to Jeddah and insufficient for Amsterdam.
c) By my calculations, the distance from 10N to BIKEN to N571 to BEKUT to 45S1 is 5217 NM. (You say it is 5525 NM.) At 10N, the fuel available in the simulation is 62,700 kg. In the recreated flight for the long segment between 10N and fuel exhaustion, the fuel rate was 13.8 kg/NM. Applying this rate to the simulated flight yields an additional range of 4,543 NM, or 673 NM short of the distance of your proposed route. To achieve this, ZS would have needed to add fuel, i.e., manually intervened.
d) You said, “If we attempt to view the simulations of 24 January 2014 contemporaneously rather than retrospectively we would at least consider that the Captain was seeking to test the operation of the PSS Boeing 777-200LR No VC program given the documented software crashes he experienced when he previously used that software in December 2013.”
The software crashes were for his use of FSX on the MK26 drive, not FS9 on the MK25 drive. We are discussing recovered data from running FS9 on the MK25 drive.
e) You said, “I’m surprised that you don’t find it even mildly curious that a deviation from a waypoint on an airway used by a flight that the Captain was most likely (re)creating just happens to precisely align with another data point ostensibly from that same flight soms 4120 nautical miles away? It’s not like I’ve just conjured any of this out of thin air.”
But you have conjured the flight from 10N to BEKUT out of thin air, and the fuel load doesn’t allow it. No to mention you added a detour away from N571 to 10N and then back to N571. On the other hand, the hypothetical flight from 10N to 45S1 to McMurdo requires no waypoints to be found. Just extend the line between 10N and 45S1 and you hit McMurdo. It’s not a matter of my not being curious about your proposal. I just don’t see how it makes sense that this was the simulated route.
@Oleksandr
“However, how to control what final data the ARINC FCA produces? In my opinion, the most logical thing is to send final result back to the processor FCA. In case if the ARINC FCA fails, the processors would know about it before the data are transmitted outside of the ADIRU.”
According to Sheffels, the processors determine which ARINC 629 FCM is active on a given ARINC 629 bus. However, data voting is done by the ARINC 629 FCMs, not the processor FCMs. The following might be helpful:
Fault-Tolerant Voter System for Output Data from a Plurality of Non-Synchronized Redundant Processors
“But this explanation does not solve the mystery of the central ARINC FCA: why does it need data from the processors? The processors can compare their results internally – the use of the ARINC FCA for this purpose would be weird.”
I have already suggested a reason: fault tolerance. According to Sheffels, the processors do not compare their results internally.
“Do GPS send data to the AIMS(DCGF)?”
GPS data is sent to the ADIRU for ADIRU sensor calibration. I assume the GPS data is routed via the AIMS DCGF and then output on the ARINC 629 Flight Controls buses.
@Victor
Victor, a very quick initial response with a more detailed one to follow.
Let’s start with my stuff-up; SPOLE is, as you have stated, the south geographic pole. My bad. A waypoint for the south magnetic pole would have to be created; an action that would surely not be beyond the capabilities of our intrepid Captain.
While on the topic of creating waypoints, McMurdo is neither a waypoint nor an airport; Phoenix (NZFX), Pegasus (NZPG) and Williams (NZWD) are airfields but aren’t in most navigational databases; NOBEY is the nearest waypoint but using it nearly doubles the offset between the track flown for 10N-NOBEY and 45S1. How do you believe the Captain went about designating his destination?
I’ll look to address your other points in due course (due course falls after the dessert course). In the interim, what makes you so sure we’re looking at FS9 files and not FSX flight file data that was saved to the MK25 drive?
@Victor. The space shuttle paper. Thanks. The excerpt of the extract does not make clear with its one yield stress (under “Nomenclature”) whether they look at compressive, tension or shear though you may know. I have not accessed the paper or those cross referenced that are more recent but will if the context warrants. I see the Journal of Impact Engineering (that’s a new one) has a 2006 paper on water entry of aerospace structures.
They do estimate ‘acceleration’ which could be useful in assessments of the decelleration for any or all MH370 stabilisers to break off forwards, debris resulting from impact with other structure and water impact after twisting.
I was hoping might be a reconstruction in the final report though this would depend on how germane that is: other evidence might render it unnecessary or redundant.
For those interested in what on this topic a final report should contain generally, go to Appendix 1 of Chapter 1, ‘Format and Content of the Final Report’, at the pdf page 11/40 of:
http://www.aaib.gov.mn/uploads/9756_p4_cons_en.pdf
If you scroll down to p15, 1.12.1 & 2 under Factual Information you will find mention of wreckage and impact information. At 1.16 tests and research should include examination of aircraft parts.
At page 18, 2.2, Analysis, you will find that should contain an evaluation of FI evidence (and also incidentally an account of the rejection of unsupported hypotheses).
At page 22, Appendices should include technical investigation reports.
@Mick Gilbert:
“While on the topic of creating waypoints, McMurdo is neither a waypoint nor an airport; Phoenix (NZFX), Pegasus (NZPG) and Williams (NZWD) are airfields but aren’t in most navigational databases; NOBEY is the nearest waypoint but using it nearly doubles the offset between the track flown for 10N-NOBEY and 45S1. How do you believe the Captain went about designating his destination?”
The McMurdo Station waypoints would have been available to ZS in a number of ways. In the FS9 base package, the following airports are available when creating a flight plan with a destination in Antarctica: SAWB, NZIR, NZPG, NZWD, NZ12, SA47, and NZSP. (NZSP is the South Pole, not the magnetic pole.) Zaharie Shah, as a serious flight simulator enthusiast, might have subscribed to a third-party package that allows the creation of detailed flight plans and complete AIRINC databases. There are lots of possibilities.
I found the best fit with NZPG. I am only using the generic location McMurdo Station because more people would be familiar with it than Pegasus Field. McMurdo is the largest base in Antarctica. The flight to Pegasus Field is described here.
“In the interim, what makes you so sure we’re looking at FS9 files and not FSX flight file data that was saved to the MK25 drive?”
I am be quite sure we are looking at FS9 files because the PSS 777 model is only available for FS9. There are also differences between FS9 and FSX flight files that allow me to be quite sure the flight files were from FS9.
@Victor
It would appear then that the RMP use the term “Flight Simulator X” when refering to FS9. Somewhat confusingly there is a report dealing with the application crash analysis and log book files that makes no reference to FS9 at all. The flight logs for what the RMP refer to as Flight Simulator X on the Mk26 drive shows that there were five B777 flights flown out of KL on 23 January 2014; the eight data sets recovered from the Mk25 drive show clear evidence of at least 5 B777 flights and three of the data sets show a B777 at KL. Surely that’s not coincidental.
Absent the link between the logs on the Mk26 and the data on the Mk25 there’s no way to know when any of the MK25 data was recorded.
Thank you for the information regarding McMurdo and the extent of the databases. As it turns out I don’t think you need invoke McMurdo; have you plotted a CMT of 180° from 10N? I think it passes pretty close to 45S1.
@sk999: You write in “FUEL MODEL SENSITIVITY TEST” sk999 May 28, 2017:
“However, from the ATPL document, we know that the MRC mach predicted by this curve is wrong, in the sense that as the difference between the HOLD and LRC Machs becomes large, the MRC Mach is closer to the middle of the Mach range rather than being skewed towards the high end. I derived a nonlinear relation that rescales the Mach number between the HOLD and LRC limits such that the predicted MRC matches the numbers in the ATPL tables for CI=0. The rescaling a parametrized as a function of the difference LRC minus HOLD Mach. I assume that this relation (which is derived for a GE90-115B1 engine) applies to Trent engines as well. This model allows me to interpolated in speed (Mach) between HOLD and LRC, and I assume it holds for Mach values above LRC as well (which happens near the end of the flight when the plane has low mass).
“
The LRC/MRC ratio decreases at high Mach numbers. There is no obvious relation to the Holding speed. Holding speed at altitudes above FL100 seems to be governed by the need to maintain an adequate manoeuvre margin relative to the buffet onset Mach number. The Holding IAS has two levels, each defined by a constant lift coefficient, and a transition between the two levels. The lower level probably represents the maximum endurance speed.
Apparently Boeing has selected a somewhat different holding speed schedule for the B777-300ER described in the ATPL booklet. Compared to the -200ER the -300ER has a different engine and a longer fuselage, but the same wing.
@Mick Gilbert: The RMP report has numerous errors. Anybody that studies it in detail will find it to be a shoddy and incomplete compilation of evidence from various sources. However, the translated “Preliminary Examination Report” clearly refers to FS9 installed on MK25 on Windows 7, and the crash analysis refers to FSX installed on MK26.
When the IG first released a preliminary analysis of the simulator data, we (really I, to put blame where it should rest) did not yet appreciate the differences between the FSX and FS9 installations on the two drives as the distinction is buried in the “Preliminary Examination Report”, which is in Malay. The distinction became apparent to me after I installed FS9 and the PSS 777 model and replicated ZS’s simulation.
The deleted recovered flight files on MK25 suggests one flight, as Yves and I have shown. I don’t understand why you claim there is evidence of at least 5 B777 flights.
I don’t recall looking at a 180° CMT path from 10N. If a CMT or other navigational mode from 10N passes close to 45S1, that is interesting and should be explored.
@Victor
Victor, I’m sorry but I can’t understand how you can say that there was only one flight. Your own analysis demonstrates that the flight that includes 10N, 45S1 and 45S1 is not part of the same flight that included 3N and 5N. Why else have you bracketed these sets of coordinates separately in red and blue in Table 2?
Regarding whether the 10N flight might have been MH150 or MH16, that flight must have departed KL with significantly more fuel than the 68,424 kg you attribute to it; it could not possibly have reached FL400 and flown some 820 nm having burned only 5,724 kg of fuel. Back of the fag packet, if the N10 flight originated from KL then it must have departed with at least 80,000 kg fuel; that’s significantly more than is required for MH150 but adequate for MH16.
In any event, let’s get down to brass tacks on the recovered data. In addition to the six data points that you and Yves refer to your paper there are two more:
1. Coordinate 6-2. Similar to 2N (ie lined up on RWY32 at KL) except with full fuel load – 135,845 kg – let’s call this 2N-2.
2. Coordinate 7. A very curious record, 0 kts at 17 ft above the keys on RWY32 at KL, gear down, with what appears to be 50% flap, at least 45,528 kg of fuel (bizarrely on top of 80.54% fuel in each of the left and right tanks there is meant to be 80.34% in the LeftAux and RightAux tanks?!) and MaxReachedEngineRPM of 17667.846 rpm. Let’s call this 2N-?.
You’ll note that the fuel loads for 2N and 2N-2 do not line up with either the 10N-45S flight nor the 3N-5N flight. So, unless you want to invoke the God of Inflight Refuelling/Dumping, there’s data supporting at least four separate flights; add 2N-? makes five. And that is consistent with the flight log for 23 January 2014 as detailed in Section 6.4 of Folder 1 of the RMP report.
As well as the additional data sets, there are some “discrepancies” with two of the “original” data sets;
1. According to your interpretation, at 10N we have an airplane with a gross weight of 247,100 kg at FL400 climbing at 3,500 fpm – that is simply not possible in the real world, is it possible using the flight simulator program?
2. 45S2 has a disagreement between Altitude (4000) and AGL (37654) that is not seen in any other data set. I don’t know if you have a view as to how this disagreement may have come about or which, if any, value should be deemed as (in)valid.
Any old how, getting back to what data sets/coordinates belong to which flight, based on the Mk25 evidence it is not unreasonable to conclude that;
– 3N and 5N are from the same flight by virtue of the shared MaximumGForce/MinimumGForce and MaxReachedEngineRPM; and
– 10N, 45S1 and 45S2 are from same flight by virtue of the shared MaximumGForce/MinimumGForce and MaxReachedEngineRPM.
And that’s it in terms of reasonable conclusions based solely on the Mk25 recovered data; anything further is just supposition. Not only is there nothing linking the 3N-5N flight to the 10N-45S flight there is in fact evidence to suggest that they are separate flights that started with different take-off weights. MaxReachedEngineRPM should be recorded at take-off and should relate to take-off weight; the 10N-45S flight has a higher MaxReachedEngineRPM than the 3N-5N flight as would be expected if it had a higher take-off weight, which it should have. In sum, they’re different flights.
Further, and I know this is splitting hairs but, based solely on the Mk25 recovered data, because there is nothing conclusively linking any of the inflight data sets – 3N, 5N, 10N, 45S1-2 – with any of the on-ground data sets – 2N and 2N-2 – you cannot reasonably conclude that either of the 3N-5N or 10N-45S flights even originated from KL. I think that they did but that is supposition beyond the evidence.
Accordingly, representing KL – 3N – 5N – 10N – 45S1-2 as one contiguous flight per figures 1, 2 and 3 of your and Yves’ paper, concluding that there is “… little doubt that a user created a simulation in which a B777 is successively positioned in the Malacca Strait, the Andaman Sea, and the SIO.” and stating that “The deleted [they weren’t deleted, they simply weren’t saved and were overwritten] recovered flight files on MK25 suggests one flight,” either goes beyond the bounds of or is directly contradicted by the Mk25 data analysis.
@Mick
Augmenting what Victor said, the older FS9 has the advantage of better computer system stability and runs faster due to simpler graphics etc. Its the FSX that typically had some more computer crash issues. Some people stuck with FS9. Although both versions have the issue the users typically apply many software patches.
FS9 has NZPG which means you can enter that FS9 flight plan into PSS777.
PSS777 itself does not have NZPG (or UPROB) but it is super easy to modify the PSS777 waypoint text file.
Also using the numerical waypoint methods Victor talks about above (eg; S90EXXXXX) any finely detailed lat/long position can be added as a waypoint, thus giving us essentially infinite SIO waypoints.
It has never been clarified how these fragments of partly overwritten files got copied in the shadow partition of the Mk25 drive. Inconvenient questions are not welcome.
@Mick Gilbert: You are incorrect in so many of your points, many that we have addressed in the paper if you read it carefully. Your points about fuel loads are wrong. Your point about vertical speed at 10N is wrong for this simulation if you consider the anomaly that occurs in the particular way the flight files were created, which was discussed in detail. The altitudes at 45S1 and 45S2 have been addressed in the paper AND recreated. You say that 2N is not linked to 3N, which is wrong. (Hint, look at the remaining fuel at 2N and 3N in Fig 3.) You statement about maximum RPM at takeoff is wrong.
I can’t continue to correct all your mistakes. Carefully re-reading the paper is a place to start.
As I said.
Has it been tested to verify the MS VSS operatablility with MS FX9 as reported by the police report with file deletions and recovery? It would be interesting to make sure these file segments can be recovered as per report.
@Mick Gilbert: “Accordingly, representing KL – 3N – 5N – 10N – 45S1-2 as one contiguous flight per figures 1, 2 and 3 of your and Yves’ paper, concluding that there is “… little doubt that a user created a simulation in which a B777 is successively positioned in the Malacca Strait, the Andaman Sea, and the SIO.” and stating that “The deleted [they weren’t deleted, they simply weren’t saved and were overwritten] recovered flight files on MK25 suggests one flight,” either goes beyond the bounds of or is directly contradicted by the Mk25 data analysis.”
A longish sentence, but well said.
@sk999
“Fuel Model tests (May 28, 2019)” note slight typo re: 2017
When you say no auto-pilot mode matches the path data, I am in agreement. I do not have a problem with that conclusion, because I am expecting likely programming of path at least through 1941 and not sure about end-of-flight yet.
@Gysbreght
“Compared to the -200ER the -300ER has a different engine and a longer fuselage, but the same wing.”
Same wing? Not true! The -200LR and -300ER have the same wing, with raked wingtips and a longer wingspan than the original wing. The -200ER has the original wing, which is the same as that found on the -200 and -300 variants.
@Andrew: Thanks for that correction.
@Andrew: The wingtip extension explains the small reduction of the low-altitude holding speeds of the -300ER relative to the -200ER. Mach-related drag rise results in lower IAS for maximum endurance, as it does for MRC and LRC. The high-altitude holding speed is not a natural characteristic, but is ‘scheduled’ to satisfy a constraint other than minimum fuel burn rate.
@Victor
“I don’t recall looking at a 180° CMT path from 10N. If a CMT or other navigational mode from 10N passes close to 45S1, that is interesting and should be explored.”
First pass attempt, pretty darn close to 45S1.
I went IGOGU DOTEN 1090E and set to 180S Mag at DISCONTINUITY.
I had set Fair Skies so that’s a helluva wind from 270 deg the whole way.
I think I am using an update to 2005 Mag headings but I have to check…orig in FS9 would be 199x vintage mag tables which would go even further East (yet more curve).
