Dear Oluf,

Was what you describe here on the ground while taxying or in flight?

I remember an accident which was caused by moderate icing, engine vibration and consequent failure of the icing panels on both engines.
Quoting from the report:

"Conduct of flight:
Simultaneously with the icing alert the crew switched the engine ant-icing and airframe anti-icing systems on. The crew attributed the increased vibrations indicated for the RH engine, after reduction of airspeed and entering cloud (icing conditions) in FL100, to ice formation on the fan. The airplane flew, for approx. 6 minutes, with reduced engine thrust under moderate icing conditions which, in all likelyhood, led to icing of the fan blades of both engines.

After a prolonged time under moderate icing conditions and low engine thrust, ice developed on the rotors of the low pressure compressors of both engines. During descent from FL90 to 3,500 ft the engines were running smoothly in a low thrust range and all engine indications were in the normal range. The crew could not recognise the reduced performance of the two engines. Thus the PIC did not hesitate to operate the airbrakes for some time, in order to reduce the airspeed to such a degree that the flaps could be extended. Once flaps and landing gear were extended it became apparent that the engines developed insufficient thrust. There was no malfunction indication at that time because in the flight warning computer monitoring of the N1 to EPR ratio is not intended. EPR indications showed, however, that dispite an RPM increase the engines developed insufficient thrust.

Cause:
The bonded joints of the ice impact panels on both engines failed due to strains caused by ice-induced vibration of the engines and by ice which had detached from the rotors of the low pressure compressor. The loose ice impact panels became trapped in front of the outlet guide vanes of the low pressure compressor and affected the airflow in the by-pass duct in such a way that the engines only produced low thrust."

Now compared to BA38, in the above mentioned case RPM (and EGT) increased but the engines developed insufficient thrust due to air flow restrictions. Increase in RPM and EGT implies that the fuel flow was getting to the engines.

In case of BA38, thrust increased initially, then rolled back even with fully opened fuel metering valves. After the rollback, no mention of increased RPM or EGT, indicating this is not very likely to be an icing problem but that the required fuel flow was not getting to the engines. What is also missing in case of BA38 is that no mention has been made of vibration due to icing or shedding of ice.

For the full report, here is the link (including FDR data, please note RPM, EPR and EGT parameters):

http://www.bfu-web.de/nn_53140/EN/Pu...MUC_Fokker.pdf


Green-dot

Question for Oluf
 

A question from a nonpilot trying to understand things.

Oluf, I can see how ice could cause the symptoms you describe above on your MD-80, but I am at a loss to see how core ice could lead to reduction of fuel flow (and we know it was reduction of fuel flow on the BA flight)

Can you explain it to me please?

Turbine engine acceleration is a pretty complex process. Simply dumping in more fuel does not assure the desired results.

The fuel controller (MEC, FADEC, whatever...) has been tailored to produce best acceleration on a reasonably healthy engine - it adds a measured amount of fuel to start the accel process, then as RPM and pressures build, a bit more fuel is added, and the cycle repeats... This is to prevent stalls, surges, overtemp, etc.

However - an ice accumulation effectively "redesigns" the aerodynamics so that the amount of fuel the controller thinks correct is now too much! Ergo, stall, lack of acceleration, etc.

A wise pilot can "milk' the throttle through this stage and not demand so much fuel - he's effectively "redesigning" the accel fuel schedule to match the new compressor aerodynamics. Once a high enough RPM is reached, the ice is either melted by heat of compression, or broken loose by centrifugal force, and things return to normal (unless the impact of ice chunks on downstream blades is significant...)

OK?

BA38s engines "Hesitated"
 

Dear barit1 and other PPruNers,

Thanks for your comments, since I am danish, you will have to excuse some of my word selections, but we seems to agree, that ice is the only "aerodynamic redesign" that melts away.

When I had my experiences with the MD-80 engines, it was on the ground, trying to get T/O power, after some taxiing in foggy weather. And that was my only experience of that kind in a 27 month period.

