Update FGTrimAnalysis wrt latest JSBSim Release

[rishi1906]

Greeting @agodemar ,

I am new to JSBSim, trying to use the trim solution provided by Dr Agostino De Marco as mentioned in his paper "A General Solution to the Aircraft Trim Problem". When I add the header and source files of FGTrimAnalysis and FGTrimAnalysisControl to CMakeLists.txt under /src/initialization folder to build, it throws several errors of incompatibility (missing/mismatched function).

Among several errors, one error looks like this

../../jsbsim-code/src/initialization/FGTrimAnalysis.cpp:335:20: error: cannot convert ‘std::shared_ptrJSBSim::FGInitialCondition’ to ‘JSBSim::FGInitialCondition*’ in assignment
335 | fgic=fdmex->GetIC();
| ~~~~~~~~~~~~^~
| |
| std::shared_ptrJSBSim::FGInitialCondition

@agodemar Sir, I have been reading and following your papers for quite some time. It's commendable the contribution/efforts made by your lab for the development of JSBSim and appreciate the recent work you have published titled
"A Deep Reinforcement Learning Control Approach for High-Performance Aircraft".

I request you to help me out with the FGTrimAnalysis code and mentioned error.

[agodemar]

@rishi1906 the FGTrimAnalysis related functionalities were written years ago, and you get errors like the one you posted above if you simply throw old code into the actual compilation workflow. This means some work is required to adapt that code to the new types we are using. If you have the possibility to do it, it shouldn't be so difficult, provided you know how to program in C++.
The compiler is so clear in the log you posted above:

In FGTrimAnalysis.cpp at line 335, there's a mismatch of types. you should modify FGTrimAnalysis.cpp by declaring fgic as std::shared_ptrJSBSim. Other errors will probably occur due to the same reason.

[rishi1906]

Thanks for the response. I anticipated the same, sure will fix the code according to the latest release of JSBSim, and yes I am comfortable with the programming environment.

[agodemar]

@rishi1906 that's great, please let me know about your progresses here in this thread.

Be advised that the trim analysis uses a variant of the Nelder-Mead algorithm that was published years ago by Virginia Torczon.
Nowadays, this algorithm is included into the fmincon and fminsearch functions (within the Optimization Toolbox).

[seanmcleod]

@rishi1906 are you planning on making use of JSBSim within a Matlab/Simulink environment via the S-Function interface, or are you planning on using JSBSim stand alone outside of a Matlab/Simulink environment?

[rishi1906]

Sir, I have started resolving the issues, I think I will be able to update them as mostly its conversion from string to SGPath and some updated Set and Get functions in addition to updating to shared_pointers as you rightly suggested.

Regarding, fmincon I suppose that's part of MATLAB environment. I am a core C++ guy.

@seanmcleod I am using JSBSim with Ardupilot SITL, so its outside MATLAB as I need to attach it with Software in the loop simulation and open source Optimizer routine.

[seanmcleod]

Thanks for the additional context. Out of interest is the current trim implementation in JSBSim not sufficient for what you need?

Regarding, fmincon I suppose that's part of MATLAB environment.

Yes, it's part of Matlab. I guess since you mentioned @agodemar's paper - "A Deep Reinforcement Learning Control Approach for High-Performance Aircraft" which made use of Matlab/Simulink he possibly assumed you were also planning on using Matlab/Simulink.

[rishi1906]

I debugged the current trim implementation in JSBSim, which uses a pairwise (state and control) method that does down the lane bisection method to update the guess, I am not fully sure why I am getting wdot not trimmable error for a custom aircraft I want to simulate in SITL. Work done by Professor @agodemar for general trim looks more promising for that matter. I am hoping it will resolve wdot no trimmable error.

The current trim solution in JSBSim, only solves for three states for example Ground Trim, but I guess all states should be trimmed according to the flight envelope as done @agodemar. I am surprised why current trim method has not been updated since 1999.

[agodemar]

@rishi1906 if you get good results by revamping the TrimAnalysis code, I hope you could pass the sources to us in order to have it back in the master branch.

Are you working in a forked repo?

[rishi1906]

Sure, I will share the updated code. Yes, I have forked working with local files. I will need your help with the code you wrote.

[seanmcleod]

I am surprised why current trim method has not been updated since 1999

That's why I was asking whether it wasn't sufficient for your needs, given it appears to have been for 90% of JSBSim users for the last 24 years 😉

If you can share your custom aircraft I could take a look and see if I can see what the trim issue is.