@David
On your comment:
” to be in tension the remnant needs to be stretched by being pulled at its ends, which requires a force separating those two hinges. Even a flat impact leading to fuselage separation above most likely would break the door by bowing it inwards. In any bending like that from a force outside (or indeed within) will place one skin in tension but the other in compression. Yet the Malaysian finding ostensibly was tension in both.”
I agree if one side of a panel is stressed under tension by bending the other side will be under compression. But under compression is not the same as compression failure.
Bending and stretching a layered piece of composite fiber at the same time will cause the layers to break under tension on one side one after another, pulling fibers out until the opposit side shears off also mostly under tension.
This pattern is what I see in the nose gear door piece and other pieces. This breaking of layers from the inside outwards.
First bending and breaking under tension pulling fibers out than finaly shearing off without substantial compression damage due to longitudal air or water forces.
But after all we just need a very detailed forensic debris damage report.. Would be very welcome if this report would be conform to the standards mentioned in the link you provided.
Screen shots of flight path for the FS9 case above
https://docs.google.com/document/d/1HfWt2jttrZp2u9oL8oDc3d3Luz_osOWlwjzVIJBGW5w/edit?usp=sharing
Gysbreght,
Not that you aren’t correct, but until someone comes up with a better way to interpolate fuel burn v. mach, I’ll stick with my “hack”. In fact, if I eliminate the nonlinear rescaling and rerun the computations, the figure of final mass v. end latitude doesn’t change a whole lot, even though individual points do move around. In any case, the general conclusion that fuel modeling at best gives a maximum Southern latitude remains.
@Gysbreght
Thanks. The increased wingspan of the 200LR/300ER helps reduce induced drag. As I recall, the raked wingtip causes the outer wing to twist slightly in a nose down direction, offloading that part of the wing and reducing the wing’s overall bending moment. The reduced bending moment allowed them to reduce the structural weight of the wing. I’ve got a reference somewhere – I’ll try to dig it out.
@Oleksandr,
I don’t know how you count, but I get 13 BFOs at 18:25-18:28.
Bobby: I only count 7. What am I missing?
1-a 18:25:27 66327 0.012520 142
1-b 18:27:03 66423 0.012560 176
1-c 18:27:04 66424 0.012520 175
1-d 18:27:08 66428 0.012520 172
1-e 18:28:06 66486 0.012500 144
1-f 18:28:10 66490 0.012520 148
1 18:28:15 66495 0.012480 143
Bobby: I found the discrepancy. As you say, there are 13 total BFO values, but the six T channel values are redundant…not independent values, so I only counted the first one. In addition, I left out the 18:25:34 value in my count since it was obviously bogus, due to the warm-up transient. I guess it would be accurate to say there are 8 independent values, including 18:25:27.
Time Stamp Channel Type Burst Frequency Offset (Hz) BFO Burst Timing Offset (microseconds) BTO
3/7/2014 18:25:27.421 R-Channel RX 142 17120
3/7/2014 18:25:34.461 R-Channel RX 273 51700
3/7/2014 18:27:03.905 R-Channel RX 176 12560
3/7/2014 18:27:04.405 R-Channel RX 175 12520
3/7/2014 18:27:08.404 R-Channel RX 172 12520
3/7/2014 18:28:05.904 R-Channel RX 144 12500
3/7/2014 18:28:10.260 T-Channel RX 148 7540
3/7/2014 18:28:10.398 T-Channel RX 148 7540
3/7/2014 18:28:10.559 T-Channel RX 147 7540
3/7/2014 18:28:10.718 T-Channel RX 147 7540
3/7/2014 18:28:10.879 T-Channel RX 147 7540
3/7/2014 18:28:11.054 T-Channel RX 147 7540
3/7/2014 18:28:14.904 R-Channel RX 143 12480
@Gysbreght
This is all I could find:
Blunt-Leading-Edge Raked Wingtips
RE: “The high-altitude holding speed is not a natural characteristic, but is ‘scheduled’ to satisfy a constraint other than minimum fuel burn rate.”
The FCOM states that the holding speed is “the higher of the maximum endurance speed and the manoeuvring speed for the selected flap setting.” At low altitudes, that means the holding speed is based on the manoeuvring speed (ie VREF30+80) rather than the maximum endurance speed. The FCOM fails to mention that the holding speed is also limited to a minimum of about M0.82 at altitudes above optimum in the -200, -200ER and -300 variants because those aircraft are thrust limited at high altitude. The -300ER, on the other hand, is not thrust limited, so its holding speed at high altitude should be closer to the maximum endurance speed. Does that seem to be the case?
@ALSM/DrB
Again, I claim it to be miraculous that the 18:25:27 BFO value is exactly what it should be based on the radar data. The other values are easily explained by a maneuver (a casual maneuver not a rip the wings off maneuver as ALSM claimed earlier). But, no matter, I do not feel like arguing especially since it makes no difference in where the terminus might be.
Ask yourself why the pilot would disable the left bus. It makes absolutely no sense in the grand scheme of things. Which is why I do not believe it happened. Why bring the bus back it 18:25:27? You are postulating events which make no sense whatever. It reminds me of Duncan’s blog where Occam’s razor and ALSM’s “that’s the way pilots like to fly airplanes carried the day”. Funny shit in retrospect.
@Andrew:
Thank you for your comments.
When not limited to the manoeuvring speed, the holding speed is apparently scheduled for an Holding Lift Coefficient shown as a red line in the chart for the B777-200ER.
The FCOM table “LRC Maximum Operating Altitude” gives the Margin to Initial Buffet. The weights/altitudes for 1.50 ‘g’ (48° bank) correspond to CL=0.505 .
The All-Engine Holding table for the B777-300ER in the ATPL(A) Flight Planning booklet gives a holding speed at FL350 – FL430 that corresponds to CL=0.55. Assuming that the buffet onset envelope is approximately the same as for the -200ER, that would correspond to about 1.4 ‘g’ (45° bank). I suppose the empty spaces in
the table indicate thrust-limited conditions.
@Gysbreght
“I suppose the empty spaces in the table indicate thrust-limited conditions.”
True – I meant thrust limited on the slow speed side. If you get below about M0.82 in the -200, -200ER or -300 aircraft at altitudes above optimum, the aircraft does not have enough thrust to accelerate again. The only way to accelerate is to descend, as some people have discovered to their embarrassment. The same is not true in the -300ER, where it’s possible to slow down to the minimum manoeuvring speed and have plenty of excess thrust to accelerate again.
@Ge Rijn. The twin skin tensile failure of the nose wheel door that you envisage would depend on whether the compressed side remained elastic while the other fails in tension. Many design and production facets would come into play there: what the design intention was, proportion of fibre, matrix type, lay, skin thickness, buckle-stabilising (eg honeycomb) etc. The tensile failure of many debris items is more easily explained by them being restrained on 2 or 3 sides/ends. In others it could be that the skin in compression might resist that before failing in tension as you imply.
As to specifics with this item most likely bending stress would be greatest at half the door’s length, which would be about the length of the remnant. However it also was fractured across its width and even though the inside edge would have been unsupported I do not see how that occurred as a result much less with exclusively tensile outcome. Besides the outside rim remains intact and the front end is broken off, again apparently in tension.
One other thing to remember is that carbon fibre is brittle, so less resistant in deformation and shock loading than metal.
For my part I do not pretend to have great insight here but I have been trying to get clear what the scope and constraints would be for those who have cause and choose to look into all this.
@DennisW,
You said: “Again, I claim it to be miraculous that the 18:25:27 BFO value is exactly what it should be based on the radar data. The other values are easily explained by a maneuver (a casual maneuver not a rip the wings off maneuver as ALSM claimed earlier).”
This is incorrect on both counts. As has been explained numerous times, no miracle is needed for the first BFO (log on request) to be in the ballpark of the steady-state value after an extended OXCO power outage in flight. After being cold soaked, the first transmission is enabled when the temperature error is small (i.e., the temperature is near the set point). The second BFO (the log on acknowledge at 273 Hz) has a larger transient error than the first BFO due to servo overshoot when the OXCO is warming up from being cold soaked.
The 18:25:34 BFO of 273 Hz cannot be explained by any ordinary aircraft maneuver, or even a fairly extraordinary one. If you think this is not the case, please provide ground speed, track, and rate of climb that would generate a 273 Hz BFO, and then show how a B777 can physically do this and then produce the subsequent BFOs and BTOs over the next 3 minutes. Otherwise, please stop wasting people’s time by repeating previously disproven claims.
@ALSM,
I noticed that the six T-channel BFOs at ~18:28:10 have two values – 148 and 147 Hz. In what way are they “redundant” or perhaps more precisely, “correlated”? Just because they occur close together in time does not necessarily make them correlated. Does the frequency measurement for each BFO look at a unique time-section of signal, or is there some overlap?
@TBill
@Victor
Victor has estimated that my hypothesised 10N to 45S1 via BEKUT would come up about 673 NM short based on starting fuel of 67,200 kg at FL400. However, I noted your comments about winds when modelling CMT 180 from 10N.
I’m not familiar with how FS9 models high level winds but if it incorporates typical patterns then much of 101059N0901328E 095356N0891351E BIKEN N571 BEKUT 641648S1363524E would be flown with favourable winds and there shouldn’t be too much exposure to unfavourable conditions.
If you ever get the chance to fly 101059N0901328E 095356N0891351E BIKEN N571 BEKUT 641648S1363524E with weather similar to that set for the Captain’s sim I’d be very interested to hear how close to 450507S1040844E fuel exhaustion occurs.
@Mick Gilbert: If 10N (DOTEN) – BEKUT – S45 doesn’t quite fit the fuel on board at 10N, perhaps you should consider
10N – P762 – MABAL (Colombo) – 45S1, or-
10N – P762 – MABAL – G465 – MQ36 (Male) – 45S1.
@Gysbreght
Thanks for those suggestions. Happenstance or otherwise BEKUT – SMP (641648S1363524E) comes within in whisker of 45S1, a couple of thousand metres or less. Once you start with points off that track the alignment with 45S1 diminishes.
BEKUT is on the route for both MH150 and MH16, and being on the west coast of India, might be where you’d decide to give either of those routes away and try something else in the sim.
The last 2200 nm of BEKUT – 45S1 should be in reasonably favourable wind conditions so I’m keen to see how much of a leg up that provides.
@Mick Gilbert: You asked before about the winds and I responded that the 5N data set is consistent with “Fair Weather” with winds of 25 kn from a direction of 270°M. The initial part of the route from 10N to BEKUT is into a headwind. The segment from BEKUT to 45S1 has a small favorable tailwind. My quick estimate is that including the wind reduces the air miles from 10N to BEKUT to 45S1 only by about 30 NM.
@Victor
Okey doke, thanks for that update on the winds in FS9, Victor.
@Victor
Victor, I have taken the time to re-read your and Yves’ paper and I note your explanations for the 10N climb and 45S2 AGL anomalies, namely user interventions.
I am not looking to argue for argument’s sake but I feel there are a few shortcoming in your analysis of the recovered data. Respectfully, I think that your conclusion over-reaches on the data that is presented; I think that you have not considered all of the evidence that was available and I think that you have not taken the opportunity to sufficiently explore alternative explanations
On the last point, you and Yves do not at any point explore the notion that any of the 8 data sets recovered from the Mk25 drive may be snapshots of different flights despite a reasonable body of evidence suggesting that that may be the case. In many instances you actually address evidence that suggests that there was more than one flight.
“There were two other data sets found on the captain’s computer showing the aircraft parked at KLIA. One (Coordinate 6-2) had all the available aircraft tanks, including the left and right auxiliary tanks, fueled to 100%. The other (Coordinate 7) had the four side tanks (main and auxiliary) fueled to 80.3% and center tank at 0%. As these fuel levels were not consistent with the levels for data set 3N (Coordinate 1), these data sets are not analyzed here. These levels might have been intermediate values before the final values were selected and the simulation was initiated.” … OR they may be evidence of more than one simulated flight.
“For the segment between 5N and 10N, the fuel efficiency for the recovered data was 2.8 kg/nm, while that for the recreated flight was 14.4 kg/nm. This suggests that during this segment, either the user positioned the aircraft forward along the path,
or fuel was added.” … OR that 5N and 10N are not related and are from two separate flights.
“For the segment between 10N and 45S1, the fuel efficiency for the recovered data was 18.5 kg/nm, while that for the recreated flight was 13.8 kg/nm. This suggests that fuel was removed by the user during this segment, reducing the range and leading to fuel exhaustion at 45S1.” … OR that 10N and 45S1 are not related and are from two separate flights OR that the airplane did not fly directly from 10N to 45S1.
“It can be seen in Table 2 that points 3N and 5N share the same values for MinimumGForce, MaximumGForce, and MaxReachedEngineRPM (red boxes), and therefore are likely from the same flight simulation. Similarly, points 10N, 45S1, and 45S2 share the same values
for these parameters (blue boxes)” … and therefore are likely from a different flight simulation to that for points 3N and 5N.
Despite addressing the matter of context in the paper, you do not take the opportunity to explore contemporaneous context, specifically that provided by the other evidence detailed in the RMP analysis of the flight simulator hard drives;
– Section 6.4.3.3 Logbook.Bin file 23 Jan 2014 @ 08:56 pm, that shows the B777 simulation was used that day for 5 flights, comprised of four originating from WMKK (Kuala Lumpur) of which one terminated back at WMKK with one landing, and one flight of an indeterminate origin (suggesting that at least one flight may have been created in mid-air).
– Section 6.4.2.1 The Flight Simulator Crash Report of the Flight Simulator, that shows that three software application crash reports were generated on 23 January 2014 at 5:44:33 PM, 7:53:21 PM, 8:56:26 PM (the last crash report coinciding with the last B777 log book entry).
– Three of the recovered data set records – 2N-2, N5 and N10 – show evidence of systems errors (NTDLL.dll, KERNRL32.dll, msvcrt.dll, etc and include the phrase “this program cannot be run in dos mode”) that suggest that they are related to the three application crashes.
I’m not going to harp on this but given the evidence that:
a. Five B777 simulations were logged;
b. Three aplication software crashes were logged;
c. The last B777 simulation and the last software crash were logged at the same time;
d. There are at least three data sets – those showing the airplane at KL – that appear unrelated to one another; and
e. Only two sets of data sets – 3N-5N and 10N-45S1-45S2 – appear to be quite clearly related to another,
I think that it is entirely unreasonable to contend that:
1. there was only one flight,
2. all the data sets were created manually, and
3. 3N-5N-10N-45S1-45S2 are part of one flight.
Any old how, right, wrong or indifferent, that’s my somewhat long-winded piece on the matter.
@Mick Gilbert: I’ve already explained that in the RMP report, the crash logs pertain to FSX on MK26, not FS9 on MK25. For all we know, he tried to use FSX, got frustrated, and used FS9. If you are going to present flight logs and crash logs as evidence, you have to correctly assign that evidence to the correct drive and installation.
Regarding evidence of moving the plane forward on the path, we know that the user manually changed the position of the plane before the save, based on the resolution of the coordinates, as explained in the paper. So we have evidence of too much fuel for continuous flight and evidence the plane was moved.
Regarding manual creation of data sets, the paper explains the specific sequence of user inputs required to produce the anomalies in velocity vectors. Automatic saving would not do this. Clearly, data sets 3N, 10N, 45S1, and 45S2 were manually created.
Regarding the references to DLLs in the file fragments, that is not evidence of a system crash of FS9. The flight files are text files, and I have never seen a dump of core memory in flight files for an application crash, nor ANY documentation suggesting that would occur. If you have ANY evidence that this occurs, please show me. Rather, those are blocks of data in the Shadow Volume not related to the flight files.
There were eight data sets found in the Shadow Volume of MK25. Of those, three have the plane on the ground. Five were in the air (3N, 5N, 10N, 45S1, and 45S2). One on the ground (2N) can be linked to one in the air (3N) by fuel level. Two in the air (3N, 5N) can be linked by one set of value markers, and three in the air (10N, 45S1, 45S2) can be linked by another set of value markers (10N, 45S1, 45S2). The sequence of the data sets are linked by the progression of coordinates and the progressively decreasing fuel levels (2N, 3N, 5N, 10N, 45S1, 45S2).
Regarding your claim that (3N, 5N) were different flights than (10N, 45S1, and 45S2) because the value markers are different, when I recreated the flight, I saw a similar grouping of value markers. There was already a discussion on this blog about this. For instance, the maximum RPM is not experienced at takeoff when a typical thrust re-rate is used by setting a typical (higher that actual) outside air temperature (OAT).
To me, there is strong evidence suggesting a single flight consisting of 2N, 3N, 5N, 10N, 45S1, and 45S2, which is what was claimed in the paper. Can the evidence be used to state with 100% certainty that the data sets are from the same flight. No, and I never made that claim.
