Doubtless your concern is not yours alone, which is perhaps one reason we don't have "the answer" from the NTSB yet. But keep in mind the current theory isn't just "fuel freezing". That is only part of the problem.

The current theory runs like this: during cruise the fuel flow was fairly light and equal, and both lines iced up about equally. Not enough to starve the engines, but enough to restrict flow to about the current flow.

Now the plane descends and warms. The ambient is warmer than the fuel, and the line going down the engine support is quite warm on the outside. The ice begins to thaw nearest the tubing wall, but is still frozen inside. We now have a lot of "loose" slush, but because it is impacted and has minimal flow past the inside (and thus minimum pressure exerted on the ice mass) it isn't going anywhere.

Now the plane has to counter some ground turbulance, and as it just happens, this results in three cycles of rythmically increasing and decreasing fuel flow. You may have experienced how one push on something stuck won't get it lose, but if you push on it rythmically a few times you will eventually get it lose. This isn't just your imagination; there are well-understood reasons why this happens. The theory seems to go that this has broken lose the ice in the lines, and probably got some (but not all) of it down to the heat exchanger.

Now finally there is a call on both engines for massive fuel flow compared to what has gone before. This causes all the ice to break loose and pile up against some obstruction, resulting in a major blockage. It is possible this obstruction was the heat exchanger. It is possible that it was some place upstream in the fuel piping that has not yet been identified. The result would be approximately the same, but not identical.

So there is a solid theory, based on both physics and experiment, that can (almost) account for the observed symptoms in the crash. There is also a demonstrated way (blocking the heat exchanger) that could cause this result. So there is reason to fix the heat exchanger post haste. The problem is that blocking the heat exchanger doesn't quite account for all of the observed symptons. So they can't say "fix the heat exchanger, case closed". More research is unfortunately needed, and probably solutions to other problems once they are discovered.

Markieboy

It wasn't at the same time, it was seven or so seconds different. And that in itself was the most powerful piece of information that said it was a common cause but not necessarily an instantaneous common mode fault like (for example) an electrical bus, relay, or fuel pump failure.

Obviously you are a statistician. I'm a fuel guy and I have a different opinion, which is that identically built engines constructed to incredibly close and repeatable tolerances operating on identical fuel in an identical environment might well be expected to suffer identical failures at more or less the identical time.

There is another thread running on PPrune concerning the amber thrust reverser deployed caution lights that came on after V1 on the take off roll at Johannesburg recently for both inboard engines on a 744. There is a hell of a lot more potential variability in those circumstances over a shorter period of time but not only was there a common cause, the lights came on within a few seconds of each other.

As Nicolas Cage said in City of Angels "just because you don't believe it doesn't mean it isn't true"

Dairyground

You have kind of answered your own criticism by explaining the difficulties that the AAIB are having replicating the incident. But introducing artefacts like cameras into a fuel icing trial is just another example of us not learning from experience: even if you could do it, you change the dynamics of the fuel system such that it is not representative of the actual conditions.

Examples:

- When the original jet engines were tested using frozen chickens fired into the rotating turbine it was wrongly assumed that the test more or less replicated what would would happen when an engine encountered a bird at altitude.

- One reason the Kegworth 737 - 400 crash happened was because CFM extrapolated the data from the -B1 engine series to the -C1 engine on the -400. And we now know that the engine didn't behave exactly as predicated by the computer model or the ground run that used a modified spinner.

- When RR did the containment tests on a trent they may have modelled a blade separation, but they were still required to write off an engine by doing a real-time engine test where the blade was actually ingested.

I wasn't advocating mandatory in vivo testing of every powerplant with every fuel: I know that in vivo tesing is expensive. I simply want to see a dedicated low temperature research facility which has the laboratory testing you cite but which then has it confirmed by the in vivo testing of actual powerplants on a variety of fuels in a variety of environments. My understand that Boeing and the USAF have been doing it for years for miliatary applications.

Fuel Tank Heating
 

IMHO, regardless of where the restriction formed, all of this still points toward a need for in-tank fuel heating.

isn't that what got TWA800 into trouble?

Nope
 

That was a missile.

