Oldlae,

Ops across the Siberian areas are not as precise as you seem to think they are. The loadsheet (and therefore calculated fuel burn) is provisional based on all male pax and predicted cargo. The final loadsheet is recieved after pushback and contains actual cargo and a Male / Female split therefore the final figures are usually a few tonnes less than the provisional figures. The flightplan fuel burn is derived from the provisional weights and ANTICIPATED atc altitude clearances (remember Siberia is metric non rvsm so there are large differences between available levels). If the crew are able to negotiate more efficient clearances enroute more fuel savings can be made. Also if the crew are able to negotiate better levels enroute the wind may also be more in their favour. All this for 10+ hours flight.

My point is that if all factors go the crews way it is not unusual to land with a few tonnes more fuel than planned. Likewise if all factors go against the crew they will have to use some or all of their contingency fuel. That's what it's there for.

Those of you that think it's significant that BA038 used less fuel than anticipated to get to LHR or that fuel quantity indications were innacurate are barking up the wrong tree.

LD

Ref the open spar valves, (LP cocks) I am surprised that the anomaly had not been picked up on the simulator. The FAA requirement to allow the correcting modification, splicing of the wiring, not to be completed until 2010 was probably based on the Boeing cockpit drill being carried out as directed. This appears to be such a simple mod that I question the thought process of however decided not to do the mod asap would not have ever envisaged that the cockpit drill was not being followed to the letter. This lack of communication is always a problem with large companies and I hope that a valuable engineering lesson has been noted. Always check any changes in procedures with the engineers.

Quoting the AAIB update bulletin S1/2008:

"The Air Accident Investigation Branch (AAIB) was informed of the accident at 1251 hrs on 17 january 2008 and the investigation began immediately."

Given the specific nature of this accident and the AAIB starting this investigation with an open mind towards any aspects that may have been contributing factors, which means securing the area and start collecting evidence as soon as possible. Is there a protocol to collect or register all personal electronic devices (PEDs) from the persons who "arrived" on this aircraft?

Since i have not read anything about this issue in the bulletin, that could mean several pieces of evidence may be missing if this has not taken place. Or are these PEDs to be traced (which seems harder to achieve) at a later stage if the investigation should have grounds to follow leads which take them in the direction of possible EMI?

Just a thought.

Regards,
Green-dot

Theory
 

I am not familiar with the 777 fuel system. If anyone is in a position to share schematics or system descriptions I would be most interested. However, with the information gleaned from this and the previous thread plus the AAIB special report, I feel that the following scenario is plausible and deserves further investigation.

This appears to be a common-mode failure. The AAIB appears to have exonerated as far as reasonably possible the engine control systems. Another potential common-mode source seems to be the centre-tank fuel system. According to my reading of NSEU's post #187, the centre tank boost pumps are "override" AKA "priority" pumps i.e. they provide a higher delivery pressure than the wing tank pumps in order to use centre tank fuel first (wing bending relief). This theory rests on that premise.

1) Significant quantities of ice accumulate in the centre tank.

-Posssibly over more than one sector if it does not have chance to thaw during turn-arounds. Maybe supported by CT excessive water warnings reported in, I think, the previous thread, from the American "leak". Maybe exacerbated by scavenge systems contaminated by FOD. Added to during the final descent.

2) Ice melts during descent, hold, approach, yielding a few litres of free water in "empty" centre tank.

3) Config change to flaps 30 at ~1000' leads to attitude change allowing water to migrate to CT boost pump inlets.

Note by design, boost pumps are typically located at lowest point in tank at typical cruise attitude. Increased flap => more nose down for same speed.

4) Fault condition (or crew action) causes CT boost pumps to re-start, or fault condition allowed them to keep running when tank emptied with no indication (or no reported indication).

5) "Override" pump system design causes CT pumps to win over the wing tank pumps and water/fuel mix from CT is consequently supplied to engines in preferance to wing tank fuel.

Note that following the landing, the open LP (spar) valves allowed the contentents of the fuel feed pipes to be deposited onto the sod. Similarly, the CT fuel contents could not be assessed for water contamination as the tanks had been disrupted and subsequently contaminated by the firefighters.

AP LEFT IN 'til fell over due (no doubt) to a/c unable to follow GS.

