johngreen

From the information supplied from this forum and some additional research, I understand the relevant components of the 777 fuel system to be; two AC powered centrifugal booster pumps mounted adjacent on the rear of each wing tank which feed into a common manifold. On the inboard side this manifold extends to become the cross feed (which has two normally closed parallel valves mounted only on the left side of the aircraft) and outboard becomes the fuel delivery pipework which passes through the spar valve, continues to the oil heat exchanger in the fan case and then on to the engine driven HP pump which is fitted with a recirculating bypass to control over pressure.
http://www.eatonaerospaceltd.com/fuel/Resources/021_2_Type9106boostpump777.pdfgives some details of the type of booster pump fitted.
I assume (dangerous?) that this unit is very similar to those fitted to some 747s of which I have better information. In order for it to be possible to service the pump without draining the adjacent fuel tank, a sprung loaded flap valve is located at the inlet of the pump case which is mechanically held fully open by the physical presence of the pump. On the outlet side, a similar flap valve – ie a disc with a hinge on one side perpendicular to the flow – is mounted. This is a typical non return valve which has a soft seat and a weak spring to ensure the valve will close when there is no forward flow and as far as I am aware no means of damping the movement.

Considering just the section of this system between the boosters and HP pump, any physical movement which even slightly changes the shape or concentricity of any part of the pipework also changes the internal volume of the pipe and hence introduces a degree of elasticity.
In the unwanted resonant state, a small surge of fuel progresses towards the engine where it is baulked by the HP pump which by its intended design strictly meters the possible flow. The excess energy contained in this moving column of fuel causes an increase in the local pressure creating a shock wave causing some slight expansion or lateral movement in the pipework which upon contraction returns a reflected shock wave back towards the booster pumps.
The non return valves are held open in proportion to the flow of fuel hence at engine idle the valves will be floating close to their seats. The shock wave hitting the back side of the valves momentarily forces them against the fuel flow direction and towards their seats. This causes a restriction to the flow creating a disparity of a depression on the engine side of the valve and an increase in pressure on the booster side. Once the shock wave has passed, this combined depression and pressure causes the valves to move rapidly to a more open position… causing a surge of fuel to progress towards the engine….

If the many and complex physical characteristics that are an influence allow the continuation of this feed back pattern, this system will continue to oscillate – ie resonate - until something either breaks the loop or as can also happen, exacerbate it although due to the enormous interdependent complexities involved, predetermining what will cause what response may be impossible.
It can be seen that on an increase in demand the HP pump allowing more fuel to pass might reduce the effect of the pressure build up and reflected shockwave and hence damp the effect. Alternately, the greater demand could result in the difference of depression and pressure across the valves causing them to open faster, resulting in an enlarged surge and a growth in the effect.
In the low demand situation with just a small amount of oscillation in the fuel column, the supply seen beyond the HP pump could remain quite adequate to cause no problem or alarm. If however on the increase of demand the oscillation is exacerbated, the supply will become increasingly restricted with all the energy being introduced to the system being absorbed in accelerating and de-accelerating the column of fuel within the pipes, generating shock waves and in accordance with Newton’s 3rd law, powering reactive movements of the mass of fuel line, components and other connected structures.
In this latter circumstance, a point will be reached where the suction of the HP pump causes such a severe pressure differential across the valves that the system will reach a limit where the flaps are moving with such speed and force that they will be bouncing between whatever mechanically limits the maximum open position and the fully closed positions. In this state the liquid managing to get past the valve will be a fraction of the normal full flow and at which time it becomes anyone’s guess as to the state of the fuel – or foam - between the HP pump and valve.

I have omitted a rather critical part of this explanation until now which is ‘So what caused the resonance to start in the first place? The simple answer is that in this particular case I really don’t know! However, if you can understand the explanation and visualise a horizontal column of pressurised liquid passing through a slightly open flap valve and then giving that pipe a little sideways shake perhaps you can see how easily it might start even though there are so many other parameters that have to be ‘just right’ for it to have any chance of self perpetuation.

If this overall scenario as described seems unlikely to begin with, it is made even more so by the fact that there are the two parallel non return valves on the booster pumps and for the effect to develop it is actually essential for them to react simultaneously. In the first instance this may seem absurd but they are both identical, operating close together in an almost identical environment and being induced by the same forces. Even if one valve does slightly lead the other, once even slight pressure variations begin, it is quite feasible that they would operate in perfect unison.