This was 19 years ago, maybe newer engines are better in trying to protect themselves from stalls, vibrations and ice damage?

Maybe the BA38 engines just never made it to the stall and vibration phase?

Oluf

Oluf please stop banging on about core icing
 

Warning: I'm non-professional; not crew, not engineer - just scientist guest and thanks.

Dear Oluf,
Please stop insisting in this thread that core icing played a role in the crash of G-YMMM.

The AAIB Special bulletin states:
The crucial words are "reduction in fuel flow" and for both engines.

Also from the AAIB, - but fuel flow did not increase.

The fuel did not flow in sufficient quantities!

Therefore, and unless I've failed to understand how these fuel systems operate, core icing cannot have contributed to the cause of BA038's crash.

Core icing is an interesting subject, probably deserving a thread of its own, but please, PLEASE, stop distracting those of us who are interested in understanding G-YMMM's crash.

Regards, Tanimbar

PS. Apologies to those who have already made these points.

Oh I wish that you hadn't left that so simple and thus feed Oluf with his ever so simple explanations.

It's true that with a significant amount of aerodynamic change (ice accretion) the fuel metering will not match the performance, but the signature is much more likely to be a mismatch in rotor spool RPMs N1, N2, N3 followed by a stall/surge. All of which are sure to be recorded on the various recorders, including non-volatile memory in the FADECS.

Add to this the extereme rareity of such non-mechanical damage events for in-flight scenarios (ground fog at ground idle is not part of this discussion).

roundy roundy.
 

This thread is going in circles, everyone is arguing their pet theory, information, DIS-information, perhaps a temporary lock on it until we ALL get some proper news from the investigators might be an idea?????

GR

More on fuel
 

A reference was made in an earlier post to (fuel) chemists seemingly having disappeared from the thread. Well this chemist has not entirely disappeared and has continued to read all the posts on this admirable thread.

I have added nothing, having nothing new to say but a reiteration may be useful – the mods will decide.

My hypothesis (- a careful use of English) remains the same i.e. that either a stratification of the fuel as a function of relative density and / or water scavenging additives such as the friends of epoxy ethane et al, may have fed either low Relative Molecular Mass fuel components or scavenged water at sufficiently high concentrations when increased thrust was commanded by the FADECs, to induce cavitation sufficient to reduce effective flow rates to the donkeys at circa 700 feet.

The hypothesis sort of fits the incomplete I am sure evidence currently available in public domain but only up to a point is the problem and I am very happy to wait for the AAIB final report. The time scale for their reportage is entirely consistent with their SOPs and I have the very highest level of respect for these chaps.

The carefully preserved circa 9 tonnes of fuel are now of course very well mixed and (just one) problem with my hypothesis is the sheer complexity of replicating the conditions. As someone who calls himself an experimental chemist I blanch at the thought of the experimental need to (i) cold soak the necessary tonnes of fresh samples of Chinese fuel (ii) putting said fuel on top of the right amount of LHR fuel left in a 777 cold soaked main wing tank (iii) using the same rigs and (iv) then replicating in full the thermal history of the fuel back to LHR as far as it is known whilst (v) pumping from this tank in line with the data and (vi) constantly measuring the Heat of Combustion of the pumped fuel.

We need to wait.:ok:

Both at once
 

Bsieker
Thats just how compressor stalls can behave.
...Also,ice by itself can upset the airflow,without necessarily shedding or causing mechanical damage
I do remember a reference to passenger quotations of grinding noises on the approach , but that may of course have been the "landing"
Lom Paseo
,,,,Extreme rarity doesnt rule it out

It may be that compressor stalls can manifest themselves in a way that the engine first accelerates and then rolls back. And also that it may happen on both engines at nearly the same time.

But that is not the point.

The AAIB specifically talks about:

(a) fuel flow reducing, and

(b) metering valve opening to fully open, in response to the low fuel flow.

(Remember that fuel flow is measured between the fuel metering unit and the HP filter, and is also recorded.)