HERO pilot 'took over' and wrote a/c off in high sink rate landing.


:ugh:

Barbiesboyfriend
 

Well if "Hero" pilot hadn't taken over, er "ken" and the aircraft had been left to it's own devices the sink rate could well have have been a darn sight higher, with even more serious consequences..... I think your comment that he "wrote the a/c off" is, putting it politely, ********.

Ah, but WHICH engineers? The operator's engineers are unlikely to be familiar with the details of the certification of the design. Nor, for that matter, are the OEM's support staff, who may be an entirely different part of the organisation to the design staff. Similarly the Airworthiness Authority personnel dealing with operational matters may have little interaction with the Certification teams, and the same applies to those approving STCs, which may not even be approved by the original state of design.

Even if you can track down the right departments (the OEm's design staff, and/or the airworthiness people who actually did the cert approval), there's still no guarantee the people there NOW know why a specific design choice was made.

Syeng, problem with number one there. Centre tank is the warmest tank, so significant quantities of ice are unlikely to accumulate there. The reason fuel temp is measured in the wing tanks is because that is where your coldest fuel will be. Also, centre tank is emptied first, so fuel spends less time in there to cool down. It is very rare to be planning to land with fuel in the centre tank, so you're not going to get buildup of ice over multiple legs.

swiss_swiss,

ref #191, previous type was C130. Yes, mainly mandraulic and clockwork, so perhaps no direct read-across.

Electrics were monitored for over / under freq (each phase), over & under volt (each phase and average across the 3). If memory serves, any phase drops below 70v or average drops to 90v across all 3, genny trips off. In the specific incident, 1 phase of a genny failed (4 EDGs supplying 4 seperate AC busses). That bus was powering the Aux Hyd pump, which was generating enough back EMF on the dead phase to keep the bus powered and prevent the genny tripping off. The problem only showed itself after the gear was raised and the Aux Hyd pump turned off.

Reason I thought it was worth a mention? It hadn't occurred previously, in some 40 years of ops. Indeed most operator's (myself included) immediate reaction was that such a scenario could not occur, yet once the problem was investigated it became clear that it could. It's an example of something nobody thought of, manifesting itself late in an ac's service life.

I appreciate that modern machinery is way more complex, however, and I imagine more closely monitored to boot - so still, may or may not have relevance! I'm sure 777 operators such as yourself could rule such a scenario in or out of possibility.

Now I'm sorry to show my ignorance here, but where is air drawn into the tanks when the fuel is pumped out.
i.e could that be blocked by ice causing too low an air pressure for the pumps to suck the fuel ?

SyEng, inlets for the center tank pumps are not located at the lowest point of the tank. In fact, there is a substantial amount of residual fuel which they can't touch. Once this level is reached, the center pumps are incapable of producing pressure.

Fuel feed then commences from main tanks, while automatic scavenge pumps slowly and automatically transfer the residual center tank fuel to the main wing tanks. Typically, the center tank is literally empty after a flight of this duration.

The ice theories are the most enticing. I talked to a QF engineer today who was totally convinced it was ice blocking the fuel-oil heat exchangers. So if we accept for a moment that this hypothesis is correct, can someone explain how this fairly large quantity of ice made it's break for freedom and evaded the air accident investigators? Is it possible it melted and ended up on the grass?

The trouble I still have with the ice theories is that, if there is ice in the fuel tanks, it can't end up all in one place. Some would have to be still in the wing tanks waiting to be discovered.

Problem with ice theories is no water was found in the main wing tanks.

Any significance to that loose union?

A Ridiculous Suggestion.....
 

Suppose....just suppose...this accident has nothing whatsoever to do with the fuel or the pumps or the pipes.

Suppose that - for some reason - all that happened is that the autopilot was never disengaged - and stayed in control all the way to the ground.

Would the outcome have been any different?

Volume and mass
 

I was intrigued by the discrepancy between planned and actual arrival fuel.

Fuel tanks measure volume (albeit it is displayed as mass on the gauges).

Pumps supply volumes.

Engines produce power depending on the mass of fuel.

If the density of the fuel had changed during flight - it may be of sufficiently low density that even with the pumps operating at full volume - the mass of fuel presented to the engine was insufficient to produce the required power.

