3db

Quoting Green-dot
“I agree, the B777 is a very safe A/C. If PEDs are only built to a price and not to a spec, does that mean there is no quality control? To my knowledge, a spec defines what a device is made of and what it should comply with (e.i. items or details included in a description of requirements or plans of a design).”

In my experience, any PED (or domestic kit) more than around 10 years old is unlikely to have been built to any EMC standard. Less than 10 years, most kit sold in the EU will conform to an EMC requirement – usually sufficient for domestic use, not necessarily suitable for aviation use. E.g. A computer with bubble jet printer was installed in an un-pressurised light twin for some monitoring work. As the A/C climbed through FL100 the printer burst into flames. It transpired the spec for the printer bubble jet cartridge had not considered extreme altitude in un-pressurised A/C and would instantaneously combust with no ignition source at around FL100. This manufacturer was responsible and modified the design. To emphasise your point, the kit was made to a spec, but the spec did not take account of all operational conditions. Similarly with the kettle, it still boiled water fine and did not emit any RFI and was within spec. “Our” radio station emitted high levels of RF energy, again within spec. However, an unusual combination of circuit elements in the kettle decoded “our” radio signal. The space inside the kettle (old type, not a jug kettle) acted like a guitar sound box and amplified it so it was audible. Again, everything in spec, but the spec did not consider all possible uses of the kit.


Green-dot:
“I agree, perhaps EMI could have been a contributing factor. I am also convinced that the B777 is a very reliable aircraft design.

Therefore, EMI seems highly unlikely. That said, so is a dual engine rollback on a design such as the B777.”

I agree totally.


Green-dot:
“There is one physical difference between the valves i refer to and the B777 spar valves. The valve i refer to has the control relay (which was sensitive to EMI, hence the filter) integrated in the shutoff valve. In the B777 the control relay and spar valve are separate components. The spar valve control relays (together with the APU fuel shutoff valve relay) are grouped together on a single panel. However, i do not see any EMI filters on the B777 spar valve schematics. Perhaps not required due to other measures taken, making the system imune for such effects, i don't know. Location of other equipment in the vicinity of the relays or valves (or absence of it) is also a factor determining system sensitivity, which obviously varies between aircraft types.”

Also worth noting anything physically mounted on the engine is not in a faraday shield (like the fuselage is). It has 2 large holes at either end, 1 for air in and the other for exhaust out. I accept it will be shielded around the sides.


Green-dot:
“Another point that may be of interest is that (theoretically assuming EMI was a factor) if the spar valve control relays were switched unscheduled from "run" to "cutoff" for a certain period of time, the APU fuel shutoff valve control relay could have switched from "closed" to "open" during the same unscheduled event due to its close proximity to the former control relays. This may perhaps explain why the APU inlet door was observed in the open position?”

An electromagnetic pulse may have caused the logic states to reverse, at least until everything re-set itself – however, I would have expected that to be recorded in the FDR somewhere or some other anomaly would make the AAIB think “what caused that reading” on the FDR?

Green-dot:
“Since EMI has occurred sporadically in the past as i have explained, why couldn't it happen again on other aircraft designs?”

I think EMC issues are happening to A/C of the FBW type with some regularity now. Often it is not possible to reproduce the event and it goes unexplained, so “something” was changed as a precaution. As more A/C types become FLB and more PED’s are taken on board the chances of an adverse interaction increase – that is simple mathematics.

Maybe its like the micro-burst situation. Years ago it was not known about and any incident/accident was put down to pilot error. Now it is recognised and specifically looked for. Maybe its time for the AAIB or others to start looking for EMI issues in accident reports?

Green-dot:
“Summarizing, i have factual information that action has been taken to prevent EMI on fuel shutoff valve control relays in the past on another aircraft type. But regarding this B777 incident, evidence (of a source) pointing in that direction would have to be found in order to determine it as a possible cause for a dual engine rollback.”

Evidence of a source pointing in that direction is positive logic. How about negative logic? If, I repeat, if, the AAIB can’t find the positive source, then should they consider EMI (and other) sources? It is clear from the AAIB information so-far (in laymen’s terms) fuel that met the spec did not get to the engines when the computers requested more and it was available on-board. Usually, the fuel system works well, but on this occasion it did not perform as expected. In the absence of other sources being positively identified, I would say that is a classic case of EMI awaiting investigation.