If a compressor stall (caused by core icing/fan icing/cowl icing/whatever) can cause low fuel flow despite high flow being scheduled by opening the metering valve, I'd love to be enlightened about a possible mechanism.


Bernd

And icing would have affected other aircraft on approach to LHR at that time, including many 777-200ERs, so why weren't they all falling out of the sky on the same flight profile?

BA38s engines "Hesitated"
 

Dear FullWings,

All other flights probably flew "straight and level" in a holding, or at some given flight level, at least three times, thus coming above flight idle for several minutes.

It has been so since 1973, when I began flying into the greater London Area.

Oluf

Dear Oluf,

How BA38 performed the holding has not (yet) been published in detail.

But suppose your theory of core icing had occurred:

- EPR, N1 and EGT are the primary engine indications.
- On the RR Trent 800 series, EPR is the thrust setting parameter (with EECs in normal mode).
- The thrust management function of AIMS calculates the reference thrust based on existing ambient conditions and the particular thrust reference mode selected by the pilot on the CDU. The pilot can override a thrust mode by manually advancing the thrust levers (which he did when the engines started to roll back).

Normally, with the thrust levers manually at the full forward position, thrust limit protection allows full rated thrust without exceeding the maximum thrust limits. EECs (in normal mode) provide EPR limit protection and also N1, N2 and N3 RPM overspeed protection.

In case of core icing and/or fan icing (and following your theory, no restriction in fuel flow), advancing the thrust levers fully forward, the EECs (in normal mode) would attempt to accelerate the engines from actual EPR to commanded EPR. With the engines contaminated by ice, the commanded EPR would likely not be reached due to restricted airflow through the engines. The EECs would allow N1, N2 and N3 to accelerate (again, with unrestricted fuel flow available) to attempt reaching the commanded EPR until an RPM reaches an overspeed protection limit. EGT would also rise considerably. Since EPR, N1 and EGT are primary indications, any abnormal indications on the EICAS display would have been noticed by the crew, certainly if an RPM limit would be reached which results in an EICAS advisory message. In this situation, advancing the trust levers will not increase thrust.

If this had occurred, the following would have been recorded on the FDR:

-Low EPR (low engine thrust);
-increased fuel flow;
-Increased RPM;
-Increased EGT.

Of course, this may have resulted in engine thrust stagnation or roll backs but the fuel flow increase would certainly have been observed, if not directly by an FDR parameter, indirectly by RPM and EGT increased parameters against a low EPR parameter.

If it had occurred that way, this investigation would by now have been in an advanced phase with the AAIB completing a final report soon.

However, according to the AAIB reports this did not occur. The AAIB, in the latest report released in February, was focussing on damaged engine HP pumps and the engine fuel system to answer why fuel flow to the engines deminished while the fuel metering valves were commanded fully open.


Regards,
Green-dot

Green-dot - All I can say to you is: please re-read my post #1028.

Simply adding fuel may not produce the desired acceleration, if there is something abnormal going on in the core flowpath. You cannot thus assume a more fuel >> more RPM scenario. :uhoh:

That said, please note that I am not taking a position pro or con icing in the BA038 case. I'm simply pointing that out as one possibility.

barit1
 

I certainly agree that "simply adding Fuel" may not produce the desired result. In Normal circumstances, however, it is basically that simple.

Lomapaseo describes turbojet basics as: Spin, spark, spray. I prefer one I heard from a flight instructor "Suck,Squeeze,Bang,Blow".

Both oversimplified, but to paraphrase your remarks, "Simply adding No Fuel will certainly not attain the desired result". I think you are correct in saying the report does not say, Fuel was/was not supplied in response to commanded level. It is a good catch, and deserves noting.

I vacillate between fascination and impatience regarding AAIB progress.
IMHO all discussion is valuable; I am acquiring knowledge in many fields free of charge, and I appreciate the accessibility of the website. This is a humbling and valuable exercise, to me.

airfoilmod
Surely "Suck,Squeeze,Bang,Blow" refers to the Otto cycle of the four-stroke reciprocating internal combustion engine - AFAIK these were not fitted....

