Air France A319 at LFPG. Engine #1 fuel starvation.

[B-HKD]

Incident: Air France A319 at Paris on Mar 12th 2014, engine starves due to lack of fuel

An Air France Airbus A319-100, registration F-GRHT performing flight AF-7665 from Marseille to Paris Charles de Gaulle (France) with 84 people on board, was descending through FL100 on approach to Paris' Charles de Gaulle Airport when the left hand engine (CFM56) starved after both left hand fuel pumps dropped offline. The crew worked the engine failure checklists, declared emergency and continued for a safe single engine landing on runway 08L about 10 minutes later.

The French BEA reported in their weekly bulletin of May 13th 2014, that the aircraft had departed Marseille with 5000kg (11,000lbs) of fuel. A post flight inspection revealed the left hand wing tank was empty, the right hand wing tank contained 1100kg (2425 lbs) of fuel, about 150kg (330 lbs) above required minimum fuel reserve. The ECAM displayed between 1000 and 1380 kg (between 2200 and 3040 lbs) of fuel for each left and right hand wing tank.

The incident aircraft remained on the ground until Apr 2nd 2014 (for 21 days) before resuming service.

ECAM showed 1000kg in left wing, and 1380kg in the right.

Upon inspection: 0kg in the left and 1100kg in the right.

[Tu.114]

Does Air France do a fuel check before departure (remaining fuel plus amount added vs. FOB indicated)?

[ManaAdaSystem]

They can't have done that on this flight, that's for sure.

[Mascot PPL]

Possible causes?

Really interested in this one as I worked on the A320 FQIS team back in the mid 80s, specifically on the test rig that emulated the rest of the aircraft systems for the FQIS (including the ECAM). Been out of the industry for 25 years so not certain if the original FQIS is still in use or has been replaced/upgraded? The thing (FQIS H/W & S/W) was built with the expectation that it would still around now but systems do get upgraded.

Never heard of one like this before so very interested in what caused it.

Reading around the web the I can't find definitive words on exactly what the ECAM was reading when they did the checks. One interesting piece of data would be if/when the left had tank reading got stuck (at either 1,000, 1,100 or 1,380 according to the various sources) and how long it had stayed at that figure before the engine pumps lost juice.

At first sight possible causes could be:

1) Faulty fuel probe(s) - didn't work on the A319 but it's the same wing so assume it still has the (IIRC) 30 capacitive probes in each tank. Would be hard (but probably not impossible) to get an over read of around 1 metric ton from a single fuel probe so there would have to be multiple failures?

2) Faulty wiring - mis-placed connector, crossed/shorted wires, giving an over read on the left wing.

3) Software glitch - some sequence of events that caused the software to go into a new "mode" and there was a bug there waiting for the day it happened that allowed the FQIS to over report (probably on both tanks, but only reported on one as the other didn't hit the "floor" value in terms of actual fuel).

I would guess the 21 days off-line included getting access to all the tank probes and checking them all, plus checking the wiring and connectors for all FQIS related feeds?

Probability of each cause?

1) Probes - probably low (unless there is a way to get such an over read from one failed probe). If over read has been there for a significant period of time and/or built up over time then could be fuel contamination blocking probes over a period of time - checking back on fuel records etc would help as well as testing the probes recovered from the A/C

2) Wiring - higher/highest likely cause (if the over read is not a round number). Would expect to see sudden appearance of over read after a maintenance schedule that included the FQIS wiring connectors being accessed - again access back on fuel records plus physical check of the wiring.

3) Software - probably low, hard to see a mode that would cause a constant over read on one tank (or both for that matter). FQIS testing was intense both in rigs and on test flights but there is always room for something to get through testing, no matter how much time/money you spend. If however, the over read was exactly 1,000Kg then I'd suspect this as the prime candidate. Reasons being 1) very hard to get wiring or fuel probe failures that would give exactly 1,000Kg reading in one tank, 2) Developers like round numbers and there will be filter values, limits, etc within the code that are set to 1,000Kg. A bug that allowed one of these values to be miss-used could result in a very exact over read.

4) Combination of probe/wiring/software - faulty probe or wiring caused the software to enter a state that allowed it to over read.

So if the over read was exactly 1,000kg I'd be digging out a lot of old FQIS code and running some new tests on it, otherwise I'd be looking at the probes/wiring.

[KBPsen]

Originally Posted by Mascot PPL

2) Wiring - higher/highest likely cause (if the over read is not a round number).

