FGTurbine don't start windmilling until powered up at least once

[Octal450]

I'm submitting a ...

• bug report
• feature request
• support request => Please do not submit support request here, see note at the top of this template.

Describe the issue

The engine stays at 0% N1/N2 and does not windmill unless it is started at least one time

What is the current behavior?

To reproduce start a simulation with FGTurbine but do not start one of the engines (example, in a twinjet, start only 1 engine), then takeoff and accelerate. Notice the engine stays at 0% N1 and N2. Then if you engage the starter and then shut down to get slightly rotation, then it seeks the windmill value properly

What is the expected behavior?

Engine should always windmill when off

What is the motivation / use case for changing the behavior?
So that he will work realistic

Please tell us about your environment:

• OS: Windows 10 Pro x64
• JSBSim version: 1.18 (FGFS next)

Other information

Should also point out, when you do start the engine, it spools down fast as if it was at 0 before the start attempt.

[seanmcleod]

Taking an initial look at the code in FGTurbine, it looks like windmilling is only implemented as part of the engine shutting down, i.e.

jsbsim/src/models/propulsion/FGTurbine.cpp

Lines 172 to 183 in 08d2f85

	double FGTurbine::Off(void)
	{
	Running = false;
	FuelFlow_pph = Seek(&FuelFlow_pph, 0, 1000.0, 10000.0);
	// some engines have inlets that close when they are off. So, if a flag is true disable windmilling
	if (disableWindmill == false) {
	N1 = Seek(&N1, in.qbar/10.0, N1/2.0, N1/N1_spindown);
	N2 = Seek(&N2, in.qbar/15.0, N2/2.0, N2/N2_spindown);
	} else {
	N1 = Seek(&N1, 0, N1/2.0, N1/N1_spindown);
	N2 = Seek(&N2, 0, N2/2.0, N2/N2_spindown);
	}

That's the only reference to windmilling in FGTurbine excluding the lookup to see if windmilling is disabled.

jsbsim/src/models/propulsion/FGTurbine.cpp

Lines 477 to 478 in 08d2f85

	if (el->FindElement("disable-windmill"))
	disableWindmill = el->FindElementValueAsBoolean("disable-windmill");

[bcoconni]

Yes, but the variable phase is initialized to tpOff:

jsbsim/src/models/propulsion/FGTurbine.cpp

Line 98 in 08d2f85

phase = tpOff;

so FGTurbine::Off should be called even if the engine has not yet been started:

jsbsim/src/models/propulsion/FGTurbine.cpp

Lines 154 to 162 in 08d2f85

	switch (phase) {
	case tpOff: thrust = Off(); break;
	case tpRun: thrust = Run(); break;
	case tpSpinUp: thrust = SpinUp(); break;
	case tpStart: thrust = Start(); break;
	case tpStall: thrust = Stall(); break;
	case tpSeize: thrust = Seize(); break;
	case tpTrim: thrust = Trim(); break;
	default: thrust = Off();

However phase is not initialized in the constructor so that might be the cause of the problem.

@Octal450, does the issue vanish if you initialize phase = tpOff in the FGTurbine constructor ?

[Octal450]

I'll try my best to test this tonight or tomorrow. A little low on time.

Thanks Bertrand!
Josh

[Octal450]

Hi @bcoconni, no difference.
Looks like the switch default is Off() anyways

Added phase = tpOff; to the constructor below disableWindmill in the latest on FlightGear next and recompiled and rebuilt.

Kind Regards,
Josh

[seanmcleod]

For testing I modified scripts\b737_runway_new.xml and added the following so that only one engine is running, then set a breakpoint in FGTurbine::Off() to see what was happening as the aircraft accelerated.