Green-dot's post is very relevant - if core icing were present the engines would have reacted completely differently.
TP

- well, I had a Harrier Pegasus do that once:)

The report notes that the fuel pumps show signs of cavitation damage.
How would engine core icing cause fuel pump cavitation?

cavitation; please explain
 

As a know-nothing, I hesitate to venture once again into this astonishingly complex discussion. I can follow the arguments in general terms but do not pretend any more than that. But I previously posted on my own experience of cavitation effects in a centrifugal pump (in a mining context) and would appreciate a layman's-level explanation.

I have seen cavitation cause a 6" Ajax water pump to repeatedly fail to deliver. The cause was restricted flow at the foot-valve which resulted in air coming out of solution and gradually accumulating in the pump casing until sufficient volume was attained to cause cavitation and loss of delivery pressure.

I have seen many references to cavitation in this thread; it seems to be a generally accepted factor in the events leading up to the accident. In my ignorance I keep thinking that if there was cavitation, where did the air/gas come from if not out of solution? And if there was cavitation in a pump, it's hardly surprising that fuel ceased to be delivered (if indeed it did cease...)
When I experienced this phenomenon, there was not a gradual drop in pressure, it was more like a sudden failure (at the head of about 200 ft where the pump delivered).

Could someone please suggest why the cavitation isn't a sufficient explanation in itself? I ask this not as a contentious "stir", merely out of puzzlement. In all the posts I recall cavitation is mentioned in passing as if it is just symptomatic rather than possibly critical - which may well be the case; I simply don't understand why.

Apologies once more for the layman's question.

I agree with Tanimbar and others who from time to time try and get the thread back on topic.

The FADEC on modern gas turbines has logic to detect comprssor stall condition.
The FADEC on most gas turbines of that generation has logic to accommodate compressor stall (fuel dipping, blead valve scheduled open etc.).
The a/c QAR catures sufficient data to enable engineers to determine if a compressor stall has occurred.
The a/c QAR and EEC both capture sufficient data for the engineers to determine if the fuel flow into the engine fuel system is less than has been demanded.
The report implies that the condition of the fuel pumps supports the theory that fuel flow from the tanks is lower than intended (suction head too low).
The report also states that despite the condition, the pumps were able to deliver full fuel flow in the correct conditions.
The official report makes no mention of core compressor icing, stall nor an reports of cor compressor damage.

I think before posting pet theories one should familiarise one's self with the facts first. Lets allow the authorities to get on with figuring why the fuel flow from the tanks was lower than it should have been despite sufficient fuel in the tanks. If anyone has new suggestions on this issue great! I think if it was easy they would have told us by now.....:ok:

If anyone wants to talk about "modified aero dynamics" - new topic?

If people want to talk about "turbine compression"....another forum :ugh:

clarification
 

Just in case it isn't clear, I have no idea what caused this accident, other than ideas stimulated by this discussion. I am completely unqualified to have an opinion, still less a "theory"!

But I'd be interested in hearing someone explain cavitation in this context.
BTW I just spoke to a former colleague who was present at the same event I described and his recollection is that we had repeated reduction in pressure just before we lost it entirely each time this cavitation occurred.

I take the point that there has been a statement that the fuel flow was not impacted in this case. Can someone suitably qualified explain if a cavitating pump can sucessfully deliver at the correct rate. This doesn't imply that I think it can't!

Cavitation, again.
 

Not so. Although the AAIB report mentions cavitation signs on the HP pump outlet, it is by no means clear that this damage was caused by an event immediately preceding this accident.

The cavities can be either air or fuel vapors. The fact that cavitation damage was apparent shows that at least some of the bubbles collapsed again on the outlet, causing abrasion. The manufacturer of the pump and experts will know whether or not it was able to sustain sufficient flow under cavitating conditions, all other things being equal. This information does not appear to be publicly available.