Regards
3db

greekdalek

but not accelerating.
This is not fuel starvation.
What else would prevent engine spool up? An airflow problem?

infrequentflyer789

Not just the general public either - try this from the Times (once arguably a respected source of news) regarding the 777 fly-by sacking:

http://www.timesonline.co.uk/tol/new...cle3427696.ece

"Jumbo Jet" in the headline and the first para, in between the video, showing... clearly not a jumbo jet. I think I would have known this wasn't a jumbo at about age 10.

sispanys ria

I would be pleased to have some comments on this scenario:

Before reaching the end of the CT, some of its fuel might have been frozen due to the small quantity, inhibiting both boost pumps due to the LOW PRESS condition. If for some reason the crew didn't manually switch off the pumps, the frozen fuel from the main tank may have melt during the descent, causing the boost pumps to restart, feeding both engines with the slush from the center tank.

Chris Scott

Quote from PBL [Feb26/08:29]:
The 8086 was a 16-bit architecture in which all processing and all data channels were 8-bit. It means you had to do everything twice: once for the lower half and once for the upper half of the 16 bits. The 80186 had full 16-bit data paths and processing.
[Unquote]

Many thanks for the links re the "Viper" chip in 1990, which I'll do my best to read later.

My post was based on information (by an engineer that played leading roles in both 8086 and 80186 design) that, notwithstanding what you have said, the 80186 CPU design was "deeply derivative" of the 8086; the changes being mainly in the "integrated peripherals". Once the 80186 had been out for a couple of years, the chances of undocumented bugs was "pretty darn low".

This thread is so much better debated than its immediate predecessor – thank goodness – even though theories are still revolving almost entirely around semi-informed minutiae of H2O and the B777 fuel system. The discussion of FBW computer architecture is, on the face of it, entirely off-topic.

No one is suggesting that there was a flight-control problem on BA038. But discussions in this area – by the like of PBL, bsieker, 3db, Green-dot, and ve3id – have given us non-engineers a fascinating insight, and could yet read across to the engine fuel system, if not the aircraft fuel system.

Perhaps you guys should all meet with 3db in the UK! So far, the arguments have mainly related to what we ordinary mortals might think of as hidden potential anomalies in logic. Apart from a few anecdotes, little attention has been given to the possible effects of EMI...

Quote from 3db [Feb26/13:22]:
Evidence of a source pointing in that direction is positive logic. How about negative logic? If, I repeat, if, the AAIB can’t find the positive source, then should they consider EMI (and other) sources? It is clear from the AAIB information so-far (in laymen’s terms) fuel that met the spec did not get to the engines when the computers requested more and it was available on-board. Usually, the fuel system works well, but on this occasion it did not perform as expected. In the absence of other sources being positively identified, I would say that is a classic case of EMI awaiting investigation.
[Unquote]

Let's hope it has not been (awaiting).

Piper_Driver

PBL, You are mistaken about the 8086 architecture. The 8086 is a true 16 bit architecture with full 16 bit data paths. it was the first member of the Intel family of processors to come out. I think you are thinking about the 8088, a chip that came out shortly after the 8086 as a lower cost alternative. The 8088 is the one with the 8 bit data path, and is incidently the chip around which the first IBM PC's were built.

The 80186 is a chip that has the identical core processor architecture as the 8086. The 80186 variant added many of the typical peripherals used in a system design to the processor chip. These included items like an interrupt controller, etc.

As far as EMC issues are concerned, all aircraft systems were tested for immunity against EMC long before 10 years ago. The FAA's certification procedures required EMC checks as specifed in thier RTCA documents long before that.

infrequentflyer789

1. Crew failure to switch off the CT boost pumps would be in the data
2. Pump switch on would probably be recorded also (I am less certain on this one)
3. CT would be reading 900kg or more, instead of empty - so you would have to have CT fuel sensors also failing (reading 900kg over)
4. If the CT didn't empty into the wings (fuel scavenge failed), then you'd have 900kg less than expected in the wing tanks - in fact they had more than expected, suspiciously so according to some previous posters.


I agree that centre tank is tempting as the source of a common problem, but it should have been empty at this point in the flight and the data so far (at least as reported) indicates that it was empty.

archae86

I was one of the four 8086 design engineers, and aided our Haifa facility in making the 8088 derivative. I also briefly was the leader of the 80186 design team during part of the product definition phase, and knew well the main design engineers who carried out the active design phase.

