Regarding the glide issue and autopilot.
I believe a modern computerized airliner already has almost all the data necessary to enable its computers to calculate in real time how it can best make the runway at the present thrust setting, be it partial thrust or zero.

The missing pieces of data are a) the airplane's lift/drag curve in various configurations, which the manufacturer should already know, and b) the wind gradient in the remaining flightpath until touchdown (usually the headwind decreases, which in this case is good and adds margin if we simply use the instantaneous wind data).

So in theory it's not that difficult to devise a software that automatically commands the autopilot into optimum glide mode at the first sign of a thrust deficit that prevents it from holding the commanded glideslope at Vref. In case thrust comes back, it would revert back to normal glideslope hold mode.

Whether that "auto glide mode" would be a truly useful feature or only serve to confuse the flight crew, is another discussion. Training would of course be essential. But a technical solution is certainly possible.

Just my $0.02

Simple software needed
 

I fully agree. There might be as well a data base of obstacles, allowing for the best way to avoid obstacles ...

However ... I have a more readily available solution :

1. Just tell us where the best glide speed is, in % of the VRef. That percentage is a constant for a given aircraft configuration. For an aircraft with a high wing aspect ratio, in a normal landing configuration, I suspect that the best glide speed is somewhere between 90% and 115% of VRef (educated guess).

2. The pilots already have a fair idea of the wind gradient, because they know the instantaneous wind ... and the ground wind as well (being "wind minded" is a part of airmanship, isn't it?)

3. The stronger the head wind, the stronger the wind gradient (rule of thumb ...)

4. The stronger the head wind ... the higher the need to inscrease the speed somewhat above the best glide speed. Add one third to one half the head wind component. Additionnal benefit : the aircraft will be low sooner, where the head wind is less. Extra speed will convert back into distance. Extra distance covered in ground effect is a bonus, regarding drag reduction.

5. Let us know also for sure the real effects of partial flaps reduction.

6. The pilots can see the main obstructions, and deviate if necessary.

Nothing is sacrilege in such an emergency.
The first available software is in the pilot brain, its common name is "basic airmanship".

Perhaps there is a need to train from time to time for the maintenance of that precious software.

That can be done but remember the BA38 was not a glider: the engines were still producing significant thrust, albeit not enough to maintain a 3deg slope in full landing config. This would have had quite an effect on the optimum speed to fly (whatever that was...).

A possibility would have been to take the aircraft into ground effect earlier but that's hard to model correctly as it is very dependent on the precise nature of the terrain and the aeroplane position relative to it. Simulators are good for many "what if?" exercises but they can throw up "false positives" unless you fully understand the limitations of the simulation environment and what the software is/isn't taking account of.

Definitely!

Agreed, but in the context of this particular accident, I don't see much training value. A slight change of circumstances might lead to a very different method to achieve an "optimal" outcome; there's not much point in training for a 1x10-9 scenario when your resources are limited and you want to achieve the best "bang for your buck". There is also the point that, generally, we don't play out scenarios in training that have a crash as the most likely outcome, despite crew actions, for obvious psychological reasons! This is reflected in the Boeing manuals, where they only deal with a dual engine failure at height and effectively assume that you get one or both restarted...

PAXboy
A little sanctimonious. I thought society had moved on from unnecessary secrecy 'in case the Public misunderstand and panic ensues'.There should be periodic updates to keep us in the picture.
But we have had no news from anyone for two months now, which is unusual to say the least.
I agree that the tenor of this thread has been amiable, but being kept in the dark for so long on such a significant issue naturally leads to a bit of impatience and an impression that something peculiar is going on.

CAAAD,

I disagree.

1. 'Periodic updates' do not give a full and complete picture (by definition) and so allow those who write headlines to publish banners which have little link to reality. Further, as an investigation progresses, things which may (earlier in the investigation) have appeared to be causal, may in fact not be whilst new things pop up; 'periodic updates' might therefore be contradictory or confusing if read in isolation.

The best 'periodic update' would be "we're still working on it and will let you know when we have the full story. Until then, naff off and stop making us waste time we could be spending on the investigation."

2. It is not at all unusual for investigations to stretch into several months (indeed, years). Not having heard anything for "two months now" is not unusual in the slightest.

In my view AAIB has been very (too) shy on basic data releasing:

1- Contrary to M.Mouse last comment, no mention of additional thrust request between TOD and 2 miles final. It has been rightfully suggested by Chris Scott here but that’s it.