Cavitation does not happen under normal conditions in a well-designed and tried-and-tested fuel delivery system with (as the AAIB said) on-spec fuel. It has to be caused by something, which may be upstream flow restriction or excessive fuel aeration. Cavitation may then exacerbate that problem, but will not be a root cause.


Bernd

I am sure the AAIB checked other 777 pumps to see if they had signs of cavitation, especially as the manufacturer stated that the pump (even with the cavitation damage/wear) was capable of meeting its design specification.

That information would establish if any deterioration in the pumps performance due cavitation was caused on the flight in question or could have been caused on a previous flight.

Biesieker
 

It may not be clear to some that cavitation was caused by events just prior to the incident, but that is exactly what Boeing Captain Carbaugh suggested and AAIB is looking for a "restriction in the Fuel system upstream as a cause of the cavitation." See post #662 and the flightglobal.com article (3-14-08) quoting Dave Carbaugh.

It may be a linguistics issue, saying cavitation couldn't be the "cause of Fuel Starvation", but it certainly would show a "restriction upstream" as evidenced by damage to pump lobes. Said restriction might be the "proximate cause of the lack of Fuel". Carbaugh said the cavitation might have been caused BY restriction, making the pump damage a RESULT of restriction. (He also mentioned ice in fuel and made reference to "temperature").

I agree that it is a likely scenario, and it fits the rest of the scenario just fine, an upstream fuel flow restriction, causing both pump cavitation and restricted flow (the former possibly exacerbating the latter). All I was saying is that we (in the uninformed public) do not (yet) know at what point in time the cavitation occurred that caused the visible signs. This is a matter of level of confidence in the factors in our causal reasoning. If I were to make a Why-Because-Graph of this, pump cavitation would be marked as an "Assumption".

Note that even the flightglobal article talks about a "possible" fuel flow restriction, and that the cavitation signs "might indicate" such a restriction. The assessment that the cavitation occurred "not long before the impact" has not yet been publicly supported by facts. It is so far just an assumption, useful for creating test scenarios, but it should not be confused with an established fact.

Yes, I almost completely agree. Except that it's not a linguistic issue, but an issue of causality.

My previous comments were in response to skridlov asking (linguistically unambiguous) why cavitation could not be a sufficient cause in itself.


Bernd

From experience observing complex fluids in experimental systems, I believe cold fuel in the lines leading to the pumps might have a tendency to 'shear' and locally develop some very abnormal physical properties if the supply flow is disrupted upstream and the downstream pumps are still doing their best to push max fuel toward the engines.

Any gasses dissolved in the fuel would increase in volume and come out of solution. Some volatile fractions of the fuel itself might gassify. Whipped around by the under-loaded and possibly over-heating pump impeller blades, the gas plus liquid would flow forward as a frothy foam, still burn-capable but much lower in density and fuel energy than normal.

Cavitation damage is hard evidence
 

The cavitation damage may or may not be relevant to this incident, but it is incontrovertible evidence that at some stage in its life that pump has had to suck harder than it should. Or put another way, at some stage in its life it has found the fuel supply system under performing.

An analysis of the pump population will indicate how rare or common it is for the fuel supply to under perform.

We know that for it to critically under perform is a rare event. To estimate the probability of this rare critical under performance, it would be useful to know how much non-critical under performance occurs, and how it correlates with aircraft history.

Then we might be able to determine how and why it happens, and how often the under performance becomes 'nearly critical'.

arcniz makes an excellent point re cavitation:
Such conditions may well exist at the HP pump inlet, but the pump discharge must still be at a much higher pressure to overcome the burner (air) pressure. Otherwise we would see backflow, starving the burner and an immediate flameout. I'd don't think that's evident on the DFDR.

Seems like a great spot for a backflow check valve, wot? Of course absence of fuel flow for any duration will tend to cause flameout also, but a pulsing flow mode might be the natural result in a fuel choking-foaming situation if some type of check valve is present inline at the outlet side of the pumps. Isn't it likely the igniters were full on at that late stage in the descent?

