Turboprop engine tuning

Roy Holmes

Resource contributor
Part 1. Start up issues.
Just about everyone complains about the awful starting behaviour of the sim Turboprop engine model.
Tales of airplanes tearing across the countryside because the parking brake was off on start are legion.
Noble efforts have been made to fix the problem, most noticeably by Heretic controlling the corrected fuel flow.

Last February I started working on the FDE for Dino Cattaneo's C-2 Greyhound. Since it is a carrier borne airplane it is more than slightly important that it behaves on start-up.
The airplane has two big turboprops with 8- bladed props. The engines are Allison T56-A-425 turboprops with 4,600 shp each.
Static Prop thrust is 800 lbs at Low speed Ground Idle ,1800 lbs at normal ground idle and 5000 lbs at full power.

So the first and biggest challenge was how to lose that 1000 lbs prop thrust by selecting Low Speed Ground Idle (LSGI).

Initially I adopted Heretic's approach of writing to Corrected Fuel Flow. I used XML Vars to set it to 330 lbs/hr though the actual value was 344.
That gave me idle prop thrust of 650 lbs at zero throttle, but opening the PLA to 33% kept the same fuel flow and increased the prop thrust to 1100, just enough to get the airplane to taxi.

This is where the second problem bites you. As soon as the airplane moves, prop thrust doubles and off you go.
The reason that happens is complex but very interesting and takes a while to explain, so I'll leave that for Part 2 in this post. Hint it, has to do with the prop tables records 511 and 512.

With the fuel flow method I could not match the PLA effects on prop thrust and Dino did not want people to have to install anything other than the airplane, so I decided to control things with mixture settings if I could. So I started on a long series of engine tests varying the mixture and checking its effects on:
N1, N2, Prop thrust, Fuel flow, Prop RPM, PBeta (prop pitch angle), Prop setting percent max torque and SHP.
What I found was that you needed 2.4% mixture to maintain ignition, below 5.4% the Engine held idle and above that value all the power indications began to rise.
With the PLA shut the power increased with mixture increase up to 100%. Above 90% the Prop RPM stabilized at 1105 which was the controlled 100% value.
I also set the propellor pitch control to 60%
Using Mixture and 60% prop pitch I was able to match the manual static prop thrust in LSGI and with LGSI off on take-off and in flight.. The airplane has two LSGI switches on the center instrument panel that affect (L:LOW SPEED GND IDLE LEFT,number) and right.
The code used was in an update section as follows:
if{ 9829 (>K:pROP_PITCH1_SET) 950 (>K:MIXTURE1_SET) }
els{ 16383 (>K:pROP_PITCH1_SET) 16383 (>K:MIXTURE1_SET) }
if{ 9829 (>K:pROP_PITCH2_SET) 950 (>K:MIXTURE2_SET) }
els{ 16383 (>K:pROP_PITCH2_SET) 16383 (>K:MIXTURE2_SET) }

With LSGI ON the start behaviour is as follows:
N1 peaks at 80% at 12 seconds after start begins
N1 has a stable idle at 65% and 18 seconds
Peak torque is 50 with 8 at idle
Peak prop thrust is 2400.
Idle prop thrust is 340
Idle prop RPM is 959 which means it is not in constant speed mode, the constant speed algorithm is inactive

With LSGI switched OFF
N1 peaks at 95% at 12 seconds after start begins
N1 has a stable idle at 70% and 30 seconds
Peak torque is 80 with 20 at idle
Peak prop thrust is 2900.
Idle prop thrust is 830
Idle prop RPM is 1105 which means the constant speed mode algorithm is actve, which causes rev instability

In general the effect is a much more controlled start with idle achieved in about half the time. It is done using the prop pitch and mixture controls.
Mixture in this context is variable high pressure fuel feed. It is a bit like what you had to do to start the Meteor Derwent engines, open the HP cock half way and slowly feed it forward as the engine lights.

I checked on the fuel flow gain influence and found the following.
No stated fuel flow gain gives no engine ignition.
I could not see a measurable difference between a single value and the multiple values recently discussed in the forum. That is not to say the multiple values are not worth considering, just that I did not see a difference.

Part 2. The constant speed issue.
Most of the wild overreactions in terms of power in the sim turboprops come from the issue of constant speed or rather, lack thereof.
To better understand this issue it is first necessary to understand a bit about propellors.
The function of a propellor is to convert engine power into prop thrust.
For a given situation with a certain amount of input power, the propellor will turn at a certain RPM. The power required to attain that RPM is absorbed by the propellor in driving air backwards, creating thrust. The amount of air driven is a function of several things but the main one is the pitch angle of the blade. If the blade angle is increased it takes more power to drive it at a set RPM and it will tend to slow down. Conversely if the blade angle is decreased it takes less power and the prop will speed up. A constant speed mechanism adjusts the blade angle until the prop is turning at the desired RPM.
There are theoretical tables of propellor power coefficients where you can enter with the blade angle and Advance ratio and get the power coefficient required. The sim does just that in table 512 and gets the power required which is then applied to table 511 to come up with the blade angle, and so on keeping the RPM constant.