Rather than rewrite the details, I explained in my original post #1165 how once started, this effect could propagate from the right to the left side taking a few seconds to do so.

Still so very many holes all to have to line up…



Regards

JG

tanimbar

The update you refer to is a previous release, 3/2008, issued in early May.

The 16 June date on the website relates to the 6/2008 AAIB Bulletin within which is published (again) S3/2008.

I might be wrong but I can't find any changes in the two documents except someone has edited the earlier document to name it S3/2008 when previously it was called 3/2008.

Might be wrong though.

Note to myself: I must get out more.

regards, Tanimbar

lomapaseo

johngreen

No sense in talking about resonance unless you also include dampening (friction losses)

Forcing functions are common place, undamped resonance is not.

johngreen

Lomapaseo

The circumstance described is most certainly unusual but I can assure you regardless of friction losses, there are occasions when very different systems in different conditions and with very different fluids will behave exactly as I describe.

regards

JG

airfoilmod

fork in the road. To damp the energy within the piping, absorbers,(damped mounts) must be tuned for all forseeable mechanical vibration. To that extent, a certain flexibility needs to be engineered into the mounting of the piping. But wait, if there is too much, the Pipe and Fuel can act in unison, which would conceivably block flow. So what is the design of the pipework relevant to the contained and flowing Fuel? And let's not forget forces other than resonance. Like Harmonics, Flutter, coef friction values dependent on fuel type and source. This is getting interesting.

Airfoil

At the point of entry of this discussion, system isolation and ETOPS are frankly irrelevant. Two systems on separate A/C could behave identically with one another given the ineluctable power of certain anomalies and reasonably similar parameters. Lose the seductive and nonrelevant impetus of ETOPS, or lose the basis of debate. This isn't about how incredible it is to entertain the thought of concurrent failure anymore, it is a given.

Carrier

Is it a fair observation that more engines would not be the solution? If there were four engines would all four give inadequate power with the same outcome?

airfoilmod

as it may seem. If two heretofore simultaneous failure proof systems have just failed, and not due to all the engineered in safety of ETOPS, then the answer is no, twelve engines would have failed the same way. Isolation from bowser to chox is possible (At ridiculous cost), so by definition we're dealing with a non Etops issue.

My mentor and his co-pilot ate the same pork sandwiches on their way to bomb Tokyo in 1945. Both got food poisoning and were unable to fly. The chicken sandwich got ate by the navigator, who was an eager ab initio student on the way back. Separate everything for each side on the ETOPS twin? Ouch, you think Fuel's expensive.....

Airfoil

trashie

At a recent safety conference attended by a Boeing accident investigator, it was revealed in social discussions that Boeing had failed to find any mechanical, environmental ice, or other aircraft malfunctions as causal factors, however, have attributed the malfunction to interference with the fuel pump electronics from a laptop computer being operated against airline rules by a senior British government official over the top of the area of the pumps.

propaganda

I doubt that theory, as no Senior British Gov't official would be sitting that far back in the cabin.

CONF iture

HP fuel pumps are mechanical ... not sure there is any electronic involved (?)

Rightbase

Hmm....
Silly late night thought. If senior government officials could be persuaded to sit directly above the high pressure fuel pumps it might solve some of our other problems.

barit1

I'm 99% certain the reference to "interference with the fuel pump electronics" must really mean EMI to the FADEC. However that does not explain the fact that the fuel metering valves were commanded open, but fuel actually delivered was a fraction of the demand.

airfoilmod

Restriction between heatexchanger and FMV would explain both HP high side cavitation and no additional flow at valves; similar result if restriction was the H/E or prior. Not a "Pull" problem but a "Pack" problem.

arcniz

I find it very gratifying to see so many informed and thoughtful people putting their heads together on the "outside the box" analysis underway. In the worst case, this discourse burns some time but stimulates flexible thinking about things that others might readily accept as "status quo". In the best case it may allow or inspire people in the official determination loop (and in the aircraft systems development stream) to work a bit harder at understanding the outer margins of probability for failure in the marvelous products they devise.

Comment on two recent posts:


I would suggest that your concept is very promising, but the conclusion is incorrectly stated:

The effect of the submerged boost pumps running will be to keep dissolved gas IN solution, because the pressure gradient across the pump is relatively gradual (with fluid present at both inlet and outlet) and the pressure added by the pump is not so large an increment to the gravity feed pressure of the surrounding fuel.

