Fuel Starvation and Autopilot

[TAAMGuy]

I am wondering what happens if an aircraft is crusing with the autopilot engaged in the event of fuel starvation. (Yes I am thinking about MH370). How will the autopilot/FMS behave when it can't hold the altitude? I envision some sort of default profile akin to best glide speed. Is this programmable or hard coded etc? Please elucidate.

[de facto]

It will drop off the sky:E
Recently done a deadstick landing in the sim..all is possible but you need pilots in front.

[18-Wheeler]

A fellow here posted this in the MH370 thread ...

As a 777 pilot I, like many others, have wondered how the 777 would perform in the scenario where the pilots were incapacitated and the aircraft ran out of fuel. I had my ideas but there is nothing like seeing it for "real" so we tried this in a 777-2 full motion zero flight time approved simulator.

We used a zero fuel weight of 175 tonnes. We let it run out of fuel at FL 250 in track hold and alt capture. However it would not make any difference what mode it was in as everything would drop out. In real life one engine uses fractionally more fuel per hour than the other and there is typically a difference between main tanks of a few hundred kilos, so we had a 300 kg difference between the contents of the left and right tank.

When the first engine failed TAC (Thrust asymmetry compensation) automatically applied rudder. The speed reduced from 320 knots indicated to 245 knots indicated. It was able to maintain 245 knots and FL250. When the second engine failed the rudder trim applied by TAC was taken out and the trim went to zero. The autopilot dropped out and the flight controls reverted to direct mode. The speed initially came back to 230 knots but then the nose started to lower. The nose continued to lower and the rate of descent increased to 4,000 feet per minute, the nose kept lowering and the descent rate increased to 7,500 feet per minute with a bank angle that increased to 25 degrees. The speed at this point had increased to 340 knots indicated, above VMO but there was no horn as it was on limited electrics. About this point the RAT (Ram air turbine) chipped in and the CDUs and copilot's PFD (Primary flight display) came alive. The flight controls stayed in direct mode.The eicas screen was full of messages like pitot heat, flight controls, APU fault (The APU had tried to autostart due double engine failure but failed due no fuel to start it) low fuel pressure etc.

Then with a max descent rate of almost 8,000 feet per minute the nose started to slowly rise and keep rising. We had dropped to about FL170 but the nose slowly rose up to 6 degrees pitch up and we started climbing at about 3000 feet per minute and the bank angle reduced to only 5 degrees. It climbed back up to FL210 at which point the speed had come back to 220 knots and then the nose dropped down again and we were soon back to descending at 8000 feet per minute. So basically a series of phugoid oscillations with bank angle between 5 and 25 degrees and pitch attitude between about 9 degrees nose down and 5 degrees pitch up. It was losing about 8000 feet and then gaining about 3 or 4000 feet with airspeed fluctuating between 220 and 340 knots.

We didn't watch it all the way down due time constraints and stopped the experiment at 10,000 feet but it was consistent all the way down. Having watched it I can say with certainty that if the pilots were incapacitated and it ran out of fuel there is no way it could have landed on the water with anything like a survivable impact. Just passing on the info.

[tmpffisch]

I don't understand 18-Wheeler. You mean the recreation that Australia's so called 'Aviation Expert' Geoff Thomas made in the mock simulator running Microsoft Flight Sim isn't accurate????? I know the quote is from a 777 pilot...but Geoff's a journalist, sorry....I mean expert. Maybe you're wrong.

Flight recreated in 777 simulator - The West Australian (http://au.news.yahoo.com/thewest/wa/a/22235213/flight-recreated-in-777-simulator/)

[john_tullamarine]

.. and, as always, any sim is only as good as the validity and fidelity of the programming.

I would wonder whether one should rely on any sim to provide usefully valid information for such a scenario .. given that it is hardly something which the OEM or sim folk would be overly concerned about ..

[Halfnut]

Yea I just don't envision the simulator manufactures spending tens of thousands of dollars in programing cost to develop a high fidelity simulator that will replicate a double engine flameout due to fuel starvation.

I could be wrong so someone go out in a real plane and test how the plane flies with no fuel vs. the simulator and report back.

[grrowler]

I could be wrong so someone go out in a real plane and test how the plane flies with no fuel vs. the simulator and report back.

Could be something useful for GT to do.

[tmpffisch]

Yea I just don't envision the simulator manufactures spending tens of thousands of dollars in programing cost to develop a high fidelity simulator that will replicate a double engine flameout due to fuel starvation.