We might agree to disagree about what probability to assign to a single flight. But I believe it is false to claim that there is no evidence to suggest the data sets are related to a single flight.
…so I see 180S CMH from 1090 with Fair Weather hits about 10nm from 45S1 at M.84 using 2005 mag headings. Wind impact is about +2 deg latitude (92 nm) and Magnetic Table version is about +1 deg latitude (45 mm) for the older mag tables in FS9 orig
@Bobby,
@ALSM,
Yes, 13 samples, not 12 – my mistake. The last one was on the next page of Inmarsat’s logs, and I missed it. But I am still puzzled why Mike counts 7 or 8 samples. Independent in terms of what? 0.5 s is a fairly long interval – a lot of things can happen during it, especially if a turn or descent was in progress.
@Bobby,
@Dennis,
Re: “The 18:25:34 BFO of 273 Hz cannot be explained by any ordinary aircraft maneuver, or even a fairly extraordinary one. If you think this is not the case, please provide ground speed, track, and rate of climb that would generate a 273 Hz BFO, and then show how a B777 can physically do this and then produce the subsequent BFOs and BTOs over the next 3 minutes. Otherwise, please stop wasting people’s time by repeating previously disproven claims.”
It looks like no single hypothesis proposed up to date is capable to explain the two sequences of BFOs during the 18:25 and 00:19 logons.
Oleksandr
T channel bursts do not produce indepedent observations. All the BTO values are identcal, and the BFO values are all within 1 lsb. Inmarsat and ATSB have confirmed that T chan bursts should be treated as one (independent) observation.
@Victor Iannello: “For instance, the maximum RPM is not experienced at takeoff when a typical thrust re-rate is used by setting a typical (higher that actual) outside air temperature (OAT). “
Of course it is entirely rational to assume that Z, rehearsing his suicide mission to the SIO, would worry about the maintenance cost of the engines that the simulator that doesn’t have.
When I asked you about the min/max GForce levels in your ‘recreation’ you replied with “NO COMMENT”.
Re T channel bursts info source: In response to some questions I sent to ATSB, here is part of the 12/19/2014 response:
“Also, I was asked to forward this on to you [from the SWG] following your latest:
It should be noted that the T-channel messages within an SU burst can’t be treated as independent bias measurements. The BTO is only estimated once at the start of the burst and then the same value is reported with each SU in that burst. It appears that rather than 392 T-channel messages there are in fact approx. 31 independent T-channel measurements. Combined with the 30 R-channel messages the “All Channel” average using the IG provided bias values decreases by roughly 6us (from IG numbers -495,657 to roughly -495,663).”
We subsequently revised our T channel BTO bias based on this feedback. The BFO values are not, strictly speaking, copies like the BTO values within a burst are, but they are so close in time that they are almost always within 1 lsb, making them redundant within the measurement error of the CUs.
Oleksandr:
Re your statement: “It looks like no single hypothesis proposed up to date is capable to explain the two sequences of BFOs during the 18:25 and 00:19 logons.”
Your statement above is 100% false, and you know it is false. Please stop with the Fake News.
I for one did propose just such a single, consistent hypothesis. Bobby not only agrees with the hypothesis explaining both logons, he has demonstrated that the 18:25 data fits the classic, predicted cold start thermal response. Many have verified that the 00:19 data fits the steep descent profile.
You are welcome to disagree with the hypothesis, even though quite compelling in my view, but you ate not free to deny the very existence of the hypothesis.
@Victor Iannello: “To me, there is strong evidence suggesting a single flight consisting of 2N, 3N, 5N, 10N, 45S1, and 45S2, which is what was claimed in the paper. “
Let’s face it, the scenario is patently absurd.
So Captain Z, planning to end his life by highjacking a B777 entrusted to his care, wanted to test fuel exhaustion in a remote location in the Southern Indean Ocean.
The simple way to do that was to take the FLT file saved from an earlier exercise, modify the location to 45S 104E, the altitude to whatever he wanted it to be, and the fuel remaining to zero, and start the simulation like you did in at least one ‘recreation’.
Instead in your hypothetical scenario he started from Kuala Lumpur International Airport, went though the entire pre-takeoff checklist to set all the necessary parameters, conducted a takeoff with derated thrust to please the virtual engines, but in rodeo style to produce Max/Min Gforces of 1.44/0.60 g, interrupted the simulated climb at 3N, FL230 to change flight parameters such as location, altitude and fuel, saved a .FLT file, continued climbing at derated climb thrust through FL230 and FL320, pausing the simulation again at 5N to add fuel. Before reaching 10N he did a step climb at maximum climb thrust (defeating the earlier derate) to FL400 while doing some aerobatic manoeuvres with Min/Max GForce levels reaching 2.2/0.15 g. Then he turned towards 45S 104E where he again stopped the simulation to change the altitude to FL357 and to reduce the fuel remaining to zero, and again saved a .FLT file.
Why did he do all that?
@Gysbreght said, Of course it is entirely rational to assume that Z, rehearsing his suicide mission to the SIO, would worry about the maintenance cost of the engines that the simulator that doesn’t have.
Flight simulator enthusiasts normally de-rate take-off thrust in the course of setting the performance parameters. Try taking off in a B777-200LR at a light weight without de-rating the take-off thrust and see what happens. Besides being snide, you are showing your ignorance.
@Gysbreght said, When I asked you about the min/max GForce levels in your ‘recreation’ you replied with “NO COMMENT”.
Now you are outright lying about me, and I won’t tolerate it. Here is the actual exchange, provided with links:
@Gysbreght said:
@Victor Iannello: With all due respect, you have stated the obvious but did not answer my question.
I’m interested in the values of acceleration in the particular simulation you did for Recreation of Simulated Flight in the particular report. I expect that these add some perspective to the values recovered from Z’s disks.
I’m also particularly interested in the engine rpm’s because they are different in your red and blue boxes. According to your report you programmed the intended flight plan into the FMC and conducted most of the simulation with the autopilot in control. I assume that you conducted a normal takeoff from Kuala Lumpur International airport, cleaned up the airplane, reduced thrust to CLB, and engaged the autopilot. The takeoff rpm would then be set by the engine control system and would not be exceeded in the remainder of the flight under AP control with CLB rating selected.
I’m not asking for replication of “min/max values of acceleration and rpm”.
Victor Iannello said:
Gysbreght: The values in automated flight are different than the values when parts of the flight are manually flown. In automated flight, the acceleration stays between 0.8g and 1.2g. With the autopilot disengaged, I can replicate the values from the sim data. In my automated simulation, max g of 1.2 and max rpm of 32,328 rpm was realized in the stepped climb between 5N and 10N, and not the takeoff, which produced a max g of 1.1 and max rpm of 29,195 rpm. The max rpm at takeoff depends on how the thrust is de-rated.
Gysbreght said:
@Victor Iannello: Thank you ever so much. That is the helpful information I was looking for.
As to “The max rpm at takeoff depends on how the thrust is de-rated.”:
It doesn’t make sense to derate takeoff thrust to less than CLB, it is something that Z wouldn’t do.
****
Not only did I provide you with an answer, you even thanked me for my reply!
Later in that exchange, I explained that pilots often enter an assumed outside air temperature (OAT) as high as 60°C to reduce thrust. You disputed this. Our B777 pilot Andrew agreed with me.
I’ve warned you before about your comments. Your comments going forward will only be approved after moderation.
@Gysbreght said, “Let’s face it, the scenario is patently absurd.”
It might seem less absurd if you kept your facts straight. Here’s some help:
“So Captain Z, planning to end his life by highjacking a B777 entrusted to his care, wanted to test fuel exhaustion in a remote location in the Southern Indean [sic] Ocean.”
I’ve said many times that I don’t know why he created the simulation. We can only speculate. That doesn’t mean he didn’t do it.
“conducted a takeoff with derated thrust to please the virtual engines,”
Take-off thrust derating is standard procedure for both real world pilots and sim pilots, as I have already explained but you refuse to accept.
“but in rodeo style to produce Max/Min Gforces of 1.44/0.60 g”
The gforces were either created in manual flight OR by the transients that occur when the MAP parameters are changed. The transients introduced by the change in MAP parameters produce abnormal vertical speeds, as explained in the paper.
“continued climbing at derated climb thrust through FL230 and FL320, pausing the simulation again at 5N to add fuel.”
Read the paper. We never said fuel was added. At times, the position was changed, as evidenced by the resolution of the coordinates in the flight files. The only time we said fuel was adjusted was to create fuel exhaustion in the SIO.
“Before reaching 10N he did a step climb at maximum climb thrust (defeating the earlier derate) to FL400”
As I explained to you before, and our B777 pilot Andrew agreed, it is not uncommon to see a climb thrust higher than the take-off thrust. That doesn’t “defeat” the take-off derate, as the air is colder at high altitudes.
“while doing some aerobatic manoeuvres with Min/Max GForce levels reaching 2.2/0.15 g.”
See my previous comments about reasons the g levels might have been high.
“Then he turned towards 45S 104E where he again stopped the simulation to change the altitude to FL357”
Wrong. We didn’t claim he changed the altitude at 45S1, as the Altitude and AGL were the same values in the flight file. However, the 45S2 data set does clearly show he changed the altitude from 37,654 ft to 4,000 ft just before saving that flight file.
“Why did he do all that?”
Nobody knows for sure why he might have created the simulation.
@ALSM,
Re: “Your statement above is 100% false, and you know it is false. Please stop with the Fake News. ”
No, your hypothesis does not hold the water, and you know about it. Earlier, I have pointed a number of inconsistencies, but you opted to keep silence. Please, until you are able to defend your hypothesis and close all the gaps, do not claim it is capable to explain everything.
“for one did propose just such a single, consistent hypothesis”
The one you proposed has been already discarded. It features too many inconsistencies. What else?
@Victor. “Nobody knows for sure why he might have created the simulation.”
Or why the simulation should be different to the actual, particularly 45S, or whether there were others like it, or where the other ground configuations come in.
As to motivaton, in the investigation human factors people might help, supposing Malaysia is apprised of all this. Are they aware do you think?
@David asked, “Are they aware do you think?”
I don’t think the Malaysians have demonstrated a strong interest in exposing the truth.
@Victor. They could claim they were unaware. Send them a copy?
@Dennis
Thanks for your clear thoughts on motive.
I think the greatest challenge is trying to come up with a single explanation. This is one way to look at multiple causes.
The flight began with an intentional diversion to negotiate over political injustice in Malaysia. Malaysia might have scrambled and pursued. When they saw they couldn’t catch the plane they fired off a missile as a warning. The missile struck the plane. The oil platform saw this. The missile depressurized the plane. The plane continued flying as a ghost flight. Kate tee saw the glow in the sky. No one expected it to stay in the air that long. No one was aware of the Inmarsat messaging. Malaysia held tight and attempted no further contact. They wouldn’t want the pilot reporting that they had been hit. No one truly knows where the plane is.
Oleksandr:
Contrary to your repeated false claims, there are no inconsistencies…zero… in the 0019 rapid descent evidence and hypothesis, or the 1825 OCXO second order thermal control loop response and effect on BFO, which fully explains the “odd BFO values”. Sorry you don’t seem to be able to comprehend the engineering analysis. I’ll leave it there.
@All:
I continue to optimize the website for speed. You should already see an improvement.
Victor:
It is working much better today. Thanks for tuning the website.
Here’s a link to Agenda Awani’s (a Malaysian news/cuurent affairs show’s) third anniversary special on MH370. It’s nearly 49 minutes long and features Blaine Gibson amongst others.
http://www.astroawani.com/video-malaysia/agenda-awani-mengenang-mh370-146070
It’s hard to follow in parts as they switch between English and Malaysian. At around the 20 minute and 36 minute marks they appear discuss the apparent anomaly associated with the signal strength of the airplane’s satellite transmissions increasing as the flight went on.
Mick:
Unfortunately, the two “experts” sitting with Blaine are absolutely clueless when it comes to the interpretation of the amplitude of those inbound C Band signals relayed by I3F1 from MH370. I found that very surprising, given their credentials. But they really don’t understand the limitations to inferring AES-I3F1 range from the amplitude of a signal relayed by satellite. The amplitude recorded by the Perth CUs is a function of many terms, most of which are orders of magnitude larger than the AES-I3F1 range term. There is no way to isolate what is a “range signal” from transponder gain changes, multipath, AES EIRP changes, AES antenna gain as a function of OBC angles, etc.
I corresponded with DR Aduwati Sali shortly after the interview aired. I explained in detail why the inbound SCPC amplitude could not be used to derive range (arcs). She did not seem to be interested in the information I provided, insisting they were on to something.
Here are 2 emails with some details:
+++++++++++++++++++
Sent March 23, 2017
Dear DR Aduwati Sali:
Thank you for responding. Unfortunately, the inbound signal strength recorded at the Perth GES cannot provide any useful information about the location or path of MH370. The effective “range signal” to “range noise” ratio is far too low to pull any useful signal from the noise.
As I’m sure you know, the propagation loss varies with range according to the inverse square of the range. In dB, the loss between isotropic antennas is = -37 – 20*log(f) -20*log(d) where f=frequency in MHz and d= range in miles. The distance from the I3-F1 satellite to the sub-satellite point (90º AES elevation angle) is 22,236 miles. The distance from the I3-F1 satellite to the horizon (0º AES elevation angle) is 25,898 miles. Thus, over the entire “full disk coverage” of the I3-F1 global beam, the range only changes from 22,236 miles to 25,898 miles. Assuming a “perfect world” with no other sources of change to the signal strength, the signal strength would change only 20*log(22,236/25,898)= -1.32 dB over the complete global area of coverage. Of course, we know that the AES elevation angle only varied from ~40º to 55º, so the change in signal strength (between ideal isotropic L band antennas) was only a few tenths of a dB for the entire flight. Meanwhile, the normal variability in signal strength due to the AES antenna pattern, the variable EIRP control, transponder loading, etc. causes the receive signal level recorded by the CUs in Perth to vary by 5-8 dB under normal conditions. These are the sources of the variability observed.
The bottom line is that the variation in signal strength recorded (and noted in the Inmarsat October 2014 JON paper) was due almost entirely to changes in AES antenna gain as a function of pointing direction, AES uplink EIRP setting, transponder loading, etc., and not due to range changes. Thus, no useful information can be squeezed from the amplitude data. If you have any questions, please let me know.
Regards, Mike
++++++++++++++++
Sent Wed, Mar 29, 2017
Dear DR Aduwati Sali:
Unfortunately, for the reasons I gave below, it is impossible to use signal strength data to calculate the range from the I3-F1 satellite to 9M-MRO. The sensitivity of the measurement is extremely low, and the noise level is 10^8 higher than the “range signal”. Thus, the signal is buried deep in the noise.
We know from the BTO data and basic physics that the range between the I3-F1 satellite and 9M-MRO varied between 22,832.7 miles and 23,526.5 miles. The average range was 23,179.6 miles. At this range and elevation angle, the sensitivity is 0.00026 dB/mile. The change in signal strength (due to range change) between the extremes over the whole flight was only 20*log(23,526.5/22,832.7) = 0.26 dB.
The noise is many orders of magnitude greater. As previously noted, the L band uplink EIRP is controlled by the system in 0.5 dB steps. It can vary several dB over a flight. The L band antennas on each end of the link have pattern ripple of several dB. Transponder loading varies over the flight. These and a dozen other link variables contribute 5-8 dB of variation in the receive signal strength for a constant range. Thus, it is impossible to “see” the 0.00026 dB/mile “range signal” against 5-8 dB in background “range noise”.
Rain attenuation is virtually zero at L band and only a few tenths of a dB at C band in strong rain. Perth was dry, so no C band variation over the time of the flight. 9M-MRO was above the boundary layer for the whole flight. Thus neither link suffered from any material rain attenuation. More importantly, even if one or both links did have some rain attenuation, that would only add to the noise, not provide any additional signal of use.
I hope this explanation will put the issue to rest. Please confirm. If you have further questions, please let me know.
Regards, Mike
@ALSM
Thank you for the explanation regarding the correct interpretation of the amplitude of the inbound signals. I could recall seeing where the “anomaly” had been explained away but was struggling to find where; you’ve saved me a lot of search time.
If nothing else the Agenda Awani segment shows that there’s at least a bit of journalistic fire in Malaysia regarding MH370 – that can’t be a bad thing.
@Victor
Do you see merit in further examination of 180S Mag from 1090E (given apparent close approach within 10 nm of McMurdo path)? I do not know if continued testing could produce closer match (eg; at 37500 alt vs. 35000 that I used).
180S fits my current thinking of a 180S flight path. Implication could be that McMurdo was not a waypoint in the sim studies. Also implies Z may have updated to 2005 mag table in FS9, which perhaps gives some insights.
@Victor
Victor, thank you for the detailed response.