TWA800. Not heating caused. Empty, 'fume filled' Center tank, and some 'arcing' internal wires, as I recall. Much talk of the need for 'inerting' tanks, like the 787 does. The need for inerting cropped up due to the extensive use of composites. No 'Faraday Cage', no 'bonding' (in the sense of 'electrical', not composite.)

Arcing Wires
 

Will is right on the cause for TWA800.

If people would read the thread rather than just skimming it, they would see that the second interim report only got indications of fuel slush icing up the heat exchanger by going totally away from both the max spec of water in fuel spec, the sample taken from the crashed plane and the probable (i.e. guess) mixture based on the Chinese fuel and likely change during flight.

The AAIB cannot (or at least not publically) replicate what happened. There have been a couple of rather inflamatory press releases from other sources which do not seem to have anything to back them up other than either wishful thinking or extreme caution.

IF (and it is a big IF) it is found that fuel slush is the problem then new certification rules will need to be established and ALL plane/engine combinations will need to be checked.

.

Too many watch the shiny object, (FAA, AAIB, etc.) The problem is not the FOHE, Water does NOT belong in Fuel. The Piping picture in the Report is Upstream, and the quantity of Ice is disturbing before any discussion of the Trent starts to distract the public. Modifying the Trent and assuming GE and PrattWhitney are invulnerable to the Ice produced on test is presumptuous. Assumptions about Fuel caused BA038's demise, not the impacted FOHE.

It's the Fuel. Tanks weren't getting sumped, water was allowed to build in tanks between flights, improper engineering and sloppy Fuel practices are a danger sign, no less troubling for the entire fleet of a/c than Twins.

Where is the GC tape, the MSA from the Accident samples? That cannot be privileged or secret in a free society, anybody privy to the complete data generated in the Investigation? Beyond a certain reasonable time interval, (which I protest, obviously) the complete results need to be made public, not sequestered behind locked doors like in some ancient Eastern regime.

Slush - NOT
 

Fuel /ice slush? Think about this. At the point where the pump goes into cavitation, the fuel pressure at the pump inlet falls to near zero. And the near zero pressure will extend back through the pipe to the blockage. On the other side of the blockage the pressure will be one atmosphere, plus any head from elevation below wing fuel surface, plus the pressure added by the wing boost pump. That is a major pressure difference across the blockage of well over 30 lb/ in. Also the small area that is not blocked is carrying all the fuel flow. If the pipe is 95% blocked per the report mock up that caused cavitation, then the velocity as it passes the blockage is 20 times normal. Still think it was slush?

Nope, Will is wrong
 

Fuel Flammability
The NTSB's theory is that some unknown spark ignited the Jet A fuel vapors inside Flight 800's center fuel tank. This theory depends on 2 things: a) sufficient fuel vapors at the proper fuel/air mixture; and b) a sufficient spark to ignite the fuel vapors. Neither condition has been found to have been present in subsequent tests. In Boeing's recent filing with the NTSB they have stated that they have not found any source of spark in Flight 800. The NTSB would have you believe that Jet A fuel vapors are a virtual bomb waiting to go off, yet every day hundreds of 747's are sitting on hot runways in places like Saudi Arabia, India, etc. with empty center tanks and none have ever exploded. Every day aircraft with empty fuel tanks are hit by lightning, a spark thousands of times greater than necessary to ignite this vapor, yet these aircraft do not explode. Why, because the fuel vapor is not explosive.

The Flight 800 Investigation

pls8xs

Technically, in the cavitation incident described it is at the Outlet of the HP where the Pressure is zero. The HP is Upstream of the FOHE, so the FOHE had exactly what to do with the Fuel Starvation ? This is why the results have been portrayed as possibly (sic) indicating upstream (of the HP) blockage.

The lack of a positive conclusion should not connote mystery, it should infer a manufacturing of escape scenarios for the Report. It also allows for leaving things the way they are (but for AD), at least in public disclosure.

FlyingCroc

Not wrong. I haven't stated I concur with NTSB, I merely stated the conclusion of the investigation. I have some reservations about the 'Explosion' of the tank. The Japanese guy who croaked in the 'seat bomb plot' was sitting in a seat that was behind (in front of?) the center tank, a blunder by the one who planted the downward exploding charge.

Thanks
 

Will, I thought you were smart enough not to believe these NTSB liars. Now back to the topic BA038.