2- How and when the APU door did open ?

3- "Fuel crossfeed valves indicated that they were closed …" but were they actually closed … ?

4- No mention of actual ZFW versus planned one.

5- No mention if the actual flight time of 10:35 was noticeably shorter than the planned one.

6- It is not clear if the planned flight level was restricted around 320 up to London, or if further steps were planned later on and if so where ?
Except from the early statement "to accommodate a request from ATC …" it is not clear if the crew choose to disregard a conservative low flight level flight plan computed to avoid "unusual low temp conditions compared to the average, but not exceptional"


I’m sorry but all these points are relevant when both engines fail to deliver almost simultaneously, especially in the last minute of a 10 hours flight … and following comments:

- "I don't discount any of the possibilities but IMHO the faulty fuel monitoring system (incorrect calibration?) appears marginally the most plausible" from snanceki

- "You don't get it much more serious than being OUTAGAS" from Spaz Modic

should not be taken lightly.

Gliding considerations
 

After this one, I'll stop!

Thank You FullWings for showing some interest in piloting techniques considerations.

Not significantly : Except for the wind, flying a jet at best L/D always provide the best angle ... whatever the thrust available. Best angle of climb, or lesser angle of descent. In a given configuration, flying best glide is simply flying minimum drag (in that configuration).

Sure, there are priorities in training ... Shame on me ... I didn't care about dual engine failures when I was in charge, I have to admit. But today I would hate to be left without any information regarding this kind of situation. Or to let pilots of my company without any hints.

How much training is at shake? Some lectures, one hour in the sim, demonstrating typical situations ... Replaying the sequence of BA38 uses less than two minutes of sim time. That should be enough to get the facts ... No need of course to demonstrate 100% success in order to get a "pass" ...

Perhaps some kids are now becoming champions at that game on MS Flight Sim ... (Do not underhestimate the value of that king of cheap simulation ... as far as I know, some aircraft models are quite realistic, and the sofware is excellent)

Let us hope that dual engine failure never become "business as usual".

But the scenario leading to the discovery of an engine failure in short final is already quite standardized - CDA it is. After all, facing a lack of engine responsiveness at the very end of flight is the result of a standard CDA scenario : using idlle power (or so) from top of descent, the trouble, if any, will show up very late. Almost allways at the same point, i-e somewhere between 500 and 1000 ft AGL if pilots and ATC really work at it ...

Nowaday, I guess there is some stress in a 777 cockpit at the end of a perfect job done during a CDA ... Idlle thrust or so from top of descent, feeling the need for aditionnal thrust only a couple of milles towards touchdown. Will the engines spool up? Are there spooling up fast enough? Let us keep an eye on them, make sure they don't roll back! Keep on, threshold coming soon! Finally :D!

I guess that 777 crews todays are a little more suspicious about low temperatures, icing messages, exotic jet fuels and ... perfect CDAs ...

Any volunteer to replay flight BA38? Exactly, I mean, in the real world.

Good flights, Gentlemen!

It has been stated ad nauseum that the aircraft had plenty of fuel on board. Around 10 tonnes I recall, certainly in that region.

While in an ideal world idle power from TOD until 1000' would be the most fuel efficient method of descending it just does not happen that often in busy airspace and almost never will it happen into LHR. The emphasis on CDAs is reserved for the descent commenced at some point after the holding fix and that emphasis is due more to the reduction in noise for those below the flight path than for simply economy.

Er......no.

CAAADLast year in the UK, the Governor of the Bank of England said that, if the Bank had been allowed to arrange bridging finance for Northern Rock in private, then he was confident that the matter could have been dealt with. He wasn't and the bank was broken.

We have no means of knowing what would have happened if they had made a private arrangement. What we do know is that the British media are sluts for trawling anything that can be sensationalised. It is not the public but the Daily Mail that you need to be worried about.

It is quite possible that we shall never know the full reasons for this failure and loss, until it happens again. Consider how long it took to discover what was wrong with the Comet. For now, I am happy to wait and feel glad that I am not working on the AAIB team, or in any of the companies involved for there is no doubt that this is a very difficult, and intractable, mystery.

@ M.Mouse
Without double checking I believe you have overlooked two points.

1. The word INDICATED was used.
2. No statement as to where the indicated fuel was stored. i.e. Wing Tanks or CWT. The fuel needed to be in the wing tanks.