If you wish to separate the flight logs on the Mk26 drive from the data recovered from the Mk25 drive then you have a problem; the Mk25 data sets have no dates associated with any of them – you don’t know when they were created. As I understand it, the only evidence that associates the data sets to 23 January 2014 is the Mk26 flight log data. If you maintain that the Mk26 log book files are not associated with Mk25 data sets then by what mechanism can you determine when they were created? or that they were even created contemporaneously?
Can the locations for storing FS9 and FSX flight log and individual flight data be set to be independent of the location that the application itself is stored? I would have thought so.
There is, however, evidence in the Mk25 data that suggests it correlates with the Mk26 flight logs, specifically the generally ignored “AG refrence (sic) – Coordinate 7″ – that data set shows a B777 with positive values for GForce and MaxReachedEngineRPM on the ground at KL (a word of caution when checking this in the RMP report, the relevant pages (61-64) are out of order, 61 and 62 are reversed). The data suggests an airplane that has flown and landed; MaximumGForce=1.4203252562828395
and MinimumGForce=0.42068584730685316 are similar values to what we see at 3N (I’m not suggesting that they are part of the same flight). The airplane also has AccBodyAxis data that is inconsistent with the 2N data showing an airplane on the runway before flight. All of that is entirely consistent with the flight log showing that a B777 flight originated at WMKK, terminated at WMKK and entailed one landing. Whether viewed in conjunction with the flight log data or independently, Coordinate 7 is evidence that there was definitely more than one flight.
With regards to how application software crashes manifest themselves in MS Fight Simulator, I have no idea. What I have learned by checking some flight sim forums is that the hardware configurations for non-virtual flight deck components such control yokes, rudder pedals, etc is a common cause of problems when running the PSS Boeing 777-200LR No VC program. I’m assuming that you have not attempted to recreate the same non-virtual cockpit set up as used by the Captain as it would be simply unreasonable to do so.
Regarding evidence of moving the plane forward on the path, we know that the user manually changed the position of the plane before the save, based on the resolution of the coordinates, as explained in the paper.”
You and Yves determined one means by which the same coordinate resolution as seen in some of the recovered data sets can be achieved. Surely you would not claim that your research in that regard was completely exhaustive? For instance, do you get the same result by retrieving a previously saved flight and dragging it to a new location to start a “new” simulation? Presumably by using the same previous flight you can drag the airplane to two different locations for two different simulations and both will “inherit” the original flight’s parameters; thus two different simulations drawn from the same saved flight may appear to be part of one flight when they are not.
Did you attempt to recreate application crashes? Just because you found one method does not mean that it is the only method. Accordingly, you most assuredly do not know that that the user manually changed the position of the plane, at all; and even if he did manually change the position of the plane that does not mean that we’re looking at just one flight.
“One on the ground (2N) can be linked to one in the air (3N) by fuel level.”
It is curious that you say that when your and Yves’ own recreated flight fails to support that and, in fact, suggests otherwise. There is no correlation between the fuel level at 3N and that at 2N and the implied fuel consumption for 2N-3N is 13.5% lower than what you and Yves recorded when you recreated that leg. So, where is the link?
Regarding your claim that (3N, 5N) were different flights than (10N, 45S1, and 45S2) because the value markers are different, when I recreated the flight, I saw a similar grouping of value markers. There was already a discussion on this blog about this. For instance, the maximum RPM is not experienced at takeoff when a typical thrust re-rate is used by setting a typical (higher that actual) outside air temperature (OAT).
While I accept that the MaxReachedEngineRPM would most likely occur in the section of the climb after 30,000 feet you would expect to see at least some evidence of increasing engine rpm between 3N (23,250 ft) and 5N (32,250 ft); that is, if we expect to see higher engine rpms at 10N we should already be seeing evidence of it at 5N. This would be particularly so given that the recovered data for 5N shows a 244,326 kg airplane climbing at 1,456 fpm; that is three times the average rate of climb between 3N and 5N.
Just by the bye, do you have any thoughts on the 1.2% difference between the MaxReachedEngineRPM numbers for the left and right engines at 3N and 5N? That asymmetry is odd, isn’t it?
I also accept that you may see slightly higher GForce numbers during the latter part of the climb compared to take-off but you would expect the difference to be in the order of 10%-15%. At 10N not only do you see a MaximumGForce that is in excess of 50% higher than at 5N but you also see a MinimumGForce number that is 75% lower. That’s an odd combination, not inexplicable by any measure, but it is not evidence to support a connection between 5N and 10N.
You can carve the evidence up whatever which way but there’s no escaping the fact that the higher fuel load and higher MaxReachedEngineRPM associated with 10N are together generally consistent with a take-off weight that is higher than that associated the fuel load and lower MaxReachedEngineRPM associated with 3N and 5N. That, in of itself, is internally consistent evidence that we may be looking at two different flights.
At the end of the day, we clearly have a different view of what constitutes “strong evidence”. By my reading;
1. there is no evidence suggesting that 2N and 3N are from the same flight and one item of evidence (fuel levels) suggesting that they are not,
2. there are two items of evidence (GForce and MaxReachedEngineRPM) suggesting that 3N and 5N are from the same flight and one item of evidence (fuel levels) suggesting that they are not,
3. there is no evidence suggesting that 5N and 10N are from the same flight and two items of evidence (GForce, fuel levels) suggesting that they are from different flights, and
4. there are two items of evidence (GForce and MaxReachedEngineRPM) suggesting that 10N and 45S1 are from the same flight and one item of evidence (fuel levels) suggesting that they are not.
For fear of oversimplification, using a simple for/against metric and giving equal weight to each element of evidence yields the following conclusions;
a. 2N and 3N are not the same flight (0/1),
b. 3N and 5N are probably the same flight (2/1),
c. 5N and 10N are not the same flight (0/2), and
d. 10N and 45S1 are probably the same flight (2/1).
For the sake of clarity and the avoidance of doubt, I don’t think that I’ve ever said that there is no evidence to suggest the data sets are related to a single flight. That said, looking at a. and c. above, I’ve got to say that I’m now inclined to do so now. I am more than happy to agree that there is some evidence linking some of the data sets but you can’t stitch them together into one flight without supposition and supposition is not evidence.
Any old how, on a related matter, I have found a reference to MAS B777 Simulator Training Lesson 3G which is described as folows:
“LESSON SUMMARY
This lesson is a flight from Kuala Lumpur (WMKK) to Chennai (VOMM) with an in-flight diversion required. The crew performs a limited preflight while the instructor starts the engines and configures the overhead panel. Normal procedures and selected non-normals will be practiced. This flight emphasizes in-flight diversion considerations and procedures. Engine failure after V1 characteristics and procedures will be introduced.”
The performance data lists GR WT 230,000KG, FUEL 70,000KG, COST INDEX 18, DERATE D TO 1. The flight profile sees a departure from RWY 32R followed by a climb to FL310 with a dual engine failure/stall occurring about 40 minutes into the flight. An engine inflight start is followed by a diversion back to KL for ILS04.
If you accept the premise that we may be looking at more than one flight, it would not be difficult to reconcile the 5N data set (amd perhaps 3N) with that training lesson – the fuel’s about right, the altitude’s about right, the heading’s about right and the location’s about right. I’m not suggesting that there’s any evidence that 3N and 5N are part of training lesson 3G but I don’t think that it would be beyond the realm of possibilities that the Captain may have been replicating MAS training lessons on his home sim.
Mick Gilbert:
“Whether viewed in conjunction with the flight log data or independently, Coordinate 7 is evidence that there was definitely more than one flight.”
I think I was clear that Coordinate 7 was not part of the flight described by sets 2N, 3N, 5N, 10N, 45S1, and 45S2. What we are discussing is whether those 6 data sets were the same flight. I maintain that the evidence shows they were. The fuel levels and coordinates are consistent with the sequence of the points, and there are other indications, as we’ve shown.
“Did you attempt to recreate application crashes? Just because you found one method does not mean that it is the only method. Accordingly, you most assuredly do not know that that the user manually changed the position of the plane, at all; and even if he did manually change the position of the plane that does not mean that we’re looking at just one flight.”
Clearly you misunderstand what we did. We did not deliberately create crashes, nor did we find a method to cause them. There is no evidence that FS9 was experiencing crashes. What we found was a way to create the anomaly in velocities that causes a transient when the simulation is restarted. More recently, I stated that when the application did crash, it did not dump memory contents into flight files, which is what you claimed. As for moving the plane, we found a way a sequence of user operations that explains the resolution of the position, the fuel level, and the abnormal velocities. I think that is more than just a coincidence.
“There is no correlation between the fuel level at 3N and that at 2N and the implied fuel consumption for 2N-3N is 13.5% lower than what you and Yves recorded when you recreated that leg.”
I don’t know how close you expect the fuel levels between the simulation and the recreation, but considering how much fuel is consumed in takeoff, and differences that may exist in time on the runway, derating, other configuration parameters, and how the plane was flown at takeoff and early climb, I think a 13% difference in fuel consumed is not very different that early in the flight. In terms of fuel levels, the recovered data was 65.1 MT, and the recreated data was 64.6 MT. On the other hand, I would be surprised to see a 13% difference in fuel mileage during cruise.
Your approach seems to be to find any possible way that the data sets are not related. My approach is to find the most likely explanation for the relationship between the data sets that were found.
Just the fact that there are three points (10N, 45S1, and 45S2) that share marker values and progress from the Andaman Islands and end with fuel exhaustion in the SIO is a strong suggestion that the simulated flight and the disappearance are related. It would be an incredible coincidence if they were not.
@VictorI
“Just the fact that there are three points (10N, 45S1, and 45S2) that share marker values and progress from the Andaman Islands and end with fuel exhaustion in the SIO is a strong suggestion that the simulated flight and the disappearance are related. It would be an incredible coincidence if they were not.”
Indeed. And all other facts and indications also point to a deliberate and well planned disapperance also. And the only one who could execute this plan with commanding power on the flight and all knowledge required was a very experienced instructor-pilot, a para-glider pilot, a sophisticated home-simulator expert/enthausiast, the captain of this flight.
Imo the discussions on BTO’s, BFO’s, fuel-data, possible flightpaths and scenarios are kind of exhausted.
Imo the hard evidence is now coming from the debris and the drift data.
They show physical proof of a most probable crash area and crash-attitude of the plane.
Like Jeff Wise latest topic on debris shows this is what’s going to tell the story much better in the end.
Like you say; the simulations on the hard-disks would be an incredible coincidence. They are obviously not regarding all the other facts that are known.
This disappearance must have been planned from the start till the very end.
Which means imo the end was also carefully planned.
And the only motivation I can think of regarding disappearance is disappear as good as possible to try and relieve your family and friends from possible consequences.
The way to achieve this could only be accomplished by ditching the plane in a very remote area leaving as little debris as possible.
All we can say wether you agree or disagree with a scenario like this, he succeeded very well till now.
I’m waiting for new debris to be found and key-reports on the already found debris.
@ALSM,
“Contrary to your repeated false claims, there are no inconsistencies…zero… in the 0019 rapid descent evidence and hypothesis, or the 1825 OCXO second order thermal control loop response and effect on BFO, which fully explains the “odd BFO values”. Sorry you don’t seem to be able to comprehend the engineering analysis. I’ll leave it there.”
It seems you are not able to understand several obvious flaws of your analysis. Yes, leave it there.
@Ge Rijn said, “Imo the hard evidence is now coming from the debris and the drift data.”
That appears to be the case.
Update on May 31, 2017.
I was recently in a discussion that included a well-known ocean explorer who happens to be a judge in the Ocean Discovery XPrize competition. We were having a general discussion about searching for MH370 and ways to scan the ocean floor at high resolution, and he told us about the capabilities of Ocean Infinity. Like the team at Virginia Tech, their approach is to employ a team of AUVs. From their website:
Six HUGIN autonomous underwater vehicles (AUVs) are capable of operating in 6,000 m water depth collecting high resolution data at record breaking speeds. Our AUV fleet is accompanied by six unmanned surface vehicles (USVs) to ensure precise position and constant communication.
With multiple autonomous vehicles working simultaneously utilizing innovative technology, we are able to survey huge swaths of the seabed, quickly and with outstanding accuracy. We can operate in shallow waters but excel in extreme depths, working in dynamic environments ranging from the tropics to the Arctic ice.
Because of the size and complexity of each AUV/USV pair, the capital cost of the technology from Ocean Infinity would greatly exceed the capital cost of Virginia Tech’s technology, which uses small AUVs with innovative navigation systems. On the other hand, both approaches benefit from having a single host vessel supporting multiple underwater vehicles, which offers significant operating cost and scan rate improvements compared to the conventional towfish technology.
Ocean Infinity’s seabed exploration system is commercially available today, including underwater and surface vehicles, on-board support equipment, and the host vessel. This is an exciting possibility for conducting the search for MH370 in the near future.
From The Australian this morning 1st June
UPDATED: Authorities are investigating an attempted aircraft hijacking after a passenger claiming to have an explosive device was detained for allegedly threatening to enter a Malaysian Airlines cockpit as it left Melbourne overnight.
Passengers were forced to overwhelm and detain the alleged attacker who struck about 20 minutes after the flight left Melbourne.
Malaysia Airlines Flight MH128 landed at Melbourne Airport shortly before midnight, with tactical response police storming the Airbus A330, which had been en route to Kuala Lumpur.
No-one was injured but the incident, in which the attacker allegedly brandished electronic equipment, has sparked chaos at Melbourne Airport and the temporary shutdown of the main runways.
0:00
/
3:40
Man Detained in Melbourne After Trying to Enter Malaysia Airlines Cockpit
Police said early today the man was in custody and that the device he was allegedly brandishing had clearly not been explosive.
The Herald Sun reported that police understood the man too have a history of mental illness.
Superintendent Tony Langdon said: “Approximately 4km out from Melbourne Airport a male from economy has left his seat and shouted that he had a bomb and tried to move towards the cockpit area.’’
”It appears he tried to gain access to the cockpit and was restrained by passengers and crew.
”The pilot had then invoked their emergency procedures and returned the aircraft to Melbourne Airport.’’
Armed police took the man off the Malaysia Airlines plane. Picture: Andrew Leoncelli
Armed police took the man off the Malaysia Airlines plane. Picture: Andrew Leoncelli
The man, reportedly a Sri Lankan national, appeared to be drunk.
Extraordinary stories have emerged of passengers overwhelming the man, including allegations that he was carrying something the size of a watermelon.
View image on TwitterView image on Twitter
Follow
saroki @saroki19
#MH128 guy drunk. In plastic cuffs. Looks like the SWAT team was despatched to the flight @MAS
2:31 AM – 1 Jun 2017
314 314 Retweets 77 77 likes
Twitter Ads info & Privacy
Radio 3AW reported a passenger it named as Andy was one of several people to swoop on the accused.
“We literally took off, 10 minutes into the take off and I was sitting in business class 4A seat, and I could hear this idiot saying that he wanted to go in and see the pilot,’’ the Herald Sun quoted him saying.
Staff had told the man to sit down but he responded that he intended to “blow up’’ the plane, with flight crew screaming for help.
”So I jumped up, undid my buckle, and approached him,” Andy told the newspaper.
“I said `Mate, what are you effing doing’?’’
“And he said: `I’m going to blow the f…king plane up, I’m going to blow the f..king plane up.
”And I was going, Oh my god, he’s f…king insane.’’
He said that two other men had grabbed the accused and “mashed the crap out of him’’, restraining him and taking the device.
”It was huge, it was black and it had two black antennas coming off it, but it also looked like an iPhone jack,’’ Andy reportedly said.
”So it could have just been a beatbox thing.’’
Malaysia Airlines said it would be investigating the circumstances surrounding the incident, with the aircraft landing safely at 11.41pm.
The plane was due to arrive in Kuala Lumpur at 5.28am (local time) on June 1.
Melb Airport started accepting flights just after 5.30am after incident on MH128. AFP + @VictoriaPolice stationed at international arrivals. pic.twitter.com/hx2XAy9JJ3
— Simone Fox Koob (@SimoneFoxKoob) May 31, 2017
Midair bomb threat
Earlier the Malaysia Airlines plane was forced to return to Melbourne’s Tullamarine Airport less than 30 minutes into a flight after the passenger allegedly tried to enter the cockpit, claiming to have explosives.
Other passengers reportedly subdued the person who tried to enter the cockpit of the Airbus A330, aviation reporter Brendan Grainger tweeted. News Corp reports MH128 bound for Kuala Lumpur left Melbourne at 11:11pm and reached 6400 feet before turning around and landing at Tullamarine Airport 11:41pm All other flights diverted to Avalon, near Geelong, with Melbourne airport reportedly in lock down.
View image on Twitter
View image on Twitter
Follow
Brendan Grainger @S118869
Armed law enforcement have boarded Malaysia Airlines flight #MH128 after a security incident at Melbourne Airport.