[rishi1906]

@seanmcleod that would be great! Till I get correct stability derivatives for modified Talon (referred to as custom aircraft), could you check the attached XML file of X8 RC UAV. Please find attached initial condition file. I have taken this XML file from github just to test it with JSBSim. For full trim I get following output

Full Trim

Trim failed
Trim Results:
Angle of Attack: 7.50 wdot: 3.07e+01 Tolerance: 1e-03 Failed
Throttle: 0.00 udot: -1.43e+00 Tolerance: 1e-03 Failed
Pitch Trim: 0.00 qdot: 7.79e-01 Tolerance: 1e-04 Failed
Roll Angle: 0.00 vdot: -1.01e-04 Tolerance: 1e-03 Passed
Ailerons: -0.00 pdot: 0.00e+00 Tolerance: 1e-04 Passed
Rudder: 0.00 rdot: 0.00e+00 Tolerance: 1e-04 Passed

x8.zip

I tried to trim it (in MATLAB) for level flight and got these values

AoA Elevator_deflection V_inf Thrust
____ ___________________ ______ ______

-0.5           1.1232          52.573    13.807
   0          0.77393          35.453    6.1801
 0.5           0.4247           28.52    3.9987
   1         0.075472           24.52    3.0018
 1.5         -0.27376          21.837    2.4545
   2         -0.62299          19.878    2.1254
 2.5         -0.97222          18.366    1.9188
   3          -1.3215          17.154    1.7878
 3.5          -1.6707          16.154    1.7067
   4          -2.0199          15.311    1.6604
 4.5          -2.3691          14.587     1.639
   5          -2.7184          13.957    1.6365
 5.5          -3.0676          13.402    1.6482
   6          -3.4168          12.909    1.6712
 6.5          -3.7661          12.466    1.7032
   7          -4.1153          12.066    1.7424
 7.5          -4.4645          11.702    1.7876
   8          -4.8138          11.369    1.8378
 8.5           -5.163          11.063    1.8921
   9          -5.5122          10.781      1.95
 9.5          -5.8615          10.519    2.0109
  10          -6.2107          10.275    2.0744

Note : I have been working on JSBSim since last 1 month only, kindly ignore if I don't know some obvious things.

[rishi1906]

Correction,

AoA Elevator_deflection V_inf Thrust
____ ___________________ ______ ______

-1.5            8.704          55.061    27.597
  -1           7.6942          36.515    12.118
-0.5           6.6844          29.234     7.833
   0           5.6746          25.076      5.87
 0.5           4.6648          22.303    4.7732
   1           3.6549          20.284    4.0933
 1.5           2.6451          18.729    3.6462
   2           1.6353          17.485    3.3419
 2.5           0.6255           16.46    3.1319
   3         -0.38432          15.596    2.9871
 3.5          -1.3941          14.856    2.8895
   4           -2.404          14.211     2.827
 4.5          -3.4138          13.644    2.7914
   5          -4.4236           13.14    2.7768
 5.5          -5.4334          12.688     2.779
   6          -6.4432          12.279    2.7948
 6.5           -7.453          11.908    2.8217
   7          -8.4629          11.568    2.8579
 7.5          -9.4727          11.256    2.9019
   8          -10.482          10.968    2.9525
 8.5          -11.492          10.701    3.0088
   9          -12.502          10.453    3.0699
 9.5          -13.512          10.221    3.1353
  10          -14.522          10.003    3.2044

<?xml version="1.0"?>
<initialize name="basic_ic">
  <!--
    Basic set of initial conditions at 1000 ft with engines running.
  -->
  <ubody unit="FT/SEC">     42.9462 </ubody> 
  <vbody unit="FT/SEC">     0.0  </vbody>
  <wbody unit="FT/SEC">     3.7573  </wbody>
  <longitude unit="DEG">    0.0 </longitude> 
  <latitude unit="DEG">     0.0  </latitude>
  <phi unit="DEG">          0.0  </phi>  <!-- roll -->
  <theta unit="DEG">        5.0  </theta>  <!-- pitch -->
  <psi unit="DEG">          0.0  </psi>  <!-- yaw -->
  <altitude unit="FT">      1000.0 </altitude>
  <elevation unit="FT">     0.00000001 </elevation> <!-- bug in JSBSim makes 0 cause NaNs -->
  <winddir unit ="DEG">     90.0 </winddir> <!-- wind direction -->
  <vwind unit ="FT/SEC">    0.0 </vwind> <!-- wind speed -->
  <running>                  -1 </running> <!-- starts all engines -->
</initialize>