    <event name="Set active engine">
      <description>Only make one of the engines active</description>
      <condition>simulation/sim-time-sec ge 19.0</condition>
      <set name="propulsion/active_engine" value="0"/>
      <notify/>
    </event>

So first off FGTurbine::Off() is being called, and the windmill section is being executed.

  if (disableWindmill == false) {
    N1 = Seek(&N1, in.qbar/10.0, N1/2.0, N1/N1_spindown);
    N2 = Seek(&N2, in.qbar/15.0, N2/2.0, N2/N2_spindown);
  }

However the issue is that the accel - N1/2.0 and the decel - N1/N1_spindown parameters to Seek() are 0 since N1 starts off with a value of 0.

[seanmcleod]

So do we add a check for N1/N2 being 0 and use something like N1_spinup for say 0 >= N1 < 15 and then switch to N1/2.0 for the acceleration?

[seanmcleod]

@Octal450 @bcoconni any suggestions in terms of a realistic spin up acceleration figure to use when N1 = 0 and the engine is off? I guess N1_spinup may be too large since I assume that's based on the APU starting up the engine?

[Octal450]

Sorry I've been away from home for my medical reasons.

Can you be more specific, is the value for starting the engine? I've already found the values are pretty wrong by default for that
If you mean for windmilling the speed is already acceptable when after start & shutdown then moving to get it to windmill. So I really think the biggest problem right now is getting it to do that (the root of this issue)

I will play with the actual values regarding the other issue (windmill value) and let you know on that when I get the chance.

Kind Regards,
Josh

[seanmcleod]

The particular issue at the moment is that when N1 = 0, applies to N2 as well, and the engine is in the off state then the third argument to Seek is the acceleration value to be used in order to seek from it's current value to the target value, in this case in.qbar/10.0.

  if (disableWindmill == false) {
    N1 = Seek(&N1, in.qbar/10.0, N1/2.0, N1/N1_spindown);
    N2 = Seek(&N2, in.qbar/15.0, N2/2.0, N2/N2_spindown);
  }

So since N1 is 0, the acceleration value N1/2.0 is also 0, so N1 never advances towards in.qbar/10.0.

[seanmcleod]

Just a reminder to possibly address issue #195 at the same time when we fix this issue.

[Octal450]

Yes sir, I am currently awaiting responses from real pilot, and possibly from a mechanic.

Kind Regards,
Josh

[bcoconni]

Alternatively we could add a shift to N1/2.0 and N2/2.0 such as N1/2.0+0.1 so that the resulting value is never 0.0

[Octal450]

I would suggest don't shift, clamp. And maybe 0.01. 0.1 is enough to show up on most aircraft displays.

Kind Regards,
Josh

[seanmcleod]

Just to be clear the addition of 0.1 that @bcoconni is suggesting is to the acceleration value as opposed to some direct N1/N2 value to be displayed.

So for example if N1 was say currently 0.5 then the acceleration value previously would've been 0.25 and now it would be 0.35.

[Octal450]

Ah I understand, thanks.
Josh

[bcoconni]

Yes and the constant 0.1 I mentioned is completely arbitrary. I guess this needs some testing before selecting a value.

[bcoconni]

I'd suggest this issue being merged with #195 as I understand they tightly linked.

I'd suggest to close #195 and keep this one open, what do you think ?

[seanmcleod]

I'd suggest that we add the 0.1 increment to the acceleration value, push a commit for that and close this issue, and leave #195 open until we get some data to decide on what the target value should be for N1/N2 under windmilling conditions.

[Octal450]

I agree.

I will test this and see if I can figure it out shortly on my build FGFS.

Kind Regards,
Josh

[seanmcleod]

I did do a local test earlier today with the 0.1 shift, running the 737 script with the aircraft accelerating down the runway with only 1 engine started and confirmed that the non-running engine's N1 increased and matched qbar/10.0

[bcoconni]

PR #509 has been pushed to the code base.
Please test the latest code from the master branch and let us know if that fixes the issue.

[Octal450]

Compiling now with the fix.

[Octal450]

Fixed, works perfectly, thank you gents!

Kind Regards,
Josh

[bcoconni]

FYI the fix is now available on the FlightGear next branch.

[Octal450]

Thank you kindly!

Josh