So why has this been left with so many possible interpretations.

Was there adequate fuel in the wing tanks? The wing tanks were undamaged plus detritus was found within the tanks (quite normal as I understand it). So how much fuel was in the wing tanks post landing?

They must know the answer to this question BUT for whatever reason they are keeping the info close to their chest! The CWT was damaged and therefore the amount in the CWT immediately prior to the event is unknown.

The AAIB also stated that a significant amount was spilled, showing that there actually was fuel on board.

As you said, the tanks were checked.

Don't you think that the AAIB would have mentioned it if there had been a stark discrepancy between indicated fuel and the amount actually found in the tanks?

"Oh, there's no fuel in the wing tanks. Do you think that's worth mentioning in the report?" --- "Nah, just an unimportant little detail. It'll only confuse the public."


Bernd

I have no experience of landing heavy aircraft like that, but if I felt that the rate of descent was too high to allow a normal landing without breaking the gear, I would definitely try and put it down on soggy grass rather than a paved runway even if the impact speed was slightly greater. Braking effect of the grass would be greater, shock on point of impact less and the chances of fire greatly reduced. In summary, I'd probaby shorten the glide to land on the grass.

James

Hi JamesCam,

The conventional wisdom, which I support, is that the hard pavement is better, provided its LCN is up to the job. The last thing you want is the wheels burying themselves into mother earth. Even on a smooth touchdown, the tyre pressures (over 200psi) of a large aeroplane mean that the latter is inevitable on grass, and this was the cause of most - if not all - of the damage to Echo-Echo.

Hi Chris, thanks for that. I take your point. I suppose in many ways the collapse of the gear under extreme heavy landings is probably the most signifcant factor in absorbing the shock of the impact?

James

Hi James,

Yes: the gear definitely absorbs vertical kinetic energy, preferably without collapsing in the process! I'm afraid you are now taking this topic well beyond my expertise. I have suggested that the aeroplane would have remained intact, had it contacted the runway at the same VS; but the truth is, I don't know. The AAIB has not stated the VS.

If the landing gear collapses on a runway, my GUESS is that there is likely to be less secondary damage to the rest of the aircraft than on grass. When the engine nacelles strike a runway, they will not dig in; likewise the nose or a wing tip.

Chris

Erm, the AAIB state quite clearly that the total fuel was:


Yes, they state quite carefully that this is indicated fuel, because that is what they have the precise figures for. They don't have precise figures for actual fuel remaining in the wing tanks because it ended up all over the crash site through the open spar valves. That's how we know the fuel was there, we don't know exactly how much, but if the indicators were right it was ten and a half tons.

Unless you are contending that someone filled a fire appliance with fuel instead of foam ?

Sorry to bring these comments up again

As someone who has a Software Eng Degree, several years commercial experience with industrial strength, fault tolerant, 'mission-critical' applications (I hate that phrase too) and an ATPL I feel that it would be almost impossible to create such an application.

For a start that the aircraft would need to be fitted with so many sensors that its drag curve would be a vertical line. A database of global obstacles would be so onerous to create that it would be Terabytes of data and commercially un-maintainable.

In short we forget how wonderful the MkI eyeball attached to the Mk10 human brain actually is to sense inputs (topography, wind, aeroplane charactersitics) and meld these with prior experience to produce a safe and effective plan.

I do agree that the best the aeroplane's onboard computer can do is target Vref (or stick shaker = Vref + a bit for the technically minded) for the instantaneous inputs it is receiving.

Hope this helps. Oh and another well done to the BA 777 boys and girls! Those MkI eyeballs and Mk10 brains sure do work well under pressure!

@bsieker and Infrequentflyer.

I was merely stating the importance of the words Indicated and that the fuel needed to be in the wing tanks in order to supply the engines.

I'm neither supporting or denying that the aircraft ran out of fuel due to lack of fuel in the wing tanks. I'm just surprised that the wording used seems to have left this whole issue "open".

Yes the spar valves were found open but is it reasonable that there was the correct amount of fuel in the wing tanks and that it all leaked out post accident or did the spill come from the damaged CWT where the fuel was not supposed to be?

We just don't have the FACTS to answer these questions but I find it hard to believe that the AAIB doesn't.

The spar valves were open but this doesn't define the rate of leakage from the wing tanks post event which would be influenced by wher/what the leak was.

We will just have to be patient.