Now I have simplified that because we have to consider the advance ratio and I skipped over that bit. Advance ratio is essentially how far the propellor moves through the air compared to how far it would move just considering its blade angle alone. The last case could be considered 100% efficiency and the advance ratio is less than that. Advance ratio (called J) is the airplane velocity in ft/second divided by the prop diameter in feet times prop revs/second
Thrust = (Engine Power/velocity) * efficiency. which means that efficiency = thrust/engine power* velocity. The problem with all this is that when the airplane is static advance ratio is zero as is propellor efficiency, so the nice constant speed equations fail. However when static, like on start up, the surge puts the prop equations in a situation when the constant speed tries to work and that accounts for much of the unsteadiness.
Worse still, when you have everything finally steadied and you go to taxi, velocity is no longer zero and efficiency comes back from zero and a surge occurs that send you tearing down the taxiway.

Controlling the mixture by and large keeps the prop at its pitch stop which helps reduce surges.

If anyone finds this interesting, I can expand it in a lot more detail with test results and stick it in the Wiki.
Please let me know


Resource contributor

Indeed it is interesting. However, and forgive me for being brutally honest, it is not possible to properly model a turboprop engine in FSX/P3D wtihout using external tables. That means a basic Simconnect connection (the least) either using XMLTools/any other C to XML interfase, or coding it in C++. Being the turboprop engine a single concept, there are so many variants that is impossible for the sim to cover them all with the default events/variables and FDE records.

Last edited:

Roy Holmes

Resource contributor
Tom, I think what you are saying is that it is not possible to model all the variants of TurboProp using the basic, single sim version. I agree. T
However I was not trying to do that, I just had one engine type to consider and worked on that with some success.
I did start with XML Tools while I tested out the fuel flow approach. When I changed to mixture it was not necessary to use XML Tools because I was adjusting a stand-alone engine parameter, mixture. This can be varied at will and there is no feedback from the engine operations. Fuel flow is a calculated parameter from thrust and Thrust specific fuel flow, so it is subject to change by the aircraft engines. As such, if you wish to have a fuel flow different from what the engine will calculate, you need to step outside the norms and use SimConnect or XML Tools.
The most important tables in the turboprop mechanisation are the prop ones and they are basied on established theory. I only had to scale up the power corefficient from the basic King Air tables to match the much more powerful prop system in the C-2 and fix an issue where the addition of a 10° blade angle curve in the wrong place, between 15 and 25 degrees was giving nonsensical results.

When I have some time I'll expand on what I did with the engine and the prop tables, I think it is interesting and I was able to give the C-2 performance within 5% of its performance manual across the board. That was my aim.


Resource contributor
Tom, I think what you are saying is that it is not possible to model all the variants of TurboProp using the basic, single sim version. I agree. T
Roy, I am saying NO ONE turboprop engine can be realistically modeled using only the basic sim version, not even the one you are working on.
Proofs are the numbers you posted of N1, Torque and RPM at startup.
There should be no N1 peaks at all, N1 rpms increase steady from light off to idle (60-70%) and stay there while idle condition remains.
There should be no Torque peaks at all, torque is almost nil during startup and idle due to props starting from lock position (~ 2 degrees of pitch).
There should be no RPM peaks either.

I'm sorry to make perhaps disturbing comments, but I prefer to be honest and tell what I've learned after a couple of years of FSIM turboprop engine design (research and code).


Roy Holmes

Resource contributor
I fully appreciate you speaking the truth as you see it.
As I see it, compared to the standard model, controlling the mixture has halved the start-up and ground idle/taxi instability which was the first thing I mentioned as a widespread complaint about turboprops in the sim.
That was just my first point, I did a lot of work on the propeller tables, PLA to SHP matching TAS effects, altitude effects on torque all of which are pretty realistic now.
I was encouraged by the developer to find solutions that did not involve having to install an external program.
As I said I found I got better results with mixture control which does not need Simconnect or any other external program than I did with fuel flow control which does.
Torque in the C-2/E-2 at ground idle is 9%, SHP is 434, Prop thrust is 800 lbs plus and the blade angle is 5 degrees. These are quite different to the figures you quoted, so it is a different engine.
Yes my start up is too rapid I'm getting idle at 18 seconds and it should be less than 35. But the prop thrust values and the airplane performance are a very good match with the manual figures.
So just because the start up and stability is a bit off does not, in my view show that the engine can not be realistically modeled. I'm more interested in flight performance than anything else.
The whole PLA set-up is somewhat unrealistic with a standard throttle in the flight idle/ground idle/reverse thrust areas, that is a bigger limitation in my view that start up/taxi instability which I have reduced to manageable proportions.
I have not seen any of the work you have done for the past two years so I can not comment on it. Which tables have you modeled externally?
That would be an interesting input, and somewhat more helpful that a blunt statement that NO turboprop can be realistically modeled without external tables. Which aspects of the engine performance absolutely need external tables?
I'd like to hear more, this was my first turbo-prop and I'm keen to learn from your experience.