Photo credit – TBA
1:47 AM – 1 Jun 2017
64 64 Retweets 27 27 likes
Twitter Ads info & Privacy
Australian Federal Police confirmed to 9NEWS they were responding to an incident but would not comment further.
@Oleksandr,
You said: “It looks like no single hypothesis proposed up to date is capable to explain the two sequences of BFOs during the 18:25 and 00:19 logons.”
On this point we agree. There are two effects occurring simultaneously at both log-ons. There is an OXCO transient event and an aircraft maneuver.
At 18:25 the OXCO was very cold, and that leads to a significant frequency overshoot (i.e., the 273 Hz BFO). Simultaneously, an aircraft maneuver appears to be taking place (based on analysis of the BTOs and BFOs). If one accepts the 18:22:12 location based on military radar, there is only one (simple) maneuver which fits all the BTOs and BFOs from 18:25 – 18:28 (after correcting the observed BFOs for the OXCO frequency transient). That maneuver is a ~15 NM offset to the right of N571. As an aside, I wonder if a pilot who intentionally hijacked the aircraft would perform such a lateral offset. Perhaps Andrew might venture an opinion on this question.
If one chooses to ignore the 18:22 location by assuming it is unreliable, as Victor has proposed doing, then an earlier turn to a heading greater than 300 degrees can also fit the remaining BFO and BTO data through 18:28. I choose not to ignore the 18:22 location, but effectively to use it for its derived BTO/BFO values. In this case, the lateral offset at a constant altitude, which is a standard ICAO emergency maneuver, is wholly consistent with the satellite data.
At 00:19 the same two things occur(an OXCO transient and an aircraft maneuver), although the details are quite different. In this case the OXCO transient effect is almost nil, affecting the BFOs by a couple of Hz at most. That is because the power outage was brief (~1 minute) and the OXCO did not cool off significantly. The rapidly decreasing BFOs measured then are dominated by the aircraft maneuver, which involved an increasing ROD. A turn may also have been underway, but a large ROD is required in order to create the observed BFOs. In my opinion, these last BFOs are valid and must have been caused by a rapidly descending aircraft.
If you or ALSM think my description above is somehow incorrect or not unique, then please provide an alternative explanation.
@DrB said, “If one chooses to ignore the 18:22 location by assuming it is unreliable, as Victor has proposed doing, then an earlier turn to a heading greater than 300 degrees can also fit the remaining BFO and BTO data through 18:28.”
As you can see in the post Possible Paths of MH370 in the Malacca Strait, when starting at 18:02, the great circle path that best fits the BTO and BFO data during the 18:25 log-on is at a track of 297° and a groundspeed of 499 kn, which is parallel to N571. I don’t dismiss the possibility that during the log-on, the plane was traveling with a right offset to N571. If you accept the 18:22 data, the offset manoeuver had to occur between 18:22 and 18:25. If you don’t accept the data, it’s possible that VAMPI-N571 was entered and then an offset of 15 NM entered at some point after 18:02. I think that’s more likely only because the timing of the offset didn’t have to be precisely timed between the end of radar coverage at 18:22 and the log-on request at 18:25. But as you say, even a track of 300° with the appropriate offset gives acceptable BTO and BFO errors.
“As an aside, I wonder if a pilot who intentionally hijacked the aircraft would perform such a lateral offset.”
It would have to be a hijacker fairly knowledgeable about air traffic. It wasn’t until Andrew told us that an offset of 15 NM is recommended if there is no traffic control clearance that most us were aware of this. (Pilot Barry Martin of the IG also advised us that a pilot would fly a busy route with an offset to avoid traffic.)
@DrB
“As an aside, I wonder if a pilot who intentionally hijacked the aircraft would perform such a lateral offset.”
I guess it would largely depend on the hijacker’s intent. If the purpose was to fly as far as possible in order to disappear without the risk of conflicting with other aircraft, then the hijacker has several options: fly at an altitude that is different to those used by other aircraft (eg standard level ±500ft), fly an offset, or a combination of both. As Victor said, it’s likely that only a pilot or someone familiar with ATC procedures would choose one of those options. If the hijacker didn’t care about conflicting traffic or was simply unaware of the procedures, then I’d say an offset or altitude change would be unlikely.
@DrB
You said: “At 18:25 the OXCO was very cold,..”
Two points:
1. It’s OCXO not OXCO.
2. The assumption that the OCXO at 18:25 was very cold can be wrong.
It is possible that the OCXO was warm, but the Left High Gain Antenna was inoperative. The 1825 log on occurs from the serviceable Right High Gain Antenna when it was finally exposed to the satellite, ie at the start of the turn from Nilam to Sanob when the aircraft rolled 25 degrees left for the programmed diversion to Banda Aceh. The first BFOs at both log ons are correct, but the second BFOs after the log ons are corrupt due to a damaged Left AIMS Cabinet.
If the APU was on prior to fuel exhaustion, the aircraft ends within a 100 NM radius to the south of the Bayesian Hotspot.
They only searched to 40 NM.
@DrB. “That is because the power outage was brief (~1 minute) and the OXCO did not cool off significantly.”
Bear in mind that that is a little uncertain. Had the right engine been powering the SDU, though adjudged unlikely, there could have been a more substantial gap.
@Mick Gilbert, @DennisW
@DennisW said:
“Ask yourself why the pilot would disable the left bus. It makes absolutely no sense in the grand scheme of things. Which is why I do not believe it happened. Why bring the bus back it 18:25:27? ”
To repeat: consider that the left bus and tie were not disabled to go dark, but to turn off something that couldn’t be disabled from the cockpit in any other way.
Later, something that had also been unintentionally disabled (ie. as a collateral effect) was needed for the next step and had to be turned back on, so the left bus and tie needed to be re-enabled.
For example, maybe ‘NO LAND 3’ (autoland) was needed? If so, that would suggest a possible landing within around 12nm? Were there any landing places within 12nm or so at 18.25?
http://support.precisionmanuals.com/kb/a40/how-do-i-perfom-an-autoland.aspx
Or… what else is (completely) disabled when the left bus (both gens) and tie is turned off?
It needs someone with a copy of the AAM to check. It’s surprising that, even so long after the event, no one has posted a complete list of what would have been disabled.
@Victor
“Your approach seems to be to find any possible way that the data sets are not related. My approach is to find the most likely explanation for the relationship between the data sets that were found.”
You’re partly right, Victor. I haven’t tried to find any possible way that the data sets are not related but I have examined whether the data might support conclusions other than that they are related. I have tried to maintain a neutral position (maybe not successfully), one that entertains the possibilities that the data sets may be related to one flight or to a number of different flights or there might be no relationship at all. I think that is the proper way to approach the evidence, it’s the way I was taught to test a hypothesis against the data – as a bare minimum you have to test the equivalent of the null hypothesis and preferably you also test the counter-hypothesis.
And when you look at the data from that perspective you see that it actually provides better support for the counter-hypothesis (viz that we are looking at data sets taken from multiple flights). And that makes statements to the effect that the data leaves “… little doubt that a user created a simulation in which a B777 is successively positioned in the Malacca Strait, the Andaman Sea, and the SIO” at the very least highly conjectural.
“Just the fact that there are three points (10N, 45S1, and 45S2) that share marker values and progress from the Andaman Islands and end with fuel exhaustion in the SIO is a strong suggestion that the simulated flight and the disappearance are related. It would be an incredible coincidence if they were not.”
I’d argue that we need to be very careful of confirmation bias and the old propter hoc trap here, particularly given the scarcity of contextual information. One piece of contextual information that we lack is how the Captain usually ended his flight simulations. We know from the limited data in his flight logs that for airplanes that he was familiar with (B737 and B777) more than half the time the simulation did not end with a landing. So, was his usual practice to simply end the simulation with the airplane mid-air in stable flight and just hit ESC or did he indulge himself in “unusual” endings? Did he sometimes just point the airplane towards empty airspace, select Time Compression and let ‘er rip? We simply don’t know. However, if we knew that running his flights to fuel exhaustion was not an uncommon way for him to end his simulations would that change how we view 10N – 45S1? Quite possibly. Just by the bye, if that all sounds quite fanciful, it was by exploring that possibility that I stumbled onto the 180 CMT connection between 10N and 45S1.
And yes, I agree that it would be an incredible coincidence for the 10N – 45S1-2 sim to be unrelated to the disappearance but it’s not like incredible coincidences don’t happen; MH370 seems to be a lightning rod for them.
For fear of waxing lyrical (or sounding like a tool), since I first became aware of the quote I have tried to be guided by Thomas Cromwell’s exhortation to the synod of the Church of Scotland back in 1650, “I beseech you, in the bowels of Christ, think it possible that you may be mistaken.” So, when I settle on some wonderful new idea, contention, hypothesis or opinion I ask myself, what if I am mistaken? how would I know? how would I prove that? what would it look like?
Any old how, thank you for your considered and courteous responses, Victor. As usual I’ve found the discourse to be useful and instructive. And thank you for maintaining this forum, I think that it might be easy for the rest of us to underestimate how much work goes into it.
The “electronic frequency device” on MH128 is really frightening in the context of MH370.
At the same time another incident:
CNN: “Flight diverted when lithium battery catches fire onboard”
http://cnn.it/2soS0LH
… with the amount of lithium batteries on MH370 being several orders of magnitude higher.
@Victor
@Andrew
@DrB
Re: “As an aside, I wonder if a pilot who intentionally hijacked the aircraft would perform such a lateral offset.”
Just setting aside malicious intent for a moment it is worth noting that flying a 15 nm offset (and descending 500 feet) is part of MAS’s lost communications procedure.
https://www.dropbox.com/s/ye1kt400yewswrj/MAS%20Lost%20Comms%20Procedure.pdf?dl=0
DrB: Re: May 31, 2017 at 4:38 pm post. I agree. At 1825 you have a cold start (maybe +25C, or maybe colder if decompression took place) resulting in some OCXO overshoot, and some maneuvering going on. We know the later is true from radar, BTO and BFO data. At 0019 we know there was a warm OCXO start (probably within 2-3C of the setpoint temp) with little BFO transient error and a large vertical velocity signal.
@Mick Gilbert:
“And yes, I agree that it would be an incredible coincidence for the 10N – 45S1-2 sim to be unrelated to the disappearance…”
We agree here.
“but it’s not like incredible coincidences don’t happen; MH370 seems to be a lightning rod for them.”
Here’s we have fundamental disagreement. My goal is always to find explanations for the coincidences, including the possibility of human intervention. To me, when an improbable event occurs in a hypothetical scenario, it means there is likely a flaw in the scenario. The more improbable events that are required, the less likely the scenario.
> Victor Iannello:
> My goal is always to find explanations for the coincidences, including
> the possibility of human intervention. To me, when an improbable event
> occurs in a hypothetical scenario, it means there is likely a flaw in
> the scenario. The more improbable events that are required, the less
> likely the scenario.
I agree with your statement,
but how do you want to catch highly improbable events then ?
Accidents very often occur as a result of a chain of highly improbable events (because commercial aviation is so secure).
@David,
You said: “Bear in mind that that is a little uncertain. Had the right engine been powering the SDU, though adjudged unlikely, there could have been a more substantial gap.”
Based on the ACARS fuel reports early in the flight, there was less fuel available to the right engine after 17:07 and the right engine had a higher fuel flow. Thus it seems highly likely that the right engine ran out of fuel first. As you say, we don’t know for certain that the SDU was powered by the left engine at 00:17 or perhaps a few minutes earlier. If the SDU were powered by the right engine then, what would be the sequence of events leading to restoration of power to the SDU after the right engine flamed out? Would SDU power be supplied by the left engine IDG or by the APU or the RAT ? What would be the time interval without power?
@Paul Onions,
Yes, it is OCXO.
I don’t know any means by which using a damaged antenna or a different antenna can cause a shift in the BFO. Your theory does not make any sense to me.
If by “Bayesian hotspot” you mean 38S, it was impossible for MH370 to reach that location. It did not have sufficient fuel.
@Victor
@Peter Norton
“Here’s we have fundamental disagreement. My goal is always to find explanations for the coincidences, including the possibility of human intervention. To me, when an improbable event occurs in a hypothetical scenario, it means there is likely a flaw in the scenario. The more improbable events that are required, the less likely the scenario.”
Victor, I understand that philosophy but I think that it is misplaced when it comes to dealing with events such as MH370. As Peter quite correctly alludes to, the history of aviation safety is replete with examples of multiple occurrences of highly improbable events giving rise to usually unforeseen outcomes that often result in the loss of an airplane. You need look no further than the chain of relatively uncomplicated but utterly improbable events that conspired to bring down AF4590. Such is the nature of aviation accidents, they are exercises in improbability.
One of the problems with eschewing improbable events is that the default remedy tends to be that all purpose, all encompassing, no-more-gaps solution, human intervention. In this case it’s malicious human intervention via malus homo. It’s the equivalent of invoking the boogey man. Moreover, it often ignores the fundamental fact that the very human interventions that are plugged in to solve the improbability problem, the individual acts and sequence of acts, are also highly improbable. However, we salve that improbability away because we see the human actor, like an artist’s mannequin, as almost infinitely flexible. The B737 rudder hard-over story is perhaps the classic example of where human intervention, in that case flight crew mishandling, was substituted (repeatedly) for the utterly improbable notion that under certain circumstances a mechanical device would perform in a manner exactly opposite to the way it was designed to perform.
DrB
You have stated that the aircraft didn’t have enough fuel to reach S38. With all due respect, can you be absolutely sure of this? Has this conclusion been corroborated by Boeing/ATSB? It is extremely important that this issue is clarified once and for all, to everyone’s satisfaction, because if it isn’t, there is a risk a subsequent search will draw another blank. There are others on this site who will now chime in and say that S38 has already been searched, conveniently ignoring the fact that the ATSB carried out the search assuming that no one was at the controls during the final descent, an assumption that may well turn out to be unsafe. The aircraft may have been glided a further 100Nm after the LH engine flamed out, indeed the nature and condition of the recovered debris tends to support a controlled impact, in my humble opinion.
@Dr B. “If the SDU were powered by the right engine then, what would be the sequence of events leading to restoration of power to the SDU after the right engine flamed out?”
Based on the Boeing fuel flows the right engine might have failed up to 15 mins before the left according to the ATSB. If the left IDG had earlier come off/been taken off line, the right IDG would then power the SDU through bus ties. On the right’s fuel exhaustion and consequent SDU shut down, that would be rebooted when the left engine finally suffered f.e., after that prompted APU autostart. That would result from AC power failure to both transfer buses (ie left engine B/U generator failure in this case).
If the left B/U gen had been were switched off with its IDG then the APU would autostart at right engine f.e. Continuing A/P engagement after right engine f.e. would also depend on the left engine’s B/U gen being on line.
@Mick
Back in January I posted the regulatory source for that MAS procedure you posted above.
http://jeffwise.net/2017/01/18/were-mh370-searchers-unlucky-or-duped/comment-page-13/#comment-202780
(..so you can guess why this comment of Victors made me chuckle;
“It wasn’t until Andrew told us that an offset of 15 NM is recommended if
there is no traffic control clearance that most of us were aware of this.” )
@Mick Gilbert: Of course improbable events can occur, especially in the context of an accident. However, the more improbable events that are required in a scenario, the lower on the list of possible scenarios. In my opinion, any MH370 scenario that doesn’t include intentional diversion requires so many improbable events to line up that it becomes much, much less likely than intentional diversion.
@buyerninety: OK. I acknowledge that you posted the information before Andrew. It’s still only recently that people (primarily DrB) began reconstructing paths that use the 15-NM offset to satisfy the BTO and BFO. When I first proposed back in April 2015 that a lateral offset manoeuver could match the satellite data, I was using an offset of 12 NM, not 15 NM.
@DrB
Did you start with the last ACARS fuel load or with the underestimated fuel load at Igari?
If the FMC was programmed Nilam-Sanob-Banda Aceh at Econ Crz/FL350, the aircraft can reach S38. That’s why they searched there.
If the APU was on prior to fuel exhaustion, the aircraft does something unexpected when the left engine fails.
The transfer of power to the Left Main AC bus from the failing Left IDG to APU causes a significant power interruption to the Satcom when the left engine spools down: Satcom depowered messages briefly appear on EICAS. (Yes, normal switching does not cause a power interruption, but fuel exhaustion does). Load shedding of IFE occurs when only on APU.
Since the APU was on, the autopilot remains engaged. The aircraft won’t go into its “death dive” until the APU has failed. The aircraft can be between 40-100 nautical miles south of the Bayesian hotspot.