I don't know the design details on the 777 specifically, but from what I do know of design of similar sized craft, the oleos will absorb vertical forces up to a point, but when it comes to horizontal (drag) forces - eg. when landing on grass - the gear is supposed to detach. My understanding is that it also is supposed to detach cleanly and without damaging the main spar and fuel tanks (for fairly obvious reasons).

In this case, it looks like the gear (at least on one side) punched up through the wing structure prior to detaching due to the drag. From that observation, my fellings are that either:

a) the 777 gear didn't detach cleanly (as, I think, it will have been designed to do)
or
b) the vertical forces were way over the capability of the gear, at the point of impact, before drag forces could detach the gear - i.e. VS was beyond the limits of the gear.

If it's (b) then my guess is that the vertical forces will be the same on a runway (maybe more since soggy grass might have absorbed some) and you would still punch the gear up through the wing. The gear may or may not detach after that, but the damage to the wing would have been done - my guess is that the wing (main spars) is the major damage that has written the hull off. Also, while the plane (with gear or not) might slide easier on the runway, wet grass would be a lot less likely to create sparks, which might be a consideration when you've got fuel everywhere...

Overall, I think it is quite possible that where they came down was the best place (from a survivability point of view) to do so for that impact.
However, I'm not sure any of this is a useful comparison since I would be very suprised if, in the same circumstances, you managed to extend the flight as far as the runway and still have the same sink rate (but now I'm well beyond my expertise...).

I don't think it has been left as open as that, whilst they have been careful to say that the fuel was indicated to be in the wing tanks, they also haven't said (or implied) anywhere that those indications are suspect. [In my view stating that the fuel figures are indicated does not imply in itself that they disbelieve the figures]. We know that they cannot have figures for fuel-on-landing from post-landing measurements because of the leakage.

Also in the report they have said that the leakage was through the spar valves, which means (if I understand the fuel system correctly) that the fuel was from the main wing tanks, not the CWT.

Hi demomonkey,
 

thanks for your comment. However, I still believe it's technically doable with reasonable effort.

The words in (my) italics are the key here. Complete engine failure on final approach is evidently judged to be such an unlikely event that it's not included in training or simulator exercises. That was most probably a contributing factor to the fact that the airplane got near the stall and in that process irrevocably lost some dearly needed energy that might have taken it to the runway.

And as others have said, the flight crew is not to blame for it. They were faced with a situation no one had prepared them for and did what they could - saving the lives of all aboard, which in the end is all that matters.

I would suggest that the loss of energy is due to the double engine failure rather than crew actions?

Jeez, you people....

Quoting infrequentflyer789:

Fuel leakage through the open spar valves, with the boost pumps switched off due to the fire drill, means there must have been a considerable amount fuel in the wing tanks. With the aircraft coming to rest wings level (which can easily be verified from published pictures), wing lower surface at the location of the spar valves has a considerable amount of dihedral downward to the rib separating wing tanks from CWT.

The spar valves are located somewhere outboard of the main landing gear and just inboard of the engine pylons and certainly not at the lowest point in the wing tanks because of spar valve location further away from inboard rib, separating wing tank from CWT, and dihedral sloping upward toward the pylon.

From the last picture taken of the aircraft while still airborne (taken from the left looking up toward the aircraft), one can clearly see the lower wing surfaces. The location of the rib separating the center tank from the right wing tank is clearly visible. The wing lower surface of the right wing tank has a frost layer beween that rib and engine pylon, meaning there is fuel in that tank, at least all the way up to the engine pylon. However, inboard of the rib toward the fuselage, the lower surface of the CWT has no frost which means no fuel is in contact with the lower wing surface in the CWT, hence the CWT is virtually empty.
Frost is also present on the lower surface of the left wing tank but barely visible because the left engine nearly obscures that particular wing surface area. You will notice the absence of frost on the left side of the CWT as well though.


Green-dot

A Plea?
 

Ahh...when the "report" is finally published or someone REALLY knows what's going on, would you please publish a separate thread on the subject?

Trying to get some sense out of this one is a bit like herding cats! :ugh:

G'day ;)

Software engineering
 

Are you really aware of the state of the art in flight simulation? Very realistic 3D rendering of landscape and obstacle around airports an beyong. No need for terrabytes of data ...

Obstacles data base are actually used (and thus maintained) to compute landing and take-off restrictions, taking into account all the relevant factors. That kind of software is in use for decades by airlines or specialised service providers. Both data base and software are "accepted" for those computations by the authorities for flight dispatching. (Grey zone, in any way).