I am not a chemist, just a retired driver airframes, is it possible that a gas (air) could have been mixed with the fuel to lower the density during flight?

This may explain the apparent fuel discrepancy.

There are many variables that affect fuel usage on long sectors but, with several thousand hours on longhaul (DC10) aircraft, I was suprised to find the aircraft made that much fuel.

I will also say that I think everyone who criticises the crew have the benefit of hindsight and I think I wouldn't have done much different to them.

I have flown several times with the captain when he was a co-pilot and he is one of the better guys!!!

MiG 15,

again and again and again,

NO!

Any kind of EMI (deliberate or inadvertent) would have shown up as anomalies, and been recorded, and the AAIB report specifically states that

Danny, thank you for moving us to the Jet Blast, where this belongs.

Hydromechanical Fuel Control vs Electronic Engine Control
 

Phil

I get the impression you are uncertain as to the extent to which electrics / electronics play a part in this process.

So, if the spill valves are actuated to actually relieve the pressure between the fuel pump and the control valves (which were confirmed as being open) are the actuation messages recorded ? If so would their operation have been covered by any of the "satisfactories" in the AAIB report ? The Special Bulletin does say that the EEC worked correctly, but the listing of items tested does not include the spill valves.

My understanding is that the EEC is the Engine Electronic Control. Now if the system is anything like the tiny little gas turbine on my aircraft, the EEC has very little to do with the Fuel Control Unit, which is a hydromechanical device that functions entirely automatically and is self-powered, using fuel pressure derived from the shaft-driven HP pump as the motive power for its various servo functions. The EEC input to the FCU is limited to driving an electric actuator which moves the throttle valve. Sensors placed elsewhere in the engine tell the EEC whether the throttle valve movement has produced the desired power, and thus complete the negative feedback control loop. Meanwhile the FCU just gets on with pumping fuel in precise quantities to meet the EEC throttle valve demand, cope with environmental changes and protect the engine from catastrophic failure. Just to give you one example of the sort of thing that goes on inside the FCU, on my aircraft there is a little centrifugal device that, when N1 gets high enough, opens a valve that bleeds fuel from the system. This stops the engine exceeding the designers limiting N1. Of course this gadget never operates unless something else has gone wrong within the system. Another feed of pump output pressure, which is related to N1 (the faster the compressor turns the faster the pump goes) is used to operate the inlet guide vanes. The FCU, with all its internal complexity, is a beautiful bit of engineering and usually extraordinarily reliable. But I would be surprised if there is monitoring by the FDR / QAR or anything in the EEC of the detail of its internal operation.

777drivers / engineers, please correct if my analogy is false.

Sven

There is a whole section in the latest AAIB report about 'weather'.

I believe that the figure of 6,900kgs was what the crew had entered into the FMS so as to generate an "insufficient fuel" message. This figure does not include the contingency fuel or any extra fuel (a nice round figure of 79,000 implies that they probably took some extra). If everything went to plan then this contingency fuel, and most of the extra fuel would still be on board at LHR which explains the recorded figure of 10,500kgs at 1000ft.
Perhaps a 777 pilot could confirm that this is correct.

Airclues

What kind of simulator are we talking about here... a flight simulator or a Boeing design program? The flight simulator would have to model damage to specific wiring (I know Level D sims are accurate, but this is ridiculous).

I'm not familiar with the layout of the Trent plumbing, but surely the fuel-oil heat exchangers are not the coldest part of the fuel feed system? The fuel line goes through the engine strut (which it probably shares with a hot bleed air duct), then through a LP engine pump which would also raise the temperature of the fuel. If it was an internal blockage in the heat exchanger, then wouldn't it have activated the differential pressure switch on the exchanger???

From the atmosphere (the tanks have a slightly positive air pressure on them courtesy of NACA scoops)

The pumps themselves don't necessarily have to be immersed in fuel. The pumps are actually well above the floors of the tanks. However, the pickups for these pumps are closer to the tank bottoms. The fuel running though the pumps, provides cooling/lubrication.

Pressure-wise, CWT Fuel pumps pumping air will be no match for fuel pumps pumping fuel. The wing tanks will simply pump fuel into the manifold causing a backpressure easily strong enough to close the check (one way) valves on the CWT pump outlets. Therefore no air. If BOTH check valves failed on the CWT pumps and they were running.... anybody's guess.