FlyinCroc sir,

Agree.

You need an oxidising agent (usually O2 from the air) to go with the reducing agent (hydrocarbons various) before you can get a combustion reaction - as well as the activation energy having been exceeded by the spark from the faulty wiring.

As there was no inerting and the tanks are vented to air, its simply an input output calculation as to whether or how much O2 is present.

My money is on very very limited amounts of O2 being there - the Vp of Jet AI being > the diffusion inwards of the air.

In Calcutta no chance - paradoxically the higher the ambient the safer the "empty" tank.

Now what was the temperature profile for TWA800 ..........

Yes we could do the sums but this is so far off thread that I confidently expect the mods to intervene ...........

CW

Your remark triggered a thought.....
Of course it doesn't belong there, and neither do other contaminants.
Yet they get in there sooner or later.
So, shouldn't a fuel system be designed from the outset for far worse crud than meets the "official" fuel specs?

Maybe a lousy analogy, but with perfect ATC, etc., there should be no other aircraft in my airspace. Yet we now have TCAS, and rightly so.
With perfect nav, CFIT should not exist. Yet we have GPWS, and rightly so.

With perfect fuel, this should not have happened. Yet it did. Maybe fuel systems should also have be designed for when 'sh!t happens'?

Whatever happened this time, the system was not "fault tolerant" enough.

CJ

You raise the 64 dollar question, Chris. For fifty years the USAF and RAF have been flying up to FL600 and higher, to my knowledge, not one a/c has been lost to Fuel ICE. That in the days when jet fuel was.... well, ask Pinkman, it was crap.

With the benefits of metallurgucal, refining, electronic and systems progress, we now worry all of a sudden about 'mysterious' effects of water at altitude? Give me a ----ing break. The 'clog' in a heat exchanger brought down a 200 million dollar a/c ?

Any Lear owner knows to add a little Pristtm, but fuel monkeys and Pilots can't be bothered to sump a big rig ? "It's Frozen, wouldn't do any good."
Right then, 777bravosierra ready on the Right. Oooh, chemical mysteries, tubesheet maladies, Amoebic slurries, micro nucleated crystalline cascades disrupted by non-cadent thrust demands............what a bunch of Bull-ess. Does Dr. Fuelinstein not really know ? Or is he just too embarrassed to say ? What does a sump exercise cost ? Mr. Beanie knows. How many add up to 200 million dollars and the possibility of the need for 200+ caskets ?

Edit: USN as well as outfits listed above, sorry

Safety Factors Are Routine
 

Yes. Engineers are taught to build in a safety factor on every component. That's why wings don't come off as you get to the published limits. Water, and other contaminants, will inevitably get into the tanks and the fuel system needs to accommodate that.

Ed,
I happen to be an aeronautical engineer, and I know that.....

But when I designed in a 1% resistor into an AFCS system, I already knew it would be at least that, and more likely 0.2%. And that anything over 1 to 2% would be found out by acceptance testing before the system ever even went on an aircraft..

Here, we may be dealing with something that the engineers did not build in enough safety factors for... quite simply because the design was based on "the published limits" for the fuel itself.

CJ

Fix It Or Restrict It
 

CJ:
I agree completely. I think they designers didn't consider very long flights across arctic airspace. The cold-soaking exacerbated the situation. I think it just happened that ice, slush, or whatever we want to call it caught on some protrusions in the piping and other components. At the risk of being slammed for repeating myself, it's time to start heating the fuel in the tanks before it has a chance to clog any downstream component. It's either that or limit the aircraft to warmer air and/or shorter flights in very cold conditions. The cost of retrofitting will be high but won't equal the cost of even one repeat accident.

May be a much cheaper and simpler solution than heating. Keep the fuel moving. If it is flowing right along, there is a pretty good chance that it isn't going to have time to build up on a decently smooth hunk of fuel line. How much flow rate you would need and how much good it would do are calculable.

Someone mentioned that the HP pump is a nice inefficient hydraulic design where it is always making lots of flow at full pressure, and the excess fuel is recirculated back to the pump inlet. Add some hose and recirculate it back to a jet in the tank directed at the suction inlet strainer. Or even to a T in the inlet strainer assembly. Or point it at the tank sump and let it mix with any water present. Whatever the CFD analysis indicates if the best point to inject the recycled flow. (And the HP pump outlet flow is going to be warmer than the inlet feed by some amount, so you will get some minimal fuel heating.)