The sim should however, have the same programming/code which would show whether the autopilot would remain engaged under the simulated conditions; as per the original post.

[anotheruser]

I would wonder whether one should rely on any sim to provide usefully valid information for such a scenarioWhy not?

The behaviour of a big jet in case of a total loss of thrust should be easily predictable. It's not like it's flying upside down or in some aerobatic maneuvre at the edge of the aerodynamic envelope of the plane. The simulator is able to simulate the plane's behaviour in case of any change of thrust, so why should it not be able to simulate a change of thrust to zero? It is able to do that in a one engine out scenario, so why not with both engines out?

The only effect which probably isn't simulated that accurately is that of the windmilling engines, because most likely it hasn't been tried in the real plane, but everything else should be replicated exactly, I believe?

[FlightDetent]

Absolute loss of thrust is just part of the scenario posted by OP. In case of complete fuel starvation, BAT/RAT electric would be the second part.

The behaviour of FBW aeroplane will be determined by the combination of both factors.

[john_tullamarine]

but everything else should be replicated exactly, I believe?

I didn't suggest that it couldn't be done .. however the sim is just a big PC with fancy bells and whistles hooked up. Unless the programming logic addresses a given situation and is validated by FT data ... it must be presumed to be of questionable validity.

For instance, going back some years ago, I was instructing on a 732 box. Post the rudder problems, the FAA mandated that US carriers had to use a box with an updated set of data. As it happened I was last man standing and did some before and after fidelity checks on the particular box .. chalk and cheese, I have to say.

[vilas]

Although principally what John says is correct and the complete flight path from autopilot disconnect to the crash may not be realistically simulated as it hasn't been tried out this particular situation should be more or less correctly replicated. Since the sequence can be visualised as first engine fail the autopilot holding with reduced speed at limit thrust then second engine fail autopilot quitting then the descent. It is closer to Dual engine failure which is a practiced manoeuvre in the SIMs.

[john_tullamarine]

Caution ... you are endeavouring to assign intelligence to computers ... and they have none unless we are looking at AI software.

[HazelNuts39]

Yea I just don't envision the simulator manufactures spending tens of thousands of dollars in programing cost to develop a high fidelity simulator that will replicate a double engine flameout due to fuel starvation. The simulator manufacturers don't have to spend 'tens of thousands of dollars' for this or any other particular scenario. The ingredients required for the simulation have to be there anyway:

- The reconfiguration of airplane systems in response to certain failures, e.g. electrical power, autopilot, autothrust, flight controls.

- The modelling of propulsion system performance, e.g. the drag of a windmilling engine.

- The modelling of aerodynamic characteristics of the airframe and control surfaces.

Only if the airplane encounters extreme angles of attack or sideslip or Mach number does the simulator enter into uncharted territory.

P.S.
I suspect that the journalist added some drama for a more 'spicy' story. I doubt if a stall would occur in a more realistic simulation.

[mm43]

The simulator manufacturers don't have to spend 'tens of thousands of dollars' for this or any other particular scenario. The ingredients required for the simulation have to be there anyway:Which leads me to the obvious question; have you had a look at the scenario described above (originally posted by suninmyeyes in the MH370 thread) and come to any likely conclusions? Speed induced phugoid, closed spiral?

Though the actual plane of the bank angle wasn't mentioned, could this be due to set rudder trim?

[HazelNuts39]

Hi mm43,

In the journalist story I doubt in particular that the autopilot would stay engaged in the event of a double engine flameout, and even if it did, that it would stay engaged until the airplane stalled.

I think it is more likely that the autopilot/autothrust would disengage, and flight path control would then transfer to the FBW flight control system. One has to know more about the flight control system to say what happens next, but a phugoid longitudinal motion associated with variable bank angles does not seem implausible to me.

[vilas]

Dual engine flame out leads to dual hydraulic failure and double generator failure in a twin. In this single hydraulic (with RAT) and limited electrics situation there is no way the autopilot is not going to be available. The flight path that will follow can be different in 777 from Airbus since AB is flight path stable and 777 is speed stable.

[Semu]

Dual hydraulic failure? I'm typed on the whale, not bigfoot, but on the whale the hydraulics are supposed to stay lit from the rotation of the engines, down 180 KIAS or so. The hot battery bus should last the rest of your life at this point too, although you may not be able to acquire a fresh cup of coffee at this point in the flight.