I've had connectors affected by cold causing the quantity in a tank to over read by round numbers (500, 1000 and 1500). It was easy to spot as the jump in readings was always a round number and never a random one.

[BOAC]

With what appears to be a tank gauging error, it looks as if the a/c quite possibly has been flying with 1100kg'ish 'missing' for a while in the left tank. I would have expected some reports of trim abnormality.

Does Air France do a fuel check before departure (remaining fuel plus amount added vs. FOB indicated)?

- this will not necessarliy pick up such an error, especially if the error came about gradually. A sudden jump of 1100kg should have shown if this process is done regularly.

It looks as if they were 'lucky' the 'engine failure' has not happened before on this ship.

[pax2908]

The text in the BEA .pdf reads slightly differently:
- "the crew decides to upload 5000 kg"
(given the readout problem, one cannot immediately know the exact quantity actually loaded, I presume)
- "[upon arrival ?] the ECAM shows 1380 kg left side and 1000 kg right side"

Also there is no mention that a developing L/R inbalance was noticed. Naively this could mean that a ~1000 kg constant offset, between reading and actual, was always present on the left side?

[Tu.114]

True; if the measuring error sneaks in gradually and keeps within the fuel check tolerances at all times, the plausibility check will not be enough to catch this. Also, if by coincidence a fuel check was improperly done at the time the fault developed, this might have stuck. So the aircraft might have flown with overindicating gauges and consequently too little fuel on board for a while already.

Dipsticks to the rescue...

[bigbird]

Very interesting post, given the increasing pressure to use plog fuel.
MOD:Could this be moved to Tech log?

[tubby linton]

Boac with a flybywire aircraft you would not notice any trim error. I was told by Airbus that you could fly the A330 with one wing empty(32t) and the aircraft would fly completely normally. The accuracy of the fuel system is discussed within Airbus FCB7 , but it gives a maximum error of about 800kg plus any apu fuel burn.

[pax2908]

tubby: "maximum error of about 800kg" that would be +/- 400 then ?

[SpringHeeledJack]

I wonder why this is in the spotters forum and not in R&N or Safety ? Perhaps not earth shattering in import, but nonetheless newsworthy and of interest to many airline workers ?


SHJ

[tubby linton]

The FCB states an error of +/- 800kg, for the total fuel system. It does not break it down per tank

[BOAC]

Originally Posted by TL

you would not notice any trim error

- this is of concern - surely somewhere in the magic system a chip or two might notice the required trim and tell mother?

Originally Posted by shj

Perhaps not earth shattering in import

- I add my puzzlement to those querying the move of this significant PILOTING event to a spotters' forum I would suggest that a gauge error equal to Final Reserve is just a little significant to us all?

[wiggy]

I add my puzzlement to those querying the move of this significant PILOTING event to a spotters' forum I would suggest that a gauge error equal to Final Reserve is just a little significant to us all?

a +1 from me - can we please move this back to the Technical/pro end of Pprune. Perhaps someone considered the incident was not "newsworthy" enough to deserve headline billing....

[BOAC]

Thanks, wiggy. You would think that 'losing' Final Reserve would fit with

Reporting Points that may affect our jobs or lives as professional pilots. Also, items that may be of interest to professional pilots.

- but maybe there are non-professional pilots playing with moderating?

[SpringHeeledJack]

Perhaps someone wanted to hide it in plain sight ?


SHJ

[tubby linton]

BOAC the FBW would notice but the effect would be transparent to the crew as in level flight the pilot has not demanded any roll. It is the same with the yaw from an engine failure., the fbw works hard to keep the wings level without any crew input.All you have to do following a V1 cut is pitch to 12.5 degrees and put in a bit of rudder and it flies happily away.

I would also like to see this topic moved to Tech log.

[BOAC]

Tl - 'transparancy' is nice and very clever, BUT surely it needs to be 'noticed' somewhere? Does a 1100kg imbalance not flag up anywhere?

[tubby linton]

BOAC, the aircraft will alert you if there is an imbalance of >1500kg but in this case the indications would have been almost symmetrical. Our aircraft have a display of fuel temperature in each tank. I wonder what is was reading as it only appears if the fuel sensor is wet? It would be very easy not to notice that it was missing. On a short flight such as this I wonder how many fuel checks they did, and was the ecam fuel page even looked at?
I see that this topic is still languishing in the spotters forum, please can a mod move it to somewhere more appropriate.