> Mick Gilbert:
> As Peter quite correctly alludes to, the history of aviation safety is
> replete with examples of multiple occurrences of highly improbable
> events giving rise to usually unforeseen outcomes that often result in
> the loss of an airplane. You need look no further than the chain of
> relatively uncomplicated but utterly improbable events that conspired
> to bring down AF4590. Such is the nature of aviation accidents, they
> are exercises in improbability.
That’s precisely what I tried to express. By way of learning from past mistakes, commercial aviation has become so safe over the decades, that a high percentage of accidents (which are in itself incredibly rare and improbable events) are caused by a chain of highly improbable events … since the likely causes of accident are already taken care of – for the most part with multiple redundancy, so that you almost need a combination of very unlikely events to be able to break through the multiple security layers.
@Victor
“In my opinion, any MH370 scenario that doesn’t include intentional diversion requires so many improbable events to line up that it becomes much, much less likely than intentional diversion.”
I agree, where we differ, I think, is that I believe the initial diversion was a legitimate, professional response to an evolving inflight emergency rather than a deliberate malicious act.
@Paul Onions said, “The transfer of power to the Left Main AC bus from the failing Left IDG to APU causes a significant power interruption to the Satcom when the left engine spools down: Satcom depowered messages briefly appear on EICAS. (Yes, normal switching does not cause a power interruption, but fuel exhaustion does).”
If you are using the PMDG 777 model as a tool to study this, I would advise that this is exactly the kind of transient, i.e., a sequence of events that depends on sub-second timing, that the PMDG 777 model would NOT model accurately. The SATCOM is designed with sufficient energy storage to ride-through a transient caused by a switch from IDG to APU power. The only way that I am aware that this switch would cause a re-boot of the SDU is if there were a malfunction. Do you have any documentation from Boeing to support your claim, or are you relying on the PMDG 777 simulation?
Comments on the 18:25:27 BFO.
http://tmex1.blogspot.com/2017/06/and-so-it-goes-182527-bfo-value.html
@Dr B.
“If by “Bayesian hotspot” you mean 38S, it was impossible for MH370 to reach that location. It did not have sufficient fuel.”
I believe that this statement is demonstrably false.
@DennisW: Considering all the time and money that has been spent, it would have been helpful to conduct some experiments with inflight log-ons under various conditions (e.g., cold and hot power-ups) to try to recreate the BFO sequence at 18:25 – 18:28. There is a disagreement about how to interpret the BFO sequence only because the proper experiments have not been conducted.
@Victor Iannello:
Please find my thoughts on the GForce ‘marker values’ in the attached draft post.
@Victor Iannello
You said: “Do you have any documentation from Boeing to support your claim, or are you relying on the PMDG 777 simulation?”
No we don’t have any Boeing documentation. They don’t tend to write about events when the aircraft runs out of fuel. A group of us have experimented in actual simulators. Yes it is a simulator and the validity could also be questioned, but it is the best option since we couldn’t try this on an actual aircraft with passengers onboard.
We think this is a viable option, to search the 40 – 100 nautical miles past the Bayesian hotspot.
We have always questioned their end of flight scenario, ie the aircraft is out of control for two minutes, then at two minutes and eight seconds it is in a massive descent with an extreme attitude. And at this attitude, the APU is amazingly still being fed fuel!
We know the satellite gurus will say, “but the last BFO suggests massive descent”, but that area has been extensively searched! They are basing their results on a system that was never designed to be used that way and is not equipped with integrity monitoring. In other words, can you fully trust that the last BFO is correct? Maybe it’s time to say NO and search 40 – 100 nautical miles past the Bayesian hotspot.
@Victor @all
I have competed a comment draft report on a Proposal for a 180S “Straight” South Flight Path:
https://docs.google.com/document/d/149krnQCSITLHDS1KPnOr__pZpiaqRL0mq0Ppc7m9oWs/edit?usp=sharing
If one assumes a 180S flight path, a 180S CTH path from just above ISBIX seems to fit the BTO/BFO data from Arc2 to Arc5. Shortly after 22:41 UT, twilight begins and this apparently coincides with a controlled descent at 300 ft/sec, possibly to manage the visual profile of the aircraft in the approaching morning light.
The end point is S33.7, although I have yet not worked out what happens between Arc6 and Arc7. The descent after 22:41 (twilight) is a new idea and I am seeking comments on the feasibility of that idea.
@Paul Onions: We DO have documentation from Honeywell saying the SATCOM should ride through power interruption lasting 200 ms. That’s an eternity compared to switching times for a bus transfer. The PMDG 777 model should not be trusted for sub-second transients. You need a Level D simulator to model a sequence of events occurring that fast.
Do you really think that Boeing would design a system whereby an emergency transfer to available APU power would cause electronic components to reboot?
@All
Interesting new article in the Australian, from Ean Higgings (I know, I can hear you saying “oh really?, Ean Higgings?, but suspend your disbelief. Apparently, Malcolm Turnbull has spoken to Najib Razak concerning the prospects of resuming the search.
Extremely important then, that all possible scenarios are considered by the parties involved, including the extended, manually controlled glide downrange of the Bayesian hotspot, and not just concentrating on the 25 thousand sq km area. It could a mistake to be overly influenced by drift studies.
@TBill: Thanks for putting that together. The 180° CTH path is interesting.
@ROB
cut-paste from Higgins//
The ATSB recently said a new “drift modelling” study by the CSIRO charting the discovery of debris from the plane found on and off the coast of Africa further supported the evidence that the aircraft lies in the proposed new target zone.
end cut-paste//
“further supported the evidence”
What evidence is being referred to? I am unaware of any evidence pointing to the new target zone.
Malaysia is very unlikely to authorize resumption of the search, IMO.
@TBill
Thanks for posting your paper on the 180-ISBIX flight path!
I have also updated my earlier paper to include the latest CSIRO drift study – everyone of course warmly welcome to comment, including criticism and the like.
https://www.docdroid.net/49FJCLb/mh370-waypoint-21.pdf.html
The way I approached this was exactly the other way round, i.e. from tail to head. I was assuming a straight path between the ultimate and penultimate arc, which leaves few options to fit the bill, and to go from there.
I agree that the 19:41 BFO can as well be explained as the result of a maneuver (a climb after leaving radar line of sight).
What is the average speed that you require for the segment between 22:41 and 0:11? Does it match a steady descent and how would that influence fuel considerations?
@Victor
Thanks for running the blog!
@Nederland: Thank you for joining the discussion!
@DennisW
Agree, I didn’t see any persuasive new evidence iro drift studies either.
There are two articles by Higgins. They give the flavour of increased enthusiasm to reopen the search.
1.”Malcolm Turnbull in tune with MH370 relatives’ requests to do more”
The Australian12:00AM June 2, 2017
ReporterSydney
@EanHiggins
Malcolm Turnbull has revealed he has taken up the cause of the Australian families of Malaysia Airlines flight MH370 victims with his Malaysian counterpart, making the disclosure following his receipt of a letter this week from a Queensland woman whose husband disappeared with the aircraft three years ago.
Danica Weeks, who was left a single mother looking after two young boys when her husband Paul was lost, wrote to the Prime Minister on Wednesday calling on him to press Malaysian Prime Minister Najib Razak to restart the search.
Mr Turnbull told The Australian he had already made representations in the interest of the families of the six Australians on board MH370, with his spokesman saying “the Prime Minister raises this issue with his Malaysian counterpart every time they speak”.
“Malaysia is the lead nation in the investigation into the disappearance of MH370, but Australia stands ready to assist in any way it can,” the spokesman said. “At present, the search for MH370 has been suspended, but if any credible evidence emerges, the Australian government will do everything it can in partnership with Malaysia to ensure the search is resumed.”
MH370 doubled back on a scheduled route from Kuala Lumpur to Beijing 40 minutes into the flight, with its radar transponder turned off and radio contact terminated. Satellite tracking data showed it ended up somewhere along a band in the southern Indian Ocean.
At the request of Malaysia, the Australian Transport Safety Bureau ran what proved to be a fruitless $200m underwater search for the aircraft which ended in January.
While the ATSB has identified a new potential search area to the north of the last one, which it says is “highly likely” to contain the aircraft, the three governments involved — Malaysia, China and Australia — have insisted no new hunt will be undertaken without evidence indicating the specific location of the aircraft.
ATSB officials are known to be keen to resume the search.
In her letter to Mr Turnbull, Ms Weeks wrote of what she called “the daily torment Malaysia Airlines and ultimately the Malaysian government have imposed upon the families … in its dealing with this situation”.
“Time is no healer for us, that is a luxury only afforded to those who get a proper goodbye and who know what happened to their loved ones,” she wrote.
“I find it extremely distressing the Malaysian government’s inclination to cease searching and the perception they wish to ‘brush MH370 under the carpet’.”
Ms Weeks also expressed her dismay at the refusal of ATSB chief commissioner Greg Hood to release key material regarding the search for the aircraft.
Mr Hood has supported ATSB general manager for strategic capability Colin McNamara’s decision to knock back a freedom of information request from The Australian for assessments of the satellite data by experts, which the bureau claims supports what has become known as its “death dive” theory that the Boeing 777 went down in a rapid, unpiloted crash with 239 passengers and crew on board.
Mr McNamara said in his rejection of the FOI request that to release the material could “cause damage to the international relations of the commonwealth.”
In her letter to Mr Turnbull, Ms Weeks claimed the ATSB “is unwilling to be open about the basis of its findings.” Mr Turnbull’s spokesman said the FOI request was a matter for the ATSB.
The developments come as a Queensland barrister acting pro bono for some of the MH370 families, Greg Williams, is preparing an online survey to gauge their “satisfaction or dissatisfaction” with the performance of the ATSB, the Australian government, and Malaysian authorities.
2.”Turnbull offers Malaysia help in new search for MH370″
The Australian12:00AM June 2, 2017
@EanHiggins
Malcolm Turnbull has discussed with Malaysian Prime Minister Najib Razak the circumstances in which the hunt for Malaysia Airlines flight MH370 could be resumed, and said Australia stands ready to do “everything it can” to see it happen.
The move comes as the Australian Transport Safety Bureau is preparing to release a report on its unsuccessful search for the aircraft, which may also present new evidence for its claim that it is “highly likely” the Boeing 777 lies in a proposed new search zone to the north.
The Prime Minister’s revelation that he has repeatedly raised the issue with his Malaysian counterpart also follows renewed activism among relatives of the Australian, Chinese and other victims who disappeared with the plane when it went down more than three years ago.
Mr Turnbull saidhe had made representations in the interest of the families of the six Australians on board, with his spokesman saying “the Prime Minister raises this issue with his Malaysian counterpart every time they speak”.
“Malaysia is the lead nation in the investigation into the disappearance of MH370, but Australia stands ready to assist in any way it can,” the spokesman said.
“At present, the search for MH370 has been suspended, but if any credible evidence emerges, the Australian government will do everything it can in partnership with Malaysia to ensure the search is resumed.”
A government source said Mr Turnbull was not pressuring the Malaysian government to restart the search, describing it more as a means of “keeping MH370 as a frontline issue”.
On March 8, 2014, with 239 passengers and crew, MH370 doubled back on a scheduled route from Kuala Lumpur to Beijing 40 minutes into the flight, with its radar transponder turned off and radio contact terminated.
Satellite tracking data showed that it ended up somewhere along a band in the southern Indian Ocean.
About the same time the ATSB’s underwater search of a 120,000sq km zone ended in January, it held a conference of international experts, which identified a new potential search area of 25,000sq km.
The three governments involved in the subsea search which cost $200 million — Malaysia, China and Australia — have taken the joint position that no new search will be undertaken without new evidence indicating the specific location of the plane.
But it is thought that Malaysia is the least enthusiastic to resume the hunt, while ATSB officials are known to be keen to do so and believe they have a strong case.
The ATSB recently said a new “drift modelling” study by the CSIRO charting the discovery of debris from the plane found on and off the coast of Africa further supported the evidence that the aircraft lies in the proposed new target zone.
Most of the passengers on MH370 were Chinese nationals. Yesterday the association representing their families issued a statement saying a letter of appeal signed by 1000 members had been sent to the Chinese Ministry of Foreign Affairs, reading, in part: “Concerning the parties responsible for the unknown fate of MH370, Malaysian government, Malaysia Airlines, aircraft manufacturer Boeing, the parties must honour and fulfil the promise of finding the plane, passengers and crew, without interruption, ceasing or abandonment.”
The Higgins articles. In the current edition the first, which iterates FOI stuff, seems to have supplanted the second. Therein may lie another tale.
@TBill and others
I respect the work you and others are putting into path modeling. I have done a lot of it myself. However, I am deeply troubled by two issues that I believe are not getting the attention they deserve. I had this conversation with Victor a couple of days ago.
1> Why the flight to Beijing?
To me this question is extremely important. If Shah wanted to dump the plane at 33S or beyond he would logically pick a flight to Jeddah or some place like Amsterdam. He could make the FMT virtually undetected and not worry about primary radar or interception. Instead Shah picked a fight where the diversion would be known early, and he was likely to be detected on primary radar. 33S-40S along with the selected flight to Beijing makes no sense. I believe he picked that flight to make the diversion known as early as possible.
2> The Malay response.
It was horrible. It was beyond casual by any S&R standards. Even ICAO was critical of it which is unusual for ICAO. It indicates to me that the Malays high up knew exactly what was going on (not where the aircraft was necessarily), and acted appropriately. I believe they fully expected Shah to land the aircraft somewhere, and were anticipating dealing with that situation after the landing occurred. When a landing did not occur, it was OMG time.
The reason for the diversion is not known by us, but it is known to several people in a position of authority in Malaysia and was known long before the aircraft ran out of fuel. I believe there was a loiter over the Andaman Sea and I believe Shah maintained a reasonable striking distance to an airfield for as long as possible. Whatever he and his colleagues in KL had in mind did not work out. I believe the aircraft is on the 7th arc near an airfield. Take your pick, but it is not in the latitude of 33S-40S or beyond.
To me there is no way to reconcile the priority search area, beyond the priority search area, or modestly North of the priority search area with the flight selected or the Malay response. Simple as that. Flame on.
Nederland:
Thanks for working up your new path scenario. It is an interesting approach. One minor suggestion. The 19:41 arc is the second arc, not the first as it is described. The cluster at 18:25-18:28 is commonly referred to as the first arc. You might want to clarify terms. Otherwise, an interesting new way to explain the data.
@Nederland: I haven’t checked your math, but BEBIM to McMurdo is an interesting possibility that I had not considered. In GE, I see a crossing of the 7th arc closer to 29.8S latitude for this great circle.
@Nederland
Looks like I need about 415 nm/hr avg ground speed from Arc5 to Arc6.
Not sure on fuel, but was assuming descent can be more fuel efficient.
@DennisW
Beijing:
Don’t forget one good answer might be 2 crew in cockpit vs. 3 crew for longer duration flight to Jeddah. Other than that it is frustrating as we do not have access to MAS staffing policy and their account of how/why Z got assigned to the MH370 flight.
Malay Response:
I am in complete “agreement” with your speculation that perhaps they knew Z was up to something that night. But I fail to see why we are thus forced to assume an extended Andaman’s loiter that precludes 32-34 South.
In mathematical terms, I currently believe the BFO data is relatively accurate at least from Arc2 thru Arc6. I see no need to discount BFO – explain it yes – but it must be matched by choice of flight conditions and course changes. Therefore the Andamans loiter/McMurdo paths fell out of favor for me, at least temporarily.
@TBill
Carry on. My rant is based on three years of frustration. Just putting it in the category of background info. Of course, I don’t have a pin in the map at 38S either.
My conviction is that a search of the new “enlightened area” will come up with the same result we have to date for the same reasons.
@DennisW
Dennis, oh Dennis. When I read your recent post re. “why divert a Bejing flight, etc”, I rolled my eyes! I’ll get back to you later today with my understanding of it.
@ALSM
Thanks for spotting that, I shall correct this in the next version. I think the graphic on p. 10 is correct in this regard.
@Victor
I put quite a bit of work in that. I don’t have the means to do fuel calculations or wind modelling, but otherwise I have discussed that for quite a while now both on the Jeff Wise Forum and a German language forum, so I believe the math should be correct. Happy if anyone wants to doublecheck again or suggest a variant to the proposed path.
@DennisW
I am broadly in agreement with your thoughts.
Why Beijing rather than Jeddah?
I think the purpose of the early flight segment could have been to demonstrate the weakness of Malaysian military response. Obviously, that would not have worked on a flight to Jeddah. This also means that MH370 was then keen to avoid being spotted by Indonesian radar within Indonesian FIR, as suggested. I think that, regardless of the exact route up to this point, a relatively late arrival at ISBIX opens up a number of viable flight paths, such as TBill’s and mine, ending up further north on the arc and agreeing with drift studies. Basically, TBill and myself have explored various possibilities in line with this assumption. Perhaps this is something worthy of further exploration?