The intended difference of the 8088 from the 8086 was only in the external bus to memory, which had only eight data bits, unlike the 16 of the 8086. However, all internal data paths (which is what a microprocessor designer means by data path, though perhaps a pilot uses the words differently) were identical in the two chips. A few slight variations in microcode were required to work correctly with the external bus difference.

It is true that the 80186 CPU internals were strongly derivative of the 8086, but not that they were identical. For one example, the need for an 80188 variant was in the plan from the beginning, so was accommodated with a few wires and a few transistors, rather than the 10% relayout required to make an 8088 out of an 8086. For another example, Jim Klovstad took my suggestion to add an "execute next micro-op during jump" capability, which gave some performance improvement at extremely low cost. (I got the idea from a paper on on of the early MIT LISP machines). There were other differences on this order, though it is correct to say that the major departure in the machine was the integrated peripherals.

I'm not a pilot.

Nipper2

Sorry guys, but I had to comment. The above post must rank near the top for 'thread creep in one step'.

By the way, as an engineering enthusiast and a humble PPL, I'm enjoying the guided tour of the 777 fuel system.

Nipper2

Sorry guys, but I had to comment. The above post must rank near the top for 'thread creep in one step'.

By the way, as an engineering enthusiast and a humble PPL, I'm enjoying the guided tour of the 777 fuel system.

PBL

Piper Driver is right that I was confusing the 8086 with the 8088. Sorry about that. The definitive post from archae86 puts the record straight.

As for Nipper2's suggestion of "thread creep", I am not yet convinced that bits and bytes are not relevant to this accident.

PBL

snowfalcon2

Thread creep or not, I feel rather humble to find myself in an aviation forum with guys such as archae86. I wonder if that team had any idea at the time about how much their work would change the world. Kind of similar pioneer teams to those that worked for Glenn Curtiss and Geoffrey de Havilland when aviation developed from an adventure to something useful.
OK, OT, I know. Next post after next info from AAIB

Bis47

Hello everybody.

I'm new to this forum. French my native language. No modern jet experience. Some airmanship (I hope). Be clement we me ...

I think that we are perhaps misleaded by the assumption that the problem occured "almost simultneously" to both engines, when they failed to spool up.

General scenario :

More likely, the flow fuel restriction problem was there well before the demand for higher engine output occured on short final. Maybe the fuel flow started to be restricted (cold related - any kind fuel of contamination, make you choice : ice, wax, bioslush ...) much earlier in the descent. The restriction built up slowly, somewhere between the LP pump in the wing fuel tank and the engine HP fuel pump. In a very cold spot? In a bend? Idlle fuel flow and very low temps at TOD might be a factor. Everything almost symetric (same cause, same effects). The restriction went unnoticed as long as only minimum fuel flow was required. Then ... when more fuel was required, i-e when the valves downstream the HP pump opened, the engines initialy accellerated, burning all the available fuel in the line, until the fuel flow became regulated by the fuel restriction upstream.

So : almost simultaneous evidence, but a relatively long standing cause.

Ok, this is only a theory, I'm not fluent with the 777 fuel system (I did learn a lot reading this thread in full). But this theory does free us from the double "sudden" failure mode : the failure were not sudden, the problem was a "progressing" one, going unnoticed for a while.

My two cents ...

PBL

Yes, they knew, very well. But they didn't know if would turn out to be a few years of glory (c.f., the Z80), or dominance that would last a quarter century.

PBL

wilyflier

High Angle of attack or highspeed stall has been known to cause engine surge, lack of acceleration or even blow out
...108 knots = high AofA
But 108 knots did not occur till well after the problem had started,
Unless perhaps there WAS sufficient wind shear coupled with a few knots slow to start with.
Stalled engines trying to spin up would make a lot of noise.Have you ever had or heard a slow start? How long can it take to reach a stable self sustaining speed?
If something like this did occur,engines being limited by high jpts on the limit should show on the FDR and QAR ? as well as airspeeds