If, however, the boost pump ceases operating, then a larger pressure gradient develops at its outlet side due to the suction of the HP pump and engine. This increased suction pressure gradient would likely be the cause of dissolved and otherwise entrained gasses increasing in volume and coming out of solution in a way that might disrupt the overall flow.


-------

Johngreen's detailed explanation of resonance is very informative as a baseline for discussion.

My understanding is that the resonant couple in a plumbing system like the 777 fuel supply path includes at least the two normal tuned systems, plus possibly other parasitic resonances. A first tuned/resonant path is the plumbing, pipes, etc, and the terminations formed by pumps and gates and valves. A second is the fluid, gas, or whatever may be rattling around inside the plumbing. Outside structures and forces may contribute to additional resonant loops.

A teasing difference of circumstance in the BA038 context is the history of very cold fuel. Someone has credibly noted that JPxx fuel viscosity is characterised only at -20c (iirc) in the standard fuel spec, meaning the viscosity might differ widely with varying fuels in the regions above and below -20c. Clearly, fuel viscosity will affect flow characteristics as well as the propagation velocity for shock waves (which controls resonance behavior). Could it be that the very cold fuel temps changed viscosity of a somewhat non-common fuel in such a manner as to put the resonance propagation velocity characteristics of the fuel outside the range of modeled behavior used in the 777 design calculations, and therefore outside the damping ability designed into the system?

If low temps, uncommon fuel, and resonance possibility were factually linked to the chain of causality, one might have a very plausible failure mode.

Flight Safety

I'm not sure I accept the resonance theory, but for the sake of discussion....

Years ago I once had a hot water shower faucet that resonated. It would start as the water got hot (temperature induced). The typical screw type faucet was many years old and the screw stem part was worn. When the resonance started, the rubber part of the valve (at the end of the stem) would oscillate from open to closed to open again on the valve seat. When the resonance occurred, the water flow would reduce (the valve being closed part of the time). I assume the hot water made the copper pipe flexible enough to allow the resonance to start. Naturally the resonance would occur at the moderate setting appropriate for a nice warm shower (if you held the handle with your hand the resonance would stop), which meant the valve had to be open or closed more to stop the terrible noise (resulting in a too hot or too cold shower), so the valve being open a certain amount contributed to the resonance. The fix was to replace the worn faucet with a new one (where the stem would not move and thus would not resonate).

Now to this accident.

There was no flow bypass in the shower example (which would have prevented or damped the resonance and provided alternate flow), and the energy feeding the resonance (since all resonances have losses) was the available water pressure. In the 777, we have 2 fuel pumps on each wing (boost and HP) capable of providing the energy to sustain a resonance. My question is in a 777 fuel system, do we have a valve in a common un-bypassed path between the tank and HP pump, that can restrict flow fuel if driven open and closed by a resonance?

airfoilmod

I think it important that other sources of energy be considered. Resonance is intriguing, and damping the pipes, etc. The energy provided by Boost and HP is substantial, but more than this, consider the introduction of massive vibratory and sonic input from core and fan. This is what happened, by report, (Initially responded....). A cold and viscous fuel in a slow, low energy green state is assaulted by enormous and rapid changes, locally, and we talk about tap water? A "length" of contained fuel may not react well to local insult of 35k kg of thrust and all the intricately timed reactions inherent in the system. Envision a long narrow mass (fuel), vibrating (in sequential expressions relative to its proximity to the energy source) and ponder what all the vibration is doing to its mass, and include sonics, flutter, (valves), etc. It is not for nothing Boeing has assembled the entire fuel system at test.

Airfoil

Re-Heat

Interesting that modern, greater precision could result in more accidents (similar perhaps to the greater navigation precision we see today that heightens danger to aircraft flying on the same airways through the airway centreline rather than scattered through the airway).

The implication is such that ever greater precision is not desirable, and requires exponentially greater understanding and testing of the small defects that arise now the larger, obvious defects have largely been solved...

pls8xx

If the oil heat exchanger is located in the pipe network of only one side, then the separate systems are not mirror images of each other. Similar resonance to dis-similar systems at about the same time is highly unlikely.

grizzled

Your view is one that, I think, will be repeated and strengthened over the next few years. Although it may contribute to "thread drift" I must opine that you have identified what I believe will become the prime aviation safety issue of the next decade.

lomapaseo

Re-Heat

--------------------------------------------------------------------------------



Agree. but only when we have acheived a factor of ten reduction in accidents due to the more obvious causes (excluding BA038.)