Malaysian response
Given the above consideration, I think it is possible that someone was spreading erroneous information about the flight on the night (why was MH370 supposed to be in Cambodia, why has the alleged conversation between ATC and military never been made public and so on?). This may imply that MH370 was supposed to land somewhere. But I couldn’t find a viable route via Cocos Island that works along waypoints. BEBIM, on the other hand, is not only the one viable waypoint in that area, but it also serves as a gateway to various destinations: Cocos and Christmas Islands, Java, probably even Learmounth (a rough calculation suggested the fuel would have been sufficient), as explained in the document. I was quite struck when I worked out that the only possible trajectory from there (assuming a straight flight path) happenend to be the one to McMurdo.
I also don’t agree with the ATSB’s assumption that this area has been combed sufficiently during the surface search. This proposed crash area is similiar to the one identified in the June/August 2014 ATSB report (note that they also assume a ‘northern hook around Sumatra’). But it was not before 28 March 2014 that the surface search shifted to this area and the area was already abandoned when alleged blackbox pings were detected on 5 April.
@Nederland: When I connect BEBIM with NOBEY in Google Earth, the crossing of the 7th arc is 29.8S at 35K ft and 29.7S at 0 ft. I’m not sure what you are using for ping arcs, but I would check this.
@Victor
As explained in the document (p. 3 and 11), I’m using the heading BEBIM – NOBEY as indicated by skyvector. This is a very long segment, and I couldn’t say in which way a given autopilot mode would affect the heading, so I took the skyvector heading of 171 as the safest option. I’m using Google Earth for the other segments as these are far shorter (and very similar to the results on skyvector). Yes, a direct connection on Google Earth is slightly off track, as far as I can say this should still work with the BFOs though.
Just checked, and I get 29.7S when drawing a straight line from BEBIM to NOBEY.
@Nederland: “This is a very long segment, and I couldn’t say in which way a given autopilot mode would affect the heading, so I took the skyvector heading of 171 as the safest option.”
If you assume the plane was navigating by waypoints, then LNAV is the roll mode, and the plane would be following great circles (geodesics) between waypoints, just as GE shows. Except for paths along constant longitude or along the equator, the track will vary along the path.
@Victor, OK, thanks for that. At a quick glance, just going by distances a heading of 170 or 171 seems to work best, at a heading of ~168 (which I get from google earth), the segment between arc 5 and 6 seems to be slightly longer than 1.5 times the distance from the one between arc 4 and 5 (not much). I’ll have another look into that and the BFOs (although a heading of 170 or 171 remains a possibility).
@Paul Onions
On your statement: ” Load shedding of IFE occurs when only on APU.”
I think this could be interesting for it’s common believe by respected experts the 00:21 IFE log-on did not take place because the plane crashed before this time.
If you are right the IFE could not log-on while the plane was only on the APU (the RAT doesn’t supply power to the SDU anyway).
Can you please supply official information on this?
@TBill
Thanks for your extensive paper. What an amount of work..
Have to read in detail still but one sentence ringed to me already..;
‘which leaves few options to fit the ‘bill’..’ I had a laugh, guess why..;-)
Personnaly, when trying to creep into a pilot’s mind who deliberately wanted to devert the plane with the goal of letting the plane vanish as good as possible, I think he would have chosen to keep things as simple as possible after he was out of possible harms way.
Meaning imo that (just) before FMT he just entered his desired end-coördinates and fly a route after that straight to his goal.
Maybe with a 1 degree descent or so, maybe not.
There would have been no need imo.
He would have had control of the fuel consumption anyway. Jettison fuel if desired (for minimizing fire/explosion and fuel traces on the surface after ‘ditching’).
I see no reason why he should have to plan a route along waypoints after FMT if one set of desired coördinates would be sufficient.
@DennisW
“My conviction is that a search of the new “enlightened area” will come up with the same result we have to date for the same reasons.”
Except they never searched the “correct” area in the first place.
Probably they should revisit Bayesian analysis first, but I feel like if they could kindly avoid putting a their preference for simple FMT ghost flight into the model, they may find 32-35S is good area.
The lack of visual debris above about 31S may be a valid limiting assumption. I am thinking Bayesean technique allows accounting for fuzzy evidence, also drift analysis.
I would make the model fit BTO/BFO and allow pilot inputs to about 19:45 or so, and after 22:41 or so.
I feel a straight flight path would not have an exactly straight BFO curve, so they need to calculate what some hypothetical straight paths would look like, then see how actual BFO/BTO deviates from the straight path ideal, and then look for proposed flight paths that match the observed deviations/inflections (with wind etc).
@DennisW
You’ve been agonizing for three years, when there was really no need.
Firstly, if Shah had expected to be tracked on primary (military) radar, why did he bother to switch off the transponders, and disable ACARS as soon as he began to divert? The answer is he didn’t expect to be tracked by primary radar. He figured the radars would be turned off after midnight (if they were ever going to be at all) and the operators would be otherwise engaged. He knew the area and the operating environment well enough, or so he thought – he knew things were lax enough for him to take a chance on not being spotted.
But as an additional protection against being spotted and tracked on military radar, he wanted it to look as if the aircraft had been blown out of the sky at 17:20. This would require the SATCOM to be disabled, and the only way that can be done from the cockpit is to isolate the LH AC Bus. Yes he could disable the IFE by simply switching off IFE power at the overhead panel but that wouldn’t prevent the SDU acknowledging the “inevitable” MAS telephone call(s) as they tried to contact the crew, letting the the ground know that the aircraft was at least still airborne – because if MAS knew he was still airborne, there was a risk they might request the military to search for him before he had got beyond range. At 18:24 he judged it safe enough to reconnect the LH AC Bus, and thus reenergize the SDU – as it turned out, a mere 2 minutes after the final radar contact, so he had obviously done his homework here.
The MH370 Beijing flight was chosen for several reasons:
1) It was a redeye flight with a relatively inexperienced First Officer, fresh out of mentoring.
2) The flight took off just hours after Anwar’s conviction had been upheld, something all observers (Shah included) had thought a forgone conclusion.
3) Most of the passengers were Chinese. Shah as an active supporter of the fundamental Islamist movement would have been angered by the way the Chinese authorities were suppressing fundamentalists in Xinjiang Province, and thought this would be a way of striking a blow for Islamic Fundamentalism. The fact that 238 innocent people would loose their lives, meant nothing to him. Just as killing and maiming innocent children enjoying a night out in Manchester meant nothing to the suicide bomber. The Malaysian authorities have essentially sat on the MH370 affair, and the Chinese authorities themselves don’t appear to have been too concerned about getting to the bottom of it either. No surprise really, is it?
Dennis, stick a pin in S38.
Dennis, stick a pin in ROB.
..”Shah, as an active supporter of the fundamental Islamist movement”..
Fake News.
@ROB
I like the story up to, and you should add to the story, sure enough 10 minutes after he powered up he got the expected sat call from MAS.
My understanding Z was not very religious and why the heck go to 38S? You need to frame out the flight plan logic of the flight path proposed.
@TBill
No problem. By aiming along a great circle path with an initial heading of 186deg (performed by extending the IGOGU, ISBIX great circle) he arrived at fuel exhaustion with the Sun about 4deg above the eastern horizon, to enable a controlled ditch in a desolate area of ocean, which turned to be at S38. The clincher for me was that the chosen flight path into the SIO ran parallel to the advancing morning terminator, ideal for matching the time of fuel exhaustion with the preferred sun angle. As long as the fuel ran out at about 00:17 (give or take a few minutes) which of course it would with this particular flight’s fuel load, he would have the right lighting conditions. It just happened that burnout was at about S37. The post FMT flight path was matched to the MH370 fuel load, no less. Pretty clever, if you ask me.
@buyer90
Yeah, Rob has been in love with 40S for a long time now.
@ROB
Your contention that Captain Shah didn’t expect to be tracked by militaryradar aside, what about civilian radar? Any professional pilot from those parts would understand that the radars that provide their feeds to the local area air traffic control operate 24/7. The locations of the radar heads and their coverage is published in the Malaysian Aeronautical Information Publication Enroute Supplement. One glance at the radar foverage chart for the Kuala Lumpur FIR shows that a track from IGARI back past Kota Bharu to Penang is one of the most surveilled sections of Malaysian airspace.
One of those civilian radars is at Kota Bharu and we know that the airplane flew straight back towards it – that’s an odd approach to evading radar detection. The other radar that any professional pilot from those parts would expect to be operating was approach control at Penang (provided by a radar head at RMAF Butterworth). Once again, the airplane flew pretty much straight towards it and then executed a banked turn that brought it into full profile, maximising its radar cross-section, essentially ensuring that it would be tracked. Again, that’s an odd approach to evading radar detection.
If a notional perpetrator intended to seize a flight and execute a complete disappearance free from radar detection then they would not have selected MH370. If for some reason said notional perpetrator was constrained to MH370 then they would have chosen a different track back across the Malay Peninsula.
@Mick Gilbert
As I, (and I think DennisW) have long contended, he “wanted”, indeed “needed” the Malaysian Authorities to “know” that he had taken the aircraft “for certain”.
He did that by deliberately “going dark from CIVIL ATC”, full well knowing that they would then “raise the alarm” and ask the Military for help (which they may or may not have done in “real time”.
In any case, he knew that even if ATC acted “quickly”, that the RMAF could not “react in time” to intercept.
If ATC or the RMAF was “slow off the mark” (which he was counting on) he would be long gone, BUT, and this is the crucial bit, he knew that they would later “check the tapes” and he would know that the Authorities would know he had gone back past Penang.
That was what he wanted and needed them to know.
BUT – where did he go “post Penang” ?
He had a few requirements.
(1) Make it hard for the RMAF to track him “post Penang”, which meant staying out of Butterworth PSR’s “view”, which can only be done by “hiding” behind Penang Island, which means continue south west at low level after a faked Kendi approach to Penang.
(2) Remain “hidden” from TNI-AU PSR’s that operate 24/7.
Going up the Malacca Strait into the full view of PSR TNI-AU Satuan Radar 233 Sabang Ache was insane.
The obvious choice, the obvious “escape rout” was to fly directly over Sumartra, through the TNI-AU “radar gap”.
See:
https://www.dropbox.com/s/01a406ypvot4p9o/Objective-Escape-Hide-from_indo-Military-PSR.jpg?dl=0
@ventus45
There is no gap in the Indonesian military radar coverage along the north coast of north Sumatra. Your “escape route” fails to show the Indonesian military radar at Soewondo, Medan – your selected flight path takes the airplane straight over the top of it – and Radar Unit 231 at Lhokseumawe – your flight path ploughs straight through its coverage. Again, that’s an odd approach to evading radar detection.
@Mick Gilbert
Do you have lat/lon and elevation for those other radars ? If so I will plot their coverage?
@Nederland
I’d be curious how your path compares on the 23:14 sat call BFO = 217?
I am thinking we have to set the date/time correctly in Column B of the CCYap spreadsheet. Check me on that calc technique. I am wondering if the only way to match well on all three BFO at 2241 2314 0011 is a descent after 2241.
Ventus45:
The radar coverage has been extensively researched and plotted by several people, including Duncan Steel, Don Thompson and Victor Iannello. For example:
1. http://www.duncansteel.com/archives/1987
2. http://www.duncansteel.com/archives/1969
3. https://goo.gl/LakWZq (locations and altitudes)
Ventus45:
Here is one more reference on radar
https://goo.gl/WkxLjW
@Ventus
The difficulty with radar with the transponder turned off is identification. There is a lot of stuff out there. How can you assign it to a particular target?
@DennisW,
In the note you posted you said:”The IG/DrB do not discard the first logged value, but insist it is simply part of the overshoot behavior associated with the OCXO warmup after turn-on, and it is serendipitous that it happens to agree perfectly with the expected BFO value. The IG/DrB are also apparently not disturbed by the fact that none of the other logon events presented by Holland exhibit any sign of overshoot behavior.”
First, you mischaracterize what I and ASLM said. We have never said it was “serendipitous” that the first BFO at 18:25:27 happens to have a fairly small error compared to the error in the 18:25:34 BFO. Quite to the contrary, we said it was virtually guaranteed to happen that way when the SDU was powered up with a cold OCXO. By the way, it is not in perfect agreement with the expected value, either, when you look at all the BTOs and BFOs circa 18:22-18:28.
I explained in my paper the reason why none of Holland’s other log-on transient events show an initial small error and then show overshoot. That is because this is the only in-flight log-on event where the power was off for an extended period of time.
Your last statement that “none of the other logon events presented by Holland exhibit any sign of overshoot behavior” is ludicrous. They all show overshoot behavior.
You should stick to your day job.
@Paul Onions.
Why does your proposed turn for Banda Aceh transit NILAM and SANOB, neither of which are part of the standard approach?
@DrB: The OCXO temperature controller overshoot is certainly a compelling theory. But without actual experimental confirmation it remains a theory, isn’t it?
@TBill
The version of the BFO calculator I am using (v. 4) does not include the BFO for the 23:14 sat call, so yes, it would be interesting to work that out. What information do you need? The heading is 171, ground speed 425 kts, do you need approximate geolocation (extrapolated)?
If I follow Victor’s advise to go for direct great circle connection between BEBIM and NOBEY, the heading is ~168. The distance between arc 4 and 5 is ~400 nm, the distance between arc 5 and 6 ~615 nm. That means the ground speed for the final segment is higher by ~10 kts on average. Is there anything to explain this other than deliberate change in speed? How about engine flameout?
On the other hand, how likely is it that NOBEY was part of the MH370 waypoint database? If it wasn’t (an unlikely landing site), and someone was plotting a route from BEBIM to NOBEY/McMurdo, he would probably have gone for the 171 heading indicated by skyvector, which seems to be the best fit in that scenario.
Also perhaps worth noting (as I’ve just seen while googling keywords) that someone came up with the possibility of a route BEBIM to NOBEY/YWKS already in 2014:
http://www.pprune.org/rumours-news/535538-malaysian-airlines-mh370-contact-lost-562.html#post8566595
http://www.pprune.org/rumours-news/535538-malaysian-airlines-mh370-contact-lost-563.html#post8566603
@Nederland
At the moment as a contrarian I am saying Z was not using NOBEY or NZPG, he may have used 180S Magnetic. You cannot work backwards from NZPG because yes there was probably a slow down/descent between Arc5 and 6. I think that means the pilot was probably still flying the aircraft, and 180S Mag or True was used.
As far as CCYap calcs, so just take the Arc5 line in Excel and change the time to 2314, then add your lat/long estimates. Any path that does not show the 2314 is hiding it because it does not fit well, but in reality it does fit well, if your flight is the same as the real MH370 flight, probably descending 300 ft/sec at twilight.
EDIT:
At the moment as a contrarian I am saying Z was not using NOBEY or NZPG in the Flight Sim, he may have used 180S Magnetic which also probably ends around 45S1.
@DrB
“Your last statement that “none of the other logon events presented by Holland exhibit any sign of overshoot behavior” is ludicrous. They all show overshoot behavior.”
Really?? Cut-paste below from Holland paper.
begin cut-paste//
With the removal of the untrustworthy BFO logged for the first message in the sequence shown in Fig. 7, the 18:25Z log-on from MH370 was also determined to follow the simple decay trend observed in other instances.
end cut-paste//
At least Holland (and ALSM) knows what an overshoot is. Apparently you do not.
You also say:
“I explained in my paper the reason why none of Holland’s other log-on transient events show an initial small error and then show overshoot. That is because this is the only in-flight log-on event where the power was off for an extended period of time.”
Why don’t you tell us why having the power off in flight makes any difference to the oscillator settling behavior?
@TBill, Victor
Actually, I just followed the advice of post #2 above (BEBIM to Wilkens Runway at -66.690833, 111.523611 according to the Wikipedia article, it also has a dot on google earth), and that gives very good results. The heading is ~171.3, and the distances between the rings work out quite well.
Give it a try!
Position at the 7th arc would be ~-31.1, 97.2
@Nederland: You need to be careful about terminology, in particular true versus magnetic, and track versus heading. What exactly do you mean?
@Victor
Yes, it’s definitely amateur-ish what I am doing, I’m simply connecting waypoints on google earth, checking this against the BFO calculator, but I do think my arcs are correctly calibrated to match the results of the calculator (and the theory makes some sense, to me at least)
With “heading” I simply mean the heading (degrees) that comes up on google earth (ruler/line menu).
@all
What is the minimum Arc6 altitude required to make the Arc7 events unfold as we think it may have happened?
@Nederland: Using GE in the manner you suggest would tell you the “true track” azimuth, which varies along the path.