777fly

From the latest AAIB update, it is clear that the problem was either fuel starvation or fuel aeration. As far as I can see, there has been plenty of speculation about fuel stavation/ icing but little about fuel aeration ( I have not followed the whole thread)
So, where can significant amounts of air get into the fuel? Apart from the tank vents and surge tank ( which aerate the tank system normally and are above the tank fuel level, anyway) the only other way in which the fuel system can be open to the air is via the fuel jettison nozzle valves. If these were to somehow be signalled to open, there would be a large diameter pipe, open to air, which is routed via the override/jettison pumps into the same area as the wing tank pumps. It might then be possible that the wing tank pumps would ingest both air and fuel, leading to fuel starvation in both engines.
The time difference in the power down of each engine could be explainable by differing amounts of residual fuel in the jettison line to each wingtip.
This speculation could explain the simultaneous failure in two 'independant' systems. However the question would have to be: what could signal the jettison nozzle valves to open independently of the jettison master switch. Maybe EMI or a software glitch? Also, I have no idea if the override/jettison pumps are an axial type without any kind of reverse flow check valve, which would allow air to pass through.

Milt

Fuel Aeration ??

The fuel system designers will have gone to a great deal of effort to prevent air entering the systems. Cannot imagine any possible entry point which would not have been leaking under normal fuel pressured operating conditions unless such a leak/s were contained within a fuel tank and hence undiscovered. Any air bubbles in the fuel at the input to the high pressure pumps could be expected to promote cavitation which will reduce pump output but only remotely possible below that being demanded by the engines at low thrust.

The AAIB is probably being simplistic in calling fuel vapourisation or partial vacuum at the HP pump inlets under cavitation conditions "aeration". Such conditions almost simultaneously in two seperate sytems is a very long shot.

I guess somone is rapidly assembling an instrumented test rig.

What a fascinating problem for our armchair Investigation Board to dwell upon and many pilots and engineers will benefit from their enhanced knowledge of jet engine fuel systems and fuel behaviour.

Missing from my know how and experience as a TP is the propensity of cold jet fuel to form vapour (not air bubbles) at the HP pump inlets at pressures close to a vacuum and can extensive vapourisation occur instantly to both damage the pumps and substantially reduce pump outputs?

Jetdoc

Both the boost pumps and ovrd/jett pumps are designed to provide vapour free fuel to the engines. The air is returned to the tank and the fuel goes to the engine. The outlets of the pumps which all connect to the engine feed manifold are have check valves which must be forced open by the fuel. When the center tank pumps are supplying fuel, the main tank outlet check valves are held closed by the higher output of the ctr tank pumps. This prevents any fuel from flowing through the main tank pumps. When the center tank pumps have emptied their tank, the main tanks are now pushing their check valves open and forcing the center tank check valves closed thus preventing any fuel from feeding back to through those pumps.
If none of the pumps are working, the fuel bypass check valves are pulled open by the suction effect from the engine driven pump and will allow fuel to pass to the engine.
The crew would actually have a warning message alerting them to turn the center pumps off when the tank has emptied and I doubt that they would ignore that.
Possibly the only thing that could have happened is that the aircraft attitude became so extreme at the low speed that the aircraft was flying that all 4 of the main tank pumps and the 2 bypass valves pickups became uncovered and fuel flow to the engines was interrupted.

NSEU

The fuel tanks NEED a slightly positive air pressure to help stop the tank collapsing when fuel is being pumped out of the tanks. The vent system provides this positive pressure.

I don't see how a jettison system can positively pressurise/aerate the fuel in the tanks. The wing nozzles face aft (no scooping effect)

How about the opposite of positive pressure in the tanks? With full fuel in the wings, maybe the float valves in the vent system iced up in the closed position (condensation + cold on the hinges and seals). Float valves have been know to stick before (but usually you only notice it when they stick open... dumping fuel in the surge tank).

Maybe the fuel being pumped out of the tanks to the engines gradually created a partial vacuum, and the fuel could not be fed to the engines fast enough at high thrust settings... hence the cavitation.
Looking at the diagrams in my manuals, it looks like all three tanks are kept at the same pressure.

Maybe just as crazy as the other theories, but I've been told it can cause problems in car fuel tanks

Looking forward to being shot down (I've probably overlooked a negative pressure relief valve in the system)

Regards
NSEU

CAAAD

I find it impossible to contribute to this excellent debate without visibility of the fuel system. Are diagrams available?
I'm sure that full illustrations, chapter and verse will be provided in the Final Report, but that's of no help now.
In the meantime it is impossible to comprehend the layout with written descriptions. Which leads to endless misapprehensions.

Where can we access the schematic?