The Pump is driven by the engine gearbox, which in GE has its own heat exchanger interface with fuel. The HP doesn't pump a great deal more than the engine requires, and the spill is a fairly closely calculated flow that is more a safety consideration than a potential Gross Tankage circulator. EPR is Fuel reliant directly; the 'Safety' factor not even a critical one, if the fuel isn't enough, the engine keeps running, but at a rate that is flow dependent, as BA038 demonstrates. What's critical is to keep the Flame lit, and not add too much fuel.

All I can do is repeat that Water needs to be out of the Fuel. Including the mystery water that comes from secret places that are mysterious and cause unknown boffins to build imaginary 'test beds' that keep the public distracted until they lose interest. Why heat fuel when sumping and caution are enough ? I say we make them choose; frankly I'd choose better Fuel supervision, but that's me.

Sumping - easier said than done sometimes...
 

No water in the fuel at cruise - too cold, it is all ice.
Descent to a surface still below freezing - atmospheric ingestion contains water vapour which at lower altitudes condenses as frost on the cold tank surface.
If ambient stays below freezing, frost stays on the tank walls. Nothing to sump.
On the return journey, same story except that at lower levels the ambient is above freezing and the frost melts.
The melt water from the centre tank is sent into the still cold fuel in the two main tanks where it freezes again...

Rightbase

I would acknowledge all your conclusions. I don't see anywhere that you think it is ok, this water in fuel. That leaves sumping after melt at a suitable location. Because something is difficult should not mean it gets a box checked, because to eliminate the water isn't 'easy' ...

It is the tolerance of water build up in 038 over time that is the problem; Boeing had to inject water into the fuel to reproduce the accident conditions.

Will

Possible solution ...
 

... warm refuelling.

Water in fuel tanks is inevitable and you are right - sumping is the answer - but it only works above freezing point.

There's already hot hydrant, but in any case the infrastructure exists, it just takes money and the will.

Turnaround Time
 

Assuming that you do pump warm fuel into the tanks, how long will it take for any water, frost, or whatever to condense and settle at the sump drains? If the next flight is only an hour or two away, is that enough time?

BA038 was on stand overnight....afaik

if it was five hours, that's 40 million an hour and a fractured leg.

Injecting water ...
 

is what happens after the frost in the centre tank melts, when the fuel purge pumps get hold of the melt water.
A bit of clever redesign might make the refreezing of this melt water in the cold main tanks slow the purge process rather than the engines.

Will F:
Not quite true, except at SL. But you gotta consider the whole envelope.

Think about this: The HP pump flow rate is directly proportional to core RPM, and is independent of altitude. Thus if there's a small flow margin at SL, there's a big excess at altitude.

Further: If idle speed is 70% (for discussion purposes...), then the HP flow rate at idle is 70%, where the actual engine burn rate is maybe 10% (or less). Therefore the extra 60% has gotta be bypassed.

STILL further: Idle at top of descent combines the altitude bypass AND the idle bypass, meaning there's a potential local overheating (through recirculation).

It's this significant excess, or bypass, fuel flow that is the question. If it's not gainfully employed, you get hot spots & cold spots in the fuel system.

Even ignoring the heat, there is significant fuel flow, several times the idle and cruise requirements. Ice has a fairly hard time sticking to reasonably smooth tube in high flow conditions. So the flow itself would likely prevent ice buidup. The heat would also help, even if a fair amount of it was lost by the time the recirculation got back to the tank inlet.

BA038 may well have been on the ground overnight, but what was the ambient temp overnight? What time was it refuelled? I bet about 3 Hrs before dept, as is standard. Not enough time after the refuel to allow any melting & settling for a meaningful water drain

I have been unable to do the sumps on a DC10 after it has been in a hangar for 12 hrs on a UK winters day because they were still frozen.

Maybe when the Glacier starts to impact Fuel capacity this problem will get attention. Was what was frozen solid in the tanks what was left ? Was the a/c actually totally OOF ??