Even if the foamy froth I described were re-compressed to a higher pressure, the new fluid likely would still have a notably lower density and some different physical properties due to the disruption of the original fuel fluid structure and consequent rejiggering of the intramolecular van der Waal's forces in the resulting froth. The aereated fuel would almost certainly have a lower density than the original JPxx for any given flow pressure.

The phenomenon is rather like those products one purchases at the store which once removed will never again fit into the original box.

Has anyone heard of RR fuel pumps being pulled for strip downs and comparison to ascertain whether this cavitation damage is a one-off or common occurrence?

Everybody is still trying to outguess the AAIB. Amongst the more bizarre theories is frothing fuel! More like frothing at the mouth, give us a break. Put the lock on this thread.:rolleyes:

Not really, as I trust the AAIB are not guessing much at this point... maybe weighing up, but not guessing as in pizzing in de vind!

Was it pilot error?

Whats goin on, the aaib must know at this stage if it was a mechanical problem.

Always a possibility. However, I can't see a realistic way of duplicating what happened to BA38, even if I had to. How could you cause both engines to 'hang' above flight idle when you're sitting on the flight deck?

Actually HotDog, I think you will find that information was offered by someone who understands more about the properties of pumped fuels at various temparatures and pressures than most of us know about hot dinners or dogs :suspect:

There's a lot you can tell from a laboratory bench that you couldn't possibly guess at from any seat in the cigar tube.

Engine Refresher

barit1 raises an interesting point concerning the pressure to be overcome by the HP engine fuel pumps at the burner nozzles at low thrust. This could be expected to be the addition of the engine core compressor output and the back pressure from the turbine/s which would not be inconsiderable. My guess would be about 100 to 200 psi.

Does anyone have a typical pressure plot through the engine core at low thrust or the range of air pressures at the burner nozzles over the thrust range at low altitude?

BA038
 

BA038

Milt, I do not know the answer to your question, but the following may be of help.

The final delivery path of fuel to a gas generator is designed to be more than capable of overcoming pressure variations in the burner. The only way (that I know of) that the final delivery can be compromised is by a defect or number of defects further up (or is it down) the line. According to the AAIB preliminary report everything between the tanks and the gas generators have been individually tested and pass muster, although cavitation damage on the impellors was noted. This might suggest a restriction in fuel flow, since pumps are carefully designed not to cavitate in the normal and extreme operating parameters that can reasonably be expected.

Hence the suggestion in my only other post (28 Feb #500) that it might be useful to run a computational fluid dynamics programme (CFD) to see where the problem might lie, which may be valuable in any subsequent physical simulation. A CFD programme may not however be able take account adequately of the effect of high frequency sound, and there is a lot of that about. High frequency sound (including structure-borne resonance) can cause disturbance in fluids, including cavitation, which may not be evident after the event, except when the entrained disturbed fluid meets a pump and leaves its mark.

Is this the area in which the AAIB should be looking? Maybe. Why hasn’t it occurred before in identical aircraft? Probably because of the infinite number of temperature/pressure/sound frequency/fuel velocity combinations.

As an aside, the incident happened about 15 minutes before I was due to land (as a passenger) but I only spent another 45 minutes aloft. Very inconsiderate all the same.
rgds

You don't make it clear who or what was 'inconsiderate'. Did the BA038 fuel not take your needs into consideration, or was it the engines themselves?

Ba 038
 

For clarity, I was not (not) on BA038, just following (as pax) from HKG. The 'inconsiderate' bit on which I should have given a fuller explanation was knowing of the incident just after it occurred (announced by pilot) but not knowing the fate of those involved until after we landed. I think we were either first or second to land after the incident/accident.

I suspect everyone was rather busy working out what to do with the large number of planes delayed and where they were going to go. Watching BBC News during the event I was quite impressed how fast the airport handled such a major incident. I suspect there wasn't spare radio bandwith to give updates on the health of the passengers in those 45 mins. It's busy enough on a good day. Which airport did they divert you to?