Any modern airliner is fitted already by an air data computer. No need for additionnal sensors. Thanks to GPS integration, and differential GPS technology, the position (of the antenna) is known within 10 ft horizontally, 30 ft vertically.

So what? When a runway is selected, a small portion of the obstacle data base should be querried and the data set made readily available for either or both :
- "visual" presentation to the crew (like in a sim)
- emergency trajectory computation by any decent computer. No need for terraflops here ... most of the computations should be done "well before" and the results stored, ready to be retrieved at the time the runway is selected.

As it was stated before, the only important missing information (today) is the "future" wind. The present wind is wel known, the ground wind (somewhat averaged) is known by the tower, but there is no direct link with the onboard computers. And no information "in between" - but good models are available.

I think that the available technology is not put to its maximum use ... Cost is the factor, but also lack of imagination.

Well, in my world an engine failure means a loss of the energy supply. Which is a good reason to use the airplane's remaining energy in the optimal way.

Biz47,

obviously you have no experience with designing or building SW, let alone SW for what would be a safety-critical application. But you presume to tell a self-identified professional IT person that there is a lack of imagination. There may be and it might well be yours. In a world which does not yet know how to deal with map shifts, building reliable, fault-tolerant, fail-safe systems which do some of those "simple" things you mention, to the required standards of reliability and fault-tolerance, is beyond state of the art.

PBL

I have some ... Not on a big scale, I do admit.
But I have some deep knowledge of flight operations.

I confirm that reliable sofware is already available for terrain visualisation and terrain avoidance :
- Flight simulation software (all levels, PC driven and higher) is amazingly realistic (terrain visualisation and aircraft handling) => allowing approved "Zero flight time" aircraft type rating (no perceived difference between simulator and real aircraft handling).
- Military software allowing flight at low altitude inside valleys ... relying more and more on stored terrain data and less on radar.

And of course, reliable software already exists regarding aircraft handling, allowing auto land in zero-zero, and flight enveloppe protection ...

Please note that airbus flybywire software does not require supercomputers : an Intel 8186 does the job pretty well. 20 years old technology ... much cheaper and much more powerfull hardware is available today - about 1000 times more powerfull ....

Note also that more than 10 years ago, cheap (but certified) general aviation GPS receivers already included some kind of advisory terrain elevation data (grid mora).

And ... today, the most popular general aviation GPS manage to display a nice picture of the runway, and some landscape features as well, when approaching an airport to land.

Where is the lack of imagination? Well nobody did imagine that a modern airliner would meet a dual engine failure on short final, so nobody saw the need for a software to handle that - neither for a pilot training actually. That is lack of imagination.

At the point at which the engine failure became obvious, the plane had X amount of energy - some was potential energy, some kinetic energy. The need was to manage this energy to conserve as long as possible, which means flying the plane in the most efficient way e.g. at best L/D. Unfortunately that's not what the autopilot did - apparently it tried to maintain the glide slope, which will have frittered the energy away.

If you fly too slowly (which is what the autopilot did) you increase the induced drag, too fast and you incrase the form drag. Both will result in the plane not flying as far as it could for the original amount of energy.

See the total drag vs. speed diagram on this URL - where the two cross is the best L/D ratio, and the best speed to fly for the situation G-YMMM was in. Unfortunatly the crew had very, very little time to think about it.

http://pilotsweb.com/principle/liftdrag.htm

I am a bit pickled, but you are all fantastic, hindsight is a wonderful thing.

I think with the situation they had they did their best in the time available.

Lets praise them instead of nitpicking about what they could have done better.

I wonder whether I and you would have done as well on the day!!!

I've just crunched the data on published formal reports by the AAIB back to 2006 (as far back as I could be bothered to go). The average length of time from incident to final report publication is 25.6 months, i.e. a little over two years. This does not and has not stopped them issuing recommendations, where appropriate, before the final report.

I'm quite happy for AAIB to resist the rabid frothing of the news media for sensational information to fill their 24-hour outpourings, and instead concentrate on trying to work out exactly what happened... :=

Green-dot, you made a good point for materializing the presence of fuel in the main tanks.

Do you think the following selected area depicts also a frozen surface, which would show in the same time where the fuel extends outboard …?
http://i45.servimg.com/u/f45/11/75/17/84/mmm_on10.gif


Now, is the total frozen area large enough to confirm an actual fuel volume of 6500 liters each side ?