Just when you need it Global Warming lets you down! Yes yes extreme weather etc ad naueum .

More seriously I very much take Wills glacier point, hyperbole in action but he's right to pose the question.

How many fueling cycles, cold soaks, limited efficiency sumpings does it take to build a glacier I wonder?

And .........of course we lost the contents of the centre tank on impact.

CW

Good question for the general case, Chris, but IIRC BA038 was pretty well water and ice free before starting the 2-way trip to Beijing and back.

So irrespective of where it collected, where did it come from?

I followed most of the thread, so my remark is not meant to be fatuous, just very summary...

Says who?
Or was this the icicle that broke off the glacier and broke the camel's back?
Or the slush puppy that finally choked the FOHE?

CJ

Edit: see Rightbase's post below. I concur. Back to base one, largely! Post #1805 is on page 91, to avoid anybody doing a tedious search.

Post #1805 refers .....

PS - it is a real pity they didn't find melt water in the FOHE in their first inspection after the episode. They will know where to look next time.

I just dont buy the simple water theories. The water-carrying capacity of a fuel is finite and varies with temperature. AFAIK the tests have not been able to reproduce the fault without direct injection of water, a situation that would not occur in real life. I am not a conspiracy theorist but I am convinced there is something else fuel related going on.

PS "hot hydrant" doesnt mean "hot" or "warm" in the temperature sense. It simply means that the fuel is continually under positive pressure in the fuel trunk line to the apron service points, in the same way that your house electricty sockets /outlets are "hot" or "live" all the time. Keeping it this way prevents water and dirt entering the fuel line although it does mean that there is an increased risk of fuel leakage which is why there are (or should be) leak detection systems running along the pipe conduit.

What Was The Water Content Of The Fuel Before The Flight?
 

I have read the whole thread but most of it was several months back so I'm vague on some of it.

As I recall, the fuel from the aircraft which was tested after the crash was within specifications with regard to water content, but that would not have revealed the water present before the flight. Any water uploaded with the fuel or present in the tanks would have begun to migrate to the drain sumps and the sumps for the in-tank pumps before and during the flight. Some of it would even have been pumped through the engines. Therefore, was it in spec, as far as water is concerned, when it was loaded aboard? While we can test the fuel in the tanks in Beijing, how we can tell what the water content of the fuel in the aircraft was before or during much of the flight. Does anyone really know? I don't think so.

Were the tests conducted with actual fuel from the aircraft or with fuel obtained from some other source? Presumably, fuel from any other source would be more or less water-free, so it would be mandatory to inject some water to get any icing in the pipes and FOHE.

The current arrangements represent a fairly good balance between cost, complexity, and specification. Thats not my opinion, its the opinion of almost everyone in the business. There are over 20 steps in the supply chain for managing quality and checking specifications (including centrifugal (coalescing) separators and microfiltration that you all are going on about). The important thing to understand is that the supply chain QC checks take place in a controlled environment. Aircraft already have filter screens against gross physical contamination. Moving water filtration systems to a aircraft is a complete non-starter because:

1) it moves the quality control operation to an uncontrolled environment
2) it adds weight and fuel burn to the aircraft (have you ever SEEN the kit that is used? weighs about half a tonne) Facet vertical coalescer separators, API-1581, Category C, Type S, Fifth Edition, aviation fuel filtration applications | CLARCOR
3) You would have difficulty inspecting and changing filters safely and inevitably it would get forgotten until, oops
4) The cost of miniaturizing the components would be astronomical

Forget it.

And yes water in fuel systems is inevitable because of the continual ingress of air through the vent which has moisture in it and no its not just a problem at sea level, all air has moisture in it which then freezes and condenses on the inside skin and forms free water which then dribbles down to the bootom of the tank and gets reabsorbed into the fuel in an free water:dissolved water equilibrium situation depending on temperature / density of the bulk fuel .

And in any case who said water was the issue?

Smilin_Ed no, the tests were not done with the actual samples (there is not enough of it) so they faked up a test batch. That didnt reproduce the problems. So they actually injected water into the fuel and eventually managed to get ice rime buiding up on the inside of hoses and on the FOHE plate.

Pinkman, do I understand it correctly that they injected more water than was found in the fuel in the plane?