The EICAS message is generated by fuel in the tanks (The low fuel warning is given by the tank totaliser system). However, the FMC also produces a message if the predicted fuel burn is below the pilot-entered minimum fuel value. There is also a message if the totaliser value varies by a certain amount from the FMC calculated level (but this is after engine start).

If the FMC was telling the pilots that the fuel at destination was below minimums, then he would surely double check the flight plan and his FMC programming and resolve the discrepancy. He certainly would NOT fool the FMC into not generating a message even if he could.

On the subject of lack of fuel in tanks.... Didn't the intial report say that one of the fuel tanks was intact? Surely a physical measurement of the remaining fuel would have been made.


Danny! Where are you???? :}

Mad (Flt) Scientist,
When a SB is issued by the manufacturer, it must go to someone in the company to decide the priority, in my company that was the Type Engineer. When reading it and it is presumed that they understand the aircraft systems, the person must have wondered why it was issued and the implication that there was a possibility that the spar valves wouldn't be closed when the fire handles are pulled. I would have also thought that the training course for the electrics on the 777 would have shown up the anomaly. Hindsight I know, but having been a TE I know how important it is to realise the implications of SB's etc on Operations.

Sounds awfully like you're trying to compare this to a GE engine, Sven?

The Rolls Royce Trent has a fuel pump and a Fuel Metering Unit. The pump is mechanically driven. The FMU is further downstream and simply meters whatever fuel comes out of the pump (according to EEC electronic input).... and delivers this metered fuel to the spray nozzles in the combustor. It would be too confusing to compare the two engine types ;)

Rgds.
NSEU

Ice chips. No doubt about it!

Ah, if BA ain't doing sump checks after fuelling at Shanghai maybe they orta... the fuel there is half water on good days... the icing lights are always coming on. Yes yes yes I know, the truck sample... I like the Chinese.

That 108 knots... um, if we had been hand flying, maybe the old girl would have just sort of naturally wandered on down at Vref or so, then a short transition in ground effect, we would have maybe gone another dozen metres.

I can well see that detail, the speed decay, being overlooked in the mental rushing about caused by the unexpected, inexplicable, sinister and potentially lethal turn of events. It is just that if it had all started say 500 feet higher then with that sort of speed bleed-off we might have had a really high sink rate going by the time we hit the ground.

I mention that only because it is the fashion these days to use the pilot regardless... the chief pilots / fleet managers etc all like it / recommend it / insist on it because they can't trust people to do things they mostly can't do themselves, namely hand-fly an approach on a sunny afternoon. So let's go back to hand flying when hand flying is an option, and keep our motor skills up, and certain piloting-related sub-systems at the back of our brains in the piloting loop.

Just a thought.

MiG 15

regarding cavitation, there are two phases, the formation of the bubbles, and their collapse.

They form at low pressure, and collapse at high pressure, so it is straightforward that the cavitation damage is on the outlet (high pressure) side, where the bubbles collapse.


Bernd

NSEU
I didn't mean by the simulator, I meant that the drill carried out by the pilots undergoing checks etc in the simulator, could have shown up the anomaly when the shut-down procedure was shared between the two.

Only if the simulation included severing of the wiring from the fire handle to the spar fuel valve. Otherwise the valve would have just closed when the fire handle was operated, and no anomaly would have shown.

Is it common in the simulator to simulate such specific damages?

Maybe, just maybe, Boeing should have objected to BA's checklist for unmodifed aircraft, but who knows.


Bernd

It is interesting that on this topic the latest report essentially repeats the information released earlier and gives no new details about the behaviour of the engines. Was the rpm steady, decreasing or increasing, were there any minor or major surges, did the engines behave identically?

This kind of details may give some clues as to the nature of the fault. For example, I'd expect the signs of a fuel flow impairment to be different than those of a fuel contamination.

My interpretation of this is that the AAIB will use the detail data to simulate any possible fault sources, but wants to do so in peace from any media speculations. And that's OK for me. We'll see the outcome sooner or later. :)

Fuel stratification - follow-up comments
 

Follow-up to my earlier posting (#195, page 10) on the possibility of fuel stratification.
Warning: I'm not crew nor engineer, just scientist.