@TBill
I just used the method you described above to get a result on the 23:14 BFO, and it comes at 219 Hz (216 to 222 Hz observed as per DCL appendix). This is for the parameters described in the paper above (171 “heading” from BEBIM), zero rate of climb.
It probably takes a while to recalculate everything for the BEBIM – Wilkins Runway route, which is currently my “favourite” (but it’s very close, so the BFOs should be very similar too).
I’m not sure about the minimum altitude (the theory is MH370 was flying just below 29,000 ft), but I’d assume a reasonable altitude is needed to allow for the descent rates observed in the final BFOs.
Dennis:
Re: “…Why don’t you tell us why having the power off in flight makes any difference to the oscillator settling behavior?…”
Well, we have tried to explain it several times, several ways. I’ll try once more.
Thermal control loops, as used in OCXOs, typically have 2 “switched modes”. At cold startup, the loop filter is essentially bypassed, and the heater is switched on 100% until the oscillator temperature gets close to the set point temperature (typically +75 or +80C). Once the temperature gets close, a 2nd or 3rd order loop filter is switched in and the temperature settles to the set point temperature, with some classic overshoot and decay…sometimes with slight exponentially decaying ringing for a minute or two. If the OCXO ambient temperature is 50C below the setpoint temp at start up, and the OCXO has been off for >30 (cold start), the oscillator temp will be closer to ambient. This means the heater will come at 100% on for several minutes to get close to the setpoint as quickly as possible. Thus, the SDU BITE algorithm will likely finish first all tests except the oven temp test. OTOH, if the OCXO power has been off for only 1 minute or so, like at 00:19, then the time to settle is much less. In fact, it may not even start in the initial 100% heat mode if the power on temp is only 2-3C below the setpoint temp. It may start with the loop already closed, and exhibit only a small transient. In any event, if the power was only off 1 minute, it is likely that the OCXO temp BITE test finished before some of the other BITE tests, resulting in little or no sign of the transient in the BFO values at 0019, but a clear transient at 1825.
Also, with Wilkins Runway (YWKS) it is clear that it was in the navigational database, otherwise the ATSB would not have considered it as a possible waypoint in the search for MH370 (Martin Dolan on 20/8/2014). It is unknown whether or not NOBEY was included in the database, however.
https://www.atsb.gov.au/infocus/posts/2014/cautious-optimism-in-search-for-mh370/
@ALSM
Thanks, but that was not my question to DrB.
@ALSM
My question was why the inflight settling behavior of the oscillator should differ from the stationary settling behavior of the oscillator. Said another way, why is logon 7 of Holland’s figure 9 the only logon which exhibits what might be incorrectly interpreted as an overshoot.
@DrB,
“If you or ALSM think my description above is somehow incorrect or not unique, then please provide an alternative explanation.”
Not to say incorrect, but very unlikely. The reason is the very first BFO 18:25:27, which is consistent with the radar data. Holland simply calls it unreliable in his paper, but does not bother to explain this remarkable coincidence. If my understanding is correct, the IG thinks that Holland did not understand what he wrote in his paper.
The first clear explanation of this effect was provided by Victor (you omitted it in your post; ALSM also prefers not to mention about it): the first BFO is correct because transaction begins upon reaching certain temperature, which is not equilibrium yet. Subsequent BFOs are abnormal, because of the temperature overshot. Fair enough. However, this explanation runs into several problems:
1. Holland presents 7 sample curves, 6 of which exhibit monotonic (within the tolerance error) behavior, and the only one, which is non-monotonic, is the one corresponding to MH370 18:25 logon.
2. From Holland’s paper it appears that the steady-state is reached within 40 to 90 seconds, not within 3-4 minutes. Can low ambient temperature be sufficient to explain the delay of 3-4 minutes? Is the heating element not powerful enough, or thermal capacity of the whole unit is too high? If yes to one of them, then why so large overshot?
3. As long as the overshot in BFO is explained by overheating, the same equally applies to 00:19 BFOs. In such a case, the duration of power outage should not matter. This means that the BFO of -2 Hz is incorrect. Based on Holland’s paper, the correct BFO should be at least 30 Hz lower, meaning even higher descent rate – less physically plausible.
On the positive side, the decay presented in Holland’s paper can help to eliminate inconsistency in the last two BFOs of 182 and -2 Hz separated by 8 seconds: perhaps the descent rate at 00:19:29 was already similar to the descent rate at 00:19:37. This helps to get rid of the sudden 0.68g acceleration over 8 seconds.
If, in addition, we assume that there was no power outage prior to 18:22, the only remaining problem is too explain the abnormal BFO of 273 Hz, as all the subsequent BFOs can be explained by a turn combined with descent.
@Oleksandr
I do not believe a power outage occurred prior to 18:22. The logon was caused by something else, IMO, and that something else (no opinion what it was) also produced the abnormal 273 Hz BFO. I agree with your explanation for the subsequent BFO’s.
@Ge Rijn
I don’t know if this works for your debris analysis, but I am suggesting the BTO/BFO’s from 2241 are possibly explained by descent at 300 ft/sec starting at twilight (approx. 22:50). The pilot then brings the aircraft to 10,000 ft (FL100) by about 23:00 or so. The FL100 brings the aircraft into the tops of the cloud layer that morning. As the fuel runs out the pilot glides to 5000ft below the clouds to survey the surroundings. Optional final maneuvers if he picked a good area from above…it would be a backwards maneuver to get back to a selected visual spot.
@Oleksandr,
@DennisW,
Neither of you have an explanation for the 273 Hz BFO (including why it is different from all the other log-on events). ALSM and I (note we are not the IG) have a quantitative explanation that is entirely consistent with ALL log-on transient BFO errors (including the 273 Hz), taking into account that the initial starting temperatures of the OCXO were different, as they generally must be.
If you ever come up with an alternative theory, please let us know. In the meantime, you have no basis for your subsequent assumption that the other BFOs circa 18:25:34 are all valid and contain no warm-up errors. In order to make that assumption, you need a theory for the 273 Hz that allows the prior and subsequent BFOs to be error-free.
@Oleksandr,
Thank you for your questions and comments related to the 273 Hz BFO.
You said: “Not to say incorrect, but very unlikely. The reason is the very first BFO 18:25:27, which is consistent with the radar data. Holland simply calls it unreliable in his paper, but does not bother to explain this remarkable coincidence. If my understanding is correct, the IG thinks that Holland did not understand what he wrote in his paper.”
I don’t know what the IG thinks of this. I am not part of that group.
The reason Holland failed to explain the 18:25 log-on sequence is that he failed to understand its uniqueness. Indeed, it is the only in-flight log-on where the power appears to have been off for a significant period of time. In addition, he failed to appreciate the multiple SDU BITE tests that must be satisfied for the first log-on request to be transmitted. To the best of my knowledge, this was first pointed out by Don Thompson (not by Victor). That low OCXO starting temperature takes a much longer time interval in full heating mode before the temperature/frequency error gets small enough to allow the closed-loop servo to be switched in.
You also said: “. . . the first BFO is correct because transaction (sic) begins upon reaching certain temperature, which is not equilibrium yet. Subsequent BFOs are abnormal, because of the temperature overshot. Fair enough. However, this explanation runs into several problems: 1. Holland presents 7 sample curves, 6 of which exhibit monotonic (within the tolerance error) behavior, and the only one, which is non-monotonic, is the one corresponding to MH370 18:25 logon.”
This is correct. Only the 18:25 log-on has an initial transmission with a small frequency error, because of the fact that the other BITE tests are already satisfied, and the initial transmission in awaiting only a sufficiently small temperature error in order to be initiated. For all the other log-ons, the first transmission occurs AFTER the temperature error has passed through zero and the peak overshoot has occurred. Thus in those cases the transmissions occur during the DECAYING portion of the transient response (after the peak overshoot).
You also said: “2. From Holland’s paper it appears that the steady-state is reached within 40 to 90 seconds, not within 3-4 minutes. Can low ambient temperature be sufficient to explain the delay of 3-4 minutes? Is the heating element not powerful enough, or thermal capacity of the whole unit is too high? If yes to one of them, then why so large overshot?”
I’m not sure where you get 3-4 minutes. As shown in Figure 3 and Table 2 in my paper, the difference in the log-on request events (relative to the first zero crossing of the temperature servo) is always less than 80 seconds. The amplitude of the frequency overshoot will depend linearly on the initial temperature error up to a maximum value (whose value depends on the velocity error when the integrating servo is switch in, and the velocity error then will depend on the maximum heating rate). For very large temperature errors the frequency overshoot will be constant, but the time to reach the operating band will depend linearly on the temperature error.
You also said: “3. As long as the overshot in BFO is explained by overheating, the same equally applies to 00:19 BFOs. In such a case, the duration of power outage should not matter. This means that the BFO of -2 Hz is incorrect. Based on Holland’s paper, the correct BFO should be at least 30 Hz lower, meaning even higher descent rate – less physically plausible.”
As I explained above, the frequency overshoot error will generally depend on the initial temperature error, and this will not be the same at 00:19 as at 18:25 because of the difference in depowered times.
You said: “On the positive side, the decay presented in Holland’s paper can help to eliminate inconsistency in the last two BFOs of 182 and -2 Hz separated by 8 seconds: perhaps the descent rate at 00:19:29 was already similar to the descent rate at 00:19:37. This helps to get rid of the sudden 0.68g acceleration over 8 seconds.”
I agree that it slightly reduces the required acceleration then. In Section 18 of my March paper I showed that the impact at 00:19 was 6 Hz or less.
@DennisW:
You said: “Why don’t you tell us why having the power off in flight makes any difference to the oscillator settling behavior?”
I have done this several times on this blog and in my paper on the subject in March. Why don’t you bother to read it? Here is a quote from my paper:
“Clearly the ambient temperature is much lower in flight than on the ground due to the -45C outside air temperature chilling the SDU compartment through the aircraft skin. That is why the transient amplitude in Figure 4 is about three times larger in flight than on the ground for the same depowered time. The percentage overshoot of the temperature controller to an input step function only depends on the damping ratio. However, the peak overshoot (in Hz) depends linearly on the size of the step and therefore on the starting temperature error. The larger the size of the temperature step (from the starting temperature to the set-point temperature), the larger the overshoot (in Hz) will be. Therefore the larger amplitude of frequency overshoot in #7 is predominantly a consequence of the larger starting temperature error.”
You also said: “ “Your last statement that “none of the other logon events presented by Holland exhibit any sign of overshoot behavior” is ludicrous. They all show overshoot behavior.”
Really?? Cut-paste below from Holland paper.
begin cut-paste//
With the removal of the untrustworthy BFO logged for the first message in the sequence shown in Fig. 7, the 18:25Z log-on from MH370 was also determined to follow the simple decay trend observed in other instances.
end cut-paste//
At least Holland (and ALSM) knows what an overshoot is. Apparently you do not.”
No, Dennis, you seem to be the only one who fails to appreciate that an overshoot means the parameter has gone too far and is past the set point. In this case the temperature and frequency are both too low when power is applied, and the servo drives both temperature and frequency past the set point to values that are TOO HIGH. Then a “DECAY” or a DECREASE, or a DECLINE occurs from a value which is TOO HIGH to eventually reach the set point. See my Figure 3 for a plot of this transient behavior. Holland only considered those BFOs that were too high (= overshoot) and were in the process of decaying to a “correct” steady-state value. Why were they too high? Because they all occurred AT or AFTER the peak overshoot in temperature/frequency had occurred, and they were settling downward to zero error. The only BFO that occurred BEFORE the peak overshoot is 18:25:27. All the remaining BFOs for all the log-ons occurred at or after the peak overshoot.
You appear to think that an “overshoot” and a “decay” are incompatible. In point of fact, you can’t have one without the other.
@TBill
You know I’m not an expert at the BTO/BFO stuff but (and therefore probably..) I don’t see why the BTO/BFO’s from 22:41 are possibly explained by a descent starting around 22:50 (~10 minutes after 22:41..).
As far as I can see the BFO’s from 22:41 till 0:11 make a fairly straight linear line which suggests a more or less level flight at least till 0:11.
Not a ~20.000ft descent (if the plane was at ~30.000) in 10 minutes between ~22:50 and 23:00 followed by another descent of 5000ft.
I mean wouldn’t this clearly show in the BFO’s at and after 22:41?
But I like the idea of a pilot bringing the plane with a steep descent to (or even preverably through) the cloudtops around twighlight. Maybe starting the descent right after the first engine stopped telling him he had maybe ~15 minutes of powered flight left. Then the 8 sec captured between 19:29 and 19:34 would only be a snapshot of this descent after the second engine stopped. After this descent leveling out at ~5000ft and choosing his best ditching options (not necessarelly backwards imo..why you think so?) seems quite possibly to me.
One problem ofcourse is the missing IFE log-on at 0:21. This has to be explained other than a crash before 0:21.
Was the IFE switched off?
Is it right what Paul Onions stated the IFE is load-shedded under only APU power?
@DrB
Overshoot refers to the shape of the transient response.
https://en.wikipedia.org/wiki/Overshoot_(signal)
My concerns with your and ALSM’s theory are:
1> None of Holland’s examples show any overshoot behavior at all except the 18:25:27 event.
2> The BFO logged at 18:25:27 is exactly what you would expect it to be based on the radar data. That is truly a remarkable occurrence if the oscillator was simply drifting through a warm up transient.
You could be correct, but there are serious warning bells going off here. I am merely pointing them out. My original blog post (which you responded rather rudely to) simply pointed these things out, and suggested that the reader draw their own conclusion. I did not take sides or suggest anyone was wrong. That is part of the new me, BTW. 🙂
you say (relative to Holland’s other six examples):
“Because they all occurred AT or AFTER the peak overshoot in temperature/frequency had occurred, and they were settling downward to zero error. The only BFO that occurred BEFORE the peak overshoot is 18:25:27. All the remaining BFOs for all the log-ons occurred at or after the peak overshoot.”
How do you know the above? As far as I am concerned Holland’s data reflects all of the BFO values collected in each transient case. None of them replicate the shape of the 18:25:27 BFO history. I think you are making an assumption that is not supported by anything in Holland’s paper.
@Ge Rijn
Sorry- a couple of sloppy errors in my orig post.
> The descent I currently see is slow and gradual: 300 ft/MIN until 00:00. Your right though he could have stepped up the rate of descent at sunrise 23:35ish.
As far as hard data, I am saying the Sat call BFO at 23:14 fits better with 300 ft/MIN descent rate, and the arrival at Arc6 is later than expected, so that’s consistent with an extended slow down from a descent. Also BFO at 00:11 is level flight (maybe not the descent proposed by IG – the IG’s observed descent at 00:11 was due to their far out 38S location- BFO suggest no descent required at 00:11 if we say end point more like 32/33S). Additional data – we know we have twilight, sunrise, fairly thick Cumulus? cloud layer, and a very stiff wind from the West.
So my current interpretation (but I need help):
1> Descent per above
2> Level flight at 00:11 maybe at FL100, so in the clouds to hide maybe.
3> In preparation for splash down, a shift to below clouds to see water surface for any unexpected company. Also maybe looking for (no traffic) spot under clouds to crash.
4> Fuel runs out, so I think there is still the needed glide time for 4 minutes to allow SDU reboot. If it was me, I am probably taking a turn to the East to ride the wind and make a misdirection under the clouds, just in case some satellite got a bead on me above the clouds.
5> The 0 altitude BFO applies for Arc7 so that means the entire *inside* Arc7 area has not been searched in the 32-24 zone I am thinking for this scenario.
Correction: the 32-34S zone is end point for this scenario
@DennisW
The whole affair has been plagued by remarkable occurrences, not just the uncanny BFO at 18:25.
Just as uncanny, nay, even more uncanny is the vertical acceleration between 00:19:29 and 00:19:37, during an arbitrarily determined 8sec interval of the descent phase.
Then there is the equally uncanny occurrence of a hijacking that takes an evidently disastrous turn for the worse when things ie negotiations don’t go to plan, but gives out no clues as to what actually went on, even after more than three years, other than flying into the distant reaches of the SIO, after doing dark, and maintaining total radio silence on reaching the first FIR boundary. No whistleblowers have come out of the woodwork. A complete zilch on that score!
But the most uncanny occurrence for me is this one: A plane takes off with a particular quantity of fuel on board, appropriate to a particular destination. However, when the putative flight path is calculated using Bayesian analysis of Satcom handshake transmissions, assumed likely aircraft dynamics, cruise speed ranges, assumed MCR etc, it turns up something very spooky. It actually appears, against all the odds you would expect on a flight subject to a set of un-preplanned, unconnected events, to end up at fuel exhaustion with sun angle amenable to a controlled ditching, in a particularly remote part of the SIO, after flying for the greater part of the journey under cover of darkness. Now that is one coincidence!
I like the new you, BTW 😊
[Comments here are closed. Please continue the discussion under the new post.]