Resource contributor

It seems we have different visions in what we consider an engine performance as realistic.
To me, it means the engine must behave like the real one, as much as possible, in every phase of flight: startup,flight and shutdown, including intermediate steps.
The idea is that an RW pilot should see no difference both in engine performance and gauge readings, with respect to what he sees when flying the real aircraft.

Actually default propeller tables in .air file are a good source to start when needing to adapt them for proper values of thrust/efficience/etc. Once they are on the spot it is necessary to start working in all the rest.

I cannot comment with details what I've done as those are commercial projects, but I cant give you some ideas on what should be custom controlled:

Custom control (external tables) of CN1 spool up rate, fuel flow rate, electrical source available, prop RPM, Oil pressure, Oil temperature, ITT, with all their relationships.
Custom control of Avars for Throttle/Prop/Mixture. Custom relationship between Trottle Lever and CN1. Custom relationship between Prop lever and Prop RPM. etc
Custom feathering control-Proper feather/unfeather spool rate.

Custom Fuel Flow managing. Custom ITT/Oil values depending on ambient situations. Custom torque display values depending on different phases of flight.Etc.

-Inflight Engine Out
Proper engine display values for normal restart, windmilling restart, etc. Custom prop feathering. Etc.

Custom spool down rates, depending on flight phases. Custom feathering/windmilling situations. Custom ITT/Oil temp-press values in a shut engine. Etc.

-In all cases, custom trapping of unwanted user commands that might interfere with internal handling of engine parameters (for example, capture user operation of prop lever(s)).

Just a few lines, but give a general idea, I hope.
I understand it is not possible to make the whole thing 100 % real, as there are no two RW engines that behave exactly the same, but indeed it can be done pretty much alike when using external resources. Please notice that I'm not saying .air/.cfg FDE files must be discarded, just complemented with corrective code, when necessary.

Anyway, the expressed is nothing more than my POV, it is perfect if you are happy with your results and they meet the requirements of the project.
With my statement (blunt indeed, you're right) I just tried to express that it is recommendable to go way off the default sim parameters to deliver ANY realistic turboprop model, something that is being required by FS market since first versions of MS Flight Sim, nowadays with almost nil options to choose from.


Roy Holmes

Resource contributor
My task with the C-2 was to develop the flight dynamics and match the Manual performance figures as close as possible. That involves setting up the engines to give the appropriate performance and creating reasonable behaviour as far as flying it is concerned. With the turboprop that involved improving start up and taxi characteristics.
My task did not include systems integration display etc.
So yes I have a different approach. I did the work in 3 months or less, fitting it in with other projects. In two years I have done over eight different FDE packages and that limits the amount of effort on each of them.
Proofs are the numbers you posted of N1, Torque and RPM at startup.
There should be no N1 peaks at all, N1 rpms increase steady from light off to idle (60-70%) and stay there while idle condition remains.
There should be no Torque peaks at all, torque is almost nil during startup and idle due to props starting from lock position (~ 2 degrees of pitch).
There should be no RPM peaks either.
All these mentioned points can be easily corrected since (at least) P3Dv4.3 with additional custom fuel_flow_gain entries in the cfg file.
Last edited:

Roy Holmes

Resource contributor
All these mentioned points can be easily corrected since (at least) P3Dv4.3 with additional custom fuel_flow_gain entries in the cfg file.
I have yet to see any examples of how that can be done.
If you found it easy maybe you could share how you did it?
The C-2/E-2 are not exclusive to P3Dv4.3 so the reported fix would not be appropriate, but I did try briefly and found it not very effective. I probably did not spend enough time on it to come to a definitive conclusion.
However, the major instability seems to come from the Prop constant speed implementation and not necessarily from N1 overshoot on start up. Controlling mixture kept the start cycle Prop RPM below the constant speed threshold.
If you found it easy maybe you could share how you did it?
Like with table 1505 on jets ;) Quite a few hours of trial and error by testing countless combinations.
IMO it's just the fuel flow that needs to be reduced during start up to achieve a realistic spool up.
I did use the fuel flow reduction method already many years ago in FSX with a payware Metro III.
For adjustment I start with the lowest RPM, e.g. 35% and then I'm working upward and experiment which is the next higher RPM which requires a higher fuel flow, and so on, until just below idle RPM.
AFAIR you can use up to 5 different values, but testing with 4 different values was time consuming enough and did result in a very nice engine start.
If the spool up is slow enough, a prop governer / constant speed 'interference' doesn't happen, or at least I haven't noticed it. It's still new terrain for me as well.
Equally, if not even more important, is that you can significantly reduce the engine response time during normal ops.
Until the is mod I always had to use the compromise of a too fast engine start and a too slow engine response when running.
Last edited:

Roy Holmes

Resource contributor
Like with table 1505 on jets
I did a few runs to get the direction best suited to slow spool up and a couple more to improve it and ended up with:
fuel_flow_gain.0 = 25.0, 0.0001
fuel_flow_gain.1 = 35.0, 0.001
fuel_flow_gain.2 = 50.0, 0.005
fuel_flow_gain.3 = 65.0, 0.01

If the 25% value was much less than 0.0001 combustion was not achieved. 62% is the idle N1
I'm sure I can improve things with a few more iterations and data study, but this PLUS the mixture control gave a nice slow build up and no more than about 2% overshoot.

I gave it the parking brake test, left it off and did a simultaneous start on both engines. The beast did not budge an inch.
Good first try but I'll document things in more detail before declaring success.
Interesting values when comparing them to e.g. my updated PC-12.
fuel_flow_gain.0 = 35, 0.001
fuel_flow_gain.1 = 40, 0.01
fuel_flow_gain.2 = 60, 0.02
fuel_flow_gain.3 = 66, 0.04
Interesting thread, and it's years since I visited this Forum :)

There's yet another old problema with turboprops in MSFS / P3D - the supposedly modeled PT-6 free-running turbine creats FF variations at constant altitude and throttle / condition settings when Prop RPM is varied - it shouldn't!

Actually so far the only desktop sim I've used that properly models this feature of a free-runing ( PT-6 like ) trubine is ELITE IFT and it's B200. Btw even X-plane 11.30 about to be released with some additional fine tuning of turboprops doesn't model it correctly!

Roy Holmes

Resource contributor
Fuel flow varies with Shaft Horse Power (SHP). So if the fuel flow increases it must be because SHP has increased. Prop RPM is constant and is controlled by the constant speed equations in the sim. There are events called Prop RPM such as CTL +F1 which is called Propeller (low RPM) but what it actually does is increase propeller pitch angle. The SDK description of the event is an increase in prop pitch. An increase in pitch angle affects the constant speed mechanism by increasing power required to maintain constant speed so the SHP has to increase and hence fuel flow increases. This is controlled by Table 512 Propeller Power Coefficient in the .air file.
Essentially, in normal flight there is no need to use the Prop RPM commands because the constant speed mechanism is doing it when necessary.
As far as the sim is concerned this behaviour is normal. My flight experience was mainly pure jets and a small amount of Piston props, but I did fly the HS748 on a couple of HUD trials after I left the service. I was not aware to any extent what was happening power wise, more interested in maintaining an accurate glide slope using the HUD.
What you are seeing would I believe also occur in real life because the constant speed mechanism principle is exactly the same. However it could be hidden by some fancy electronic control.

that's because MSFS's turboprop model for a free-running turbine like the Pt-6 used on a Kingair is wrong ( actually just as in X-plane, even with the upcoming 11.30 update to the turboprop model, but Austin told me he'll try to address it in future updates - he's turboprop also uses a Pt-6... )

SHP = Torque in FT/lbs * RPM * K
K = .00015

Power is controlled by the Power lever.

If one changes the prop RPM the Power does not chance so the torque must. If the RPM is decreased the Torque will rise. No Fuel flow variations... And it makes sense to reduce RPM when transitioning from climb to cruise, at constant throttle...

As I pointed out, you can check it correctly modeled in ELITE IFT FNPTs, as well as the FRASCAs...
that's because MSFS's turboprop model for a free-running turbine like the Pt-6 used on a Kingair is wrong
As I pointed out, you can check it correctly modeled in ELITE IFT FNPTs, as well as the FRASCAs...
I don't know where you've got the info from that FF shouldn't change, because IRL is definitely does.

PA-31T S.L. ISA 2000 RPM 246ktas 1628lbft = 754lb/hr
PA-31T S.L. ISA 1900 RPM 246ktas 1628lbft = 736lb/hr