Hand Solo, you wrote:
"I would think stratification would be unlikely given the aircrafts frequent manouvring and encounters with turbulence. The fuel would have been well and truly mixed up for most of the flight and certainly for the last 30 minutes."

I think you might be mistaken given that the g forces applied to the aircraft during manoeuvers are very low and the resulting force is typically normal to the wing surface (apologies, being very simplistic). The g forces would not be sufficient to disturb a stratification to cause it to break-up, i.e. the fuel becomes fully mixed.

Turbulence could cause a break-up of stratification if violent and with the imposed g forces being applied largely parallel to the wing surface. Moderate turbulence, i.e. the aircraft falls and rises rapidly, and applying g forces normal to the wing surface, might cause disturbance of stratified layers, even cause disruption, but I don't think it would take long for the full stratification to return in calm air. Plus, there are no reports of such events on this flight.

Bsieker, you wrote:
"It is my understanding of the water scavenge jet pumps that they would, while trying to keep the water emulsified in the fuel, would also mix the fuel, preventing stratification."

Thanks for that. Can anyone comment on how these pumps operate and to what 'depth' in the fuel their effect is propogated? Are they designed to fully mix fuel within the total volume of the tanks and so prevent stratification?

Bsieker also wrote:
"A minor slip: the imbalance was 300kg, not 30kg, which makes it still less likely that both engines would be fed the same undigestible type of stratum at roughly the same."

Sorry, a typo. When considering the imbalance, and its possible implications, I had in mind the 300Kg figure. A 5.8% weight imbalance equates to a few centimeters in depth of fuel (for example, 2.9 cm for a fuel depth of 0.5m). Of course, I don't know the dimensions of the fuel tank and so cannot calculate the exact difference in fuel depth between the two wing tanks but, I'd be surprised if the actual difference is greater than 2-3cm. This would possibly suggest that stratification, if it exists, in
both wing tanks would be essentially identical and might be fed into the tank outlet ports at the nearly the same time. Fascinating!

Further, I note now, having missed the text before, that the AAIB Special Bulletin 1/2008 concludes with:
"In addition, comprehensive examination and analysis is to be conducted on the entire aircraft and engine fuel system; including the modelling of fuel flows taking account of the environmental and aerodynamic effects."

Also, someone elsewhere on this thread mentioned that it had been previously thought that the mixing of fuels, provenanced from different sources, was not likely to cause any problems but was now being investigated.

And finally to politics, the elephant in the room: the fuel was sourced from Peking in an Olympic
year.

Regards, Tanimbar

How can one believe that ice obstruction can lead to such a result ?!
First we have 4 LP pumps in the tanks, meaning a quadruple simultaneous obstruction, reducing the fuel flow to a SIMILAR value on both sides (resulting in this just above flight idle thrust). This looks unrealistic to me.
In case of ice obstructions, the fuel restrictions wouldn't have been exactly similar, and thrust values would have been different on each engine.

"Further, I note now, having missed the text before, that the AAIB Special Bulletin 1/2008 concludes with:
"In addition, comprehensive examination and analysis is to be conducted on the entire aircraft and engine fuel system; including the modelling of fuel flows taking account of the environmental and aerodynamic effects.""

It seems to me this is the most likely way the AAIB will uncover the cause.

I dont know how much data they have recovered from the flight recorders, but must be enough for some detailed modeling. If the fuel flow is the common mode failure, it was asynchronous by 7 seconds between the engines. Is it not possible the initial response to a demand for thrust was using the fuel in the pipework and any other volumes after the restriction blockage.

So if we know the diameter of the pipework, the speed of the engines, caculated quantity of fuel used in the 3 seconds one engine and 7 seconds the other, cant one calculate the distance back along the fuel lines when the fuel stopped coming? Is there an offcentre pump, tank, point or junction that fits the data?

Just an idea

Quote (NSEU) :-

......... The Rolls Royce Trent has a fuel pump and a Fuel Metering Unit. The pump is mechanically driven. The FMU is further downstream and simply meters whatever fuel comes out of the pump (according to EEC electronic input).... and delivers this metered fuel to the spray nozzles in the combustor. ........

unquote

So, the Fuel Metering Unit should have its own data recorded - and hence shouldn't the AAIB Special Bulletin have comented on its staisfactory performance ? The bulletin comments that the correct commands were given by the EEC, but not what data was being recorded by either the FMU or spill valves as to what actions they were performing.