 Full Trim
  Sorry, qdot doesn't appear to be trimmable

  Trim failed
  Trim Results: 
       Angle of Attack:   4.81  wdot: -1.01e+00 Tolerance: 1e-03  Failed
              Throttle:   0.00  udot: -2.52e+00 Tolerance: 1e-03  Failed
            Pitch Trim:   0.00  qdot: -2.86e+00 Tolerance: 1e-04  Failed
            Roll Angle:   0.00  vdot: -8.20e-10 Tolerance: 1e-03  Passed
              Ailerons:   0.00  pdot:  1.29e-14 Tolerance: 1e-04  Passed
                Rudder:   0.00  rdot: -1.45e-14 Tolerance: 1e-04  Passed

[seanmcleod]

You didn't supply the engine file.

  Propulsion:
No filename given.


In file ../aircraft/x8/x8.xml: line 79
Could not open file: electric800W.xml

Aircraft propulsion element has problems in file Path "../aircraft/x8/x8.xml"

Is it identical to the existing electric engine files in JSBSim, just with the power set to 800W?

In your Matlab examples above, are all of the entries valid trim points?

What is V_inf? And what units?

Also the units for the Thrust column?

[rishi1906]

Oh yes it's a simple JSBSim electric engine file.

<?xml version="1.0"?>

<electric_engine name="electric">
 <!-- This is a basic model of the electric motor used to power the x8
 -->
 <power unit="WATTS">  800.0 </power>
</electric_engine>

Yes, all are valid trim points.
V_inf is velocity in x direction (basically ubody in init file (XML), V_inf is in m/sec.
Thrust is in Newton.
Attaching matlab code.

% code to tabulate the equilibrium conditions of X8 aircraft

clear all; close all; clc

S = 0.75;           % Wing area in m2
m = 3.364;          % Aircraft weight in kg
rho = 1.225;        % Density of air in kg/m3
g = 9.81;           % Acceleration due to gravity in m/s2

% Aerodynamic coefficients

CL0 = 0.0867;
CLalp = 4.0203;
CLde = 0.2781;

CD0 = 0.0197;
CDalp1 = 0.079;
CDalp2 = 1.0555;
CDde = 0.0633;

Cm0 = 0.0227;
Cmalp = -0.4629;
Cmde = -0.2292;

% CD0 = 0.0107;
% CDalp1 = -0.00955;
% CDalp2 = 1.1;
% CDde = 0.0196;
% 
% CL0 = 0.0477;
% CLalp = 4.06;
% CLde = 0.7;
% 
% Cm0 = 0.00439;
% Cmalp = -0.227;
% Cmde = -0.325;

% Assume different values of alpha
AoA = -1.5:0.5:10;         % aoa in degrees
alpha = AoA*pi/180;     % aoa in radians

% zero moment
de = -(Cm0 + Cmalp*alpha)/Cmde;
Elevator_deflection = de*180/pi;        % in degrees

% Lift balance
Vsq = (m*g/(0.5*rho*S))./(CL0 + CLalp*alpha + CLde*de);
V_inf = sqrt(Vsq);

% Thrust measurement in Newtons
Thrust = (0.5*Vsq*rho*S).*(CD0 + CDalp1*alpha + CDalp2*(alpha.^2) + CDde*(de.^2));

AoA = AoA';
Elevator_deflection = Elevator_deflection';
V_inf = V_inf';
Thrust = Thrust';

X8_Equilibrium = table(AoA, Elevator_deflection, V_inf, Thrust)

[seanmcleod]

Ah, so you're not actually getting Matlab to come up with a trim solution, you've come up with a calculation to for a given AoA work out the elevator deflection needed to zero out the pitch moment due to $C_{m{_\alpha}}$ then the airspeed needed for the lift to equal the weight for the given alpha and finally the thrust needed for the drag for the given alpha.

You're assuming the MRP (Moment Reference Point) is coincident with the cg, which is the case for this model, although I see a lot of stuff is coincident at (0, 0, 0), the cg, the MRP and the engine thruster.

One minor thing you haven't taken into account in your calculation is the thrust angle.

Currently JSBSim is saying qdot doesn't appear to be trimmable, i.e. it isn't able to zero out the pitch moment.

[rishi1906]

Ah, so you're not actually getting Matlab to come up with a trim solution, you've come up with a calculation to for a given AoA work out the elevator deflection needed to zero out the pitch moment due to then the airspeed