The crew did a remarkable job of using optimum flaps to get close to the runway end. Leaving them in final landing position would have resulted in a bad situation. They should be commended on their skills to save those people.

As this is the rumour network... And I do understand the inverse square law, and all that good stuff. I have to report the "dark" rumors on the street are, that it was electronic interference of some sort.

I have a real struggle to see how it could possibly be, yet the rumor is persistent, and independent.

RTFM - Spot on! :ok:

Biz47,

the difference between our views comes very likely because you are, obviously, amateur and I am expert in the field of embedded safety-critical digital systems.

PBL

PBL, Demomonkey
 

I think you are sidetracked by this proposed obstacle-avoidance software. Let's look at the feasibility of the original proposal, which was an autopilot sub-mode that would fly the plane to its best glide speed in case of a serious engine failure.

From a flight safety perspective, it can IMO be treated much like the stick pusher stall prevention device. The functions are in fact quite similar.
The basic difference is that in this case the primary trigger signal comes not from the AoA sensor but from a thrust deficit sensor. The autothrottle already incorporates the function of such a sensor, presumably engineered to flight-critical reliability standards, thus it already exists.

This document from Transport Canada specifies the reliability requirements for stick pushers in transport category aircraft. It says the probability for failure of the device to operate shall not exceed 10^-4. Furthermore, regarding the probability of unwanted operation, a very pertinent issue in this kind of device, no single failure shall cause this and the probability of unwanted operation from all sources shall not exceed 10^-5. IMHO that is easily achievable with well known methods.

In fact, the only additional component not part of standard airplane hardware would be the lift/drag polar curve data. I presume any transport category aircraft-certified computer would have error-correcting memory as a standard feature, so no additional reliability issue there. And the software algorithm is very simple.

Furthermore, this suggested device would only need to be armed at final approach and when at or below the glideslope, greatly reducing its potential to do any harm. Last but not least, it would only ever be triggered when a flight critical "10^-9" event has already happened.

To summarize, if industry can build safe stick pushers, building this device is not a problem.

So please tell me if I've forgotten something crucial. And to repeat: this is about the technical feasibility. The actual usefulness of this suggested device is another discussion. :)

Wing tank fuel.
 

Posted by CONF iture:
Yes, i believe that is also fuel induced frost.

I will do some calculations using scale drawings of the fuel tanks.

At first glance, using a B777 front plan view and drawing a horizontal line between the bottom of the wing tanks just outside the left and right pylons with the line disecting the wing tank/CWT rib, the fuel column measured from the bottom skin up reaches two-thirds to the top of the wing tank/CWT rib (i.e. top wing skin). Only one-third of total wing thickness at the rib (from fuel level to upper wing surface) is filled with air.

I could place the drawing here, if I only knew how. Apparently i do not have an attachment option, which according to PPRuNe FAQ i should have.

Should not be too difficult to calculate fuel contents (taking the dihedral of the wing bottom surface into account) if exact dimensions of wing thickness at the ref. rib, chord of the rib between front and rear spar, and distance from rib to just outboard of the pylons are known.

Green-dot

Interesting line of enquiry, you two, but I would suggest that the resulting figure might be slightly less than the amount on board. It seems likely that the wing bottom surface would have been experiencing positive TATs for at least 10 minutes before the photo was taken.

The aircraft left the Lambourne Hold, according to the AAIB, at FL90. The surface/sea-level temperature was +10C, suggesting something of the order of –8C at FL90. At a typical hold-exit speed of 210kts IAS, I think (in the absence of a Dalton computer) that this would produce a positive TAT. Perhaps someone can tell us? [It is conceivable that the TAT may have become positive prior to the speed-limit point, which would have been 8 minutes earlier. This may have been short-lived, however, and reversed as they slowed down into the hold.]

Time from LAM to touchdown is likely to have been 10 – 15 minutes, during which the skin below the dry part of the fuel tank would have been conducting relative heat towards the ‘wet’ part. This would be eroding the margin of the frost below the ‘wet’ part of the tank, I think. How much would depend partly on the fuel temperature; partly on the TAT.

You should treat whatever figure you arrive with as a MINIMUM estimate. Presence of frost means presence of fuel; conversely, absence of frost does not prove absence of fuel.

snowfalcon2,

pertinent comments, with which I do not entirely agree, but for reasons unrelated to this thread I probably won't be contributing further.

PBL