.

phil gollin.. did you miss this bit?
Regards
TP

No, it says the units worked when tested, it did not state that the meausurements recorded by those units (and the spill valves

For this to be an interference (or software) fault, with the information the AAIB have released now, it would have to be a fault which:

- reduced engine thrust to below that commanded
- simultaneously faked the data to the FDR etc. to show that everything right through to the fuel valves was responding to commands
- did all this without leaving any other trace of interference in the recorded data

To say that is not credible (for software bug or RFI) is putting it mildly.

Yes it could still have been software failure, or RFI, just as it could still have been aliens. I note the AAIB haven't explicitly ruled out alien involvement either...

I don't think mixing of different fuel grades or fuel from different sources should be an issue: refineries routinely blend to produce jet fuel in the first place and transfers of different grades between storage tanks are common place.

I'm not convinced by arguments based on fuel freezing - if the TAT was -37 and the fuel freeze point -57 how did it freeze? The freeze point is the appearance of crystals, not the point at which the fuel goes solid, and the margin was 20°C, not the 3°C required.

Stratification of products in storage tanks (caused by density differences) is well known in the liquid gas industry. However, stratification does not occur in ships, and this is attributed to motion. Note the design of the ships minimises sloshing at the liquid surface, it is the induced motion in the bulk fluid that prevents stratification occurring. I would suggest the continuous motion of an aircraft in flight would have a similar effect.

The odd bit in all this is that whatever happened affected both engines at virtually the same time, though the systems are supposedly completely independent. Is there some kind of (hidden) master- slave relationship between the engines buried in the software?

If I understand correctly, there are two pumps per tank - even with a master slave relationship between the engines this seems to rule out obstruction of the pump inlet screens by a foreign object or pump failure.

I think if the investigators knew the answer it would have leaked by now - which makes the whole affair ever more curious.

Indeed, but how about the fuel temperature in the pipework? could it, somewhere down the line, get below the temp measured in the coldest spot of the tank? could part of the fuel, during the descent towards LHR, when idling, begin to freeze and cause an (temporary) obstructon?

borgha
 

Fuel, leaving the Tankage, is under ever increasing pressure from LP pumps toward the HP pumps not to mention, it is moving, both actions would increase its resistance to solidification by introducing friction due to transit, and pressure, which lowers the Freezing Point. At the Low Side of the HP pumps, given a restriction upstream, the Pressure could be drastically reduced, raising the Freezing Point and potentiating fuel solidification or, more likely, transient emulsified water turning to ice which could have blocked critical pathways. This is my current theory, notwithstanding AAIB testing of several samples of collected Fuel.

Again, with a "homogeneous Fuel supply" isolation/separation of Fuel source might have not "protected" either powerplant. The restriction would be timed emphatically on temperatures, demand, and pressure, all of which may have been so close, the isolation would be "overridden" by parameters that could not be prevented by merely "separating" the Fuel sources. The five second lag actually supports this eventuation.

If my theory has weight, it might be better to Fuel the Engines differently.
One from a Wing the other from the Center, given that the location and other factors of the different tanks increase the heterogeneity of the Fuel source.

Thanks airfoilmod, I see your point, but what would be the fuel pressure/freezing point gradient? this might be of some importance on the LP side of any of the pumps. is there any expert out there in fluid dynamics?

PLEASE PLEASE can we stop talking about BA doing sump checks after refuelling.
I am sure 95% or more do not after refuelling, if l understand correctly the final fuel figure is worked out after final pax & cargo figures are known.
You would then have to wait 2 hours or more for the fuel to settle before the suspended water sinks to the sump drain points.:ugh:

Restriction
 

I think we may be going in the same direction. Additionally, it might be more harmful than not to be so focussed on precision in FADEC. I think the system probably worked perfectly so I use it only to suggest that in ETOPS it may actually be helpful to introduce anomalies of benign nature into the control system. If a Fuel source is Homogeneous, sourcing it in different locations in the A/C creates a tolerable "discontinuity" that may introduce a higher level of safety, rather than a lower, albeit predictable, one.

The authority I would suggest, Daniel Bernoulli, dead, but his work lives on.