17
$\begingroup$

There seems to be a certain class of airline accident where an airliner engine flames out followed by further unfortunate events. Very often, pilots are recorded as having attempted to restart a failed engine by windmilling, which seems to involve attempting to restart an engine from the rotation due to the passage of air due to airspeed. In accident reports this rarely ever succeeds, leaving the impression of a rather worthless exercise.

But, of course, there's a massive bias here because these are the reports of accidents and if the windmill start worked, things wouldn't have deteriorated.

So, in rough terms, how desperate is attempting a windmill start. Is it in the land-on-water level of optimism, or something which usually succeeds which we never hear about?

$\endgroup$
  • 2
    $\begingroup$ As you observe, there's a huge selection bias, there. $\endgroup$ – David Richerby Sep 1 '16 at 8:23
15
$\begingroup$

Windmill starts are a rather last-resort type of thing. Lots of things have to go wrong before pilots will be attempting that.

First, an engine has to fail. Since airliners are required to have at least 2 engines, and be able to fly on one engine, this isn't a dire emergency. Pilots will work the associated checklists. If there is no apparent issue they may attempt to restart the engine using bleed air from the running engine. If there is an issue they will leave it shut down. With thousands of planes in the air, this sort of thing happens pretty much daily.

The chances of the other engine failing are very low, and generally due to some common circumstances that affect both engines. Pilots may be more aggressive in trying to get even a damaged engine started, as a damaged one is better then none at all.

The APU may be started and used to help start the engines as well. But this takes time, which tends to be running out when gliding with no engines running.

As ymb1 pointed out, engines are required as part of certification to demonstrate windmill starts. Issues that can prevent this from working tend to get the attention of the safety folks. This establishes limits and guidelines that are put into the aircraft checklists. Pilots know that if they maintain a certain air speed, the engines should reach a certain RPM, which will allow them to be started. Of course if multiple engines have already failed, there is a chance that this issue will prevent a restart.

$\endgroup$
16
$\begingroup$

If the engine isn't damaged, and the restart altitude/speed are followed, the engine should start. Fuel starvation, bird ingestion, etc., do not count.

enter image description here
(YouTube) In-flight shutdown.

In-flight restart is tested as part of the aircraft certification, and it must work. In the link above is a video showing the procedure in a 777F simulator.

Engine Restart Capability - § 25.903 (e).

Tests should be conducted to determine that inflight restarting can be accomplished within the envelope provided. Restarts at the conditions of the critical corners of the envelope and at or near the high altitude extremes of the envelope should be conducted to verify the boundary conditions of the envelope (FAA).


Examples of successful restarts:

  • On British Airways Flight 9, all four jets on a 747 failed after entering a volcanic ash cloud, all 4 were restarted after exiting the ash cloud. One of the four failed shorty after, but volcanic ash is sticky, the particles are very small, they melt, and they clog moving parts. So three out of four isn't bad at all.

  • For KLM Flight 867 (a mirror of BA 9) it was four out of four successful restarts.

$\endgroup$
  • 3
    $\begingroup$ BA-9: "The engines had enough electrical power to restart because one generator and the on-board batteries were still operating; electrical power was required for ignition of the engines." (Not windmilling.) $\endgroup$ – Jacob Krall Aug 31 '16 at 21:16
  • 2
    $\begingroup$ KLM 867 also appeared to have functioning standby power. $\endgroup$ – Jacob Krall Aug 31 '16 at 21:18
  • 7
    $\begingroup$ @JacobKrall source of ignition is needed for any type of restart. $\endgroup$ – ymb1 Aug 31 '16 at 21:18
13
$\begingroup$

As Oscar Wilde said, "pure and simple truth" is rarely pure and never simple...

The "not-so-simple" part is that engines don't just flame out for no reason. They only flame out because something is abnormal, either with the engine itself or with the environment (ash clouds, sandstorms, torrential rain, fuel supply, etc).

Therefore, the probability of a successful engine restart is already compromised in some unknown manner by the fact that it stopped.

Deliberately shutting down an undamaged, low-life engine and demonstrating that it will relight, as part of the flight test certification program, is about as good a test as is practicable (there is no way any flight test program will go around looking for volcanic ash clouds to fly through, water ingestion tests done on an engine on a test stand on the ground can't be followed by a windmill relight test, etc) but it's not necessarily completely representative of a real-life engine relight situation.

$\endgroup$
  • 3
    $\begingroup$ "water ingestion tests done on an engine on a test stand on the ground can't be followed by a windmill relight test" why not? $\endgroup$ – Random832 Sep 1 '16 at 6:29
  • 6
    $\begingroup$ @Random832 Because the engine is stationary on the ground so you can't windmill it, unless you have a really big hairdryer to push 550mph air through it. $\endgroup$ – David Richerby Sep 1 '16 at 8:26
  • $\begingroup$ @DavidRicherby Like the fan on another jet engine? $\endgroup$ – user253751 Sep 1 '16 at 11:57
  • 3
    $\begingroup$ @DavidRicherby You can't do the test in a wind tunnel? $\endgroup$ – Random832 Sep 1 '16 at 14:12
  • 3
    $\begingroup$ In actual accidents pilots confusing one engine for another seems to be a relatively common occurrence. I'm not an accident investigator but I seem to remember when I was growing up there were at least two major disasters on UK soil (Kegworth and Manchester Airport) when that happened. OK, so Manchester was on the ground, but in a "oops, wrong engine" accident, presumably windmill starting an undamaged engine would have been useful? $\endgroup$ – Dan Sheppard Sep 1 '16 at 17:48
4
$\begingroup$

The problem with windmill starts on a gas turbine is - it requires a fairly high airspeed, as mentioned in this bit on starting aircraft engines.

Typically, the APU, itself a smaller turboshaft engine, has an electric/battery start, while high pressure bleed air from it or another running engine or a ground supply is used to start a main engine.

If your only option is a windmill start, it means you have no bleed air from another running engine, or the APU, as in all of them have failed. Quite probably, you've run out of fuel, so a windmill start is a waste of time. APU's tend to start far more quickly than main engines, so if you don't have the time to start the APU, you probably don't have the altitude to swap for airspeed to try a windmill start.

The last thing you want to do if you're in that shape is lose more altitude than necessary when you have no power, and no other way to restart your engines.

Windmill starts on piston/prop planes in the 1930's and 1940's were performed, usually when the electric starter on one engine of a four engine aircraft had failed while the aircraft was parked. They would take off unladen on three engines, with the fourth prop feathered, and use the airflow to crank that engine to a start, then land and pick up whatever they were carrying and go on their way.

Or, WW2 fighter aircraft would windmill start when the pilot dropped external tanks but forgot to switch fuel supply to the main tanks. They'd just change the fuel settings, nose down and reset prop pitch, and fire the engine back up. In some cases, it was quicker than an electric start, or in the case of naval aircraft that used a high pressure shotgun shell starter, it was the only way to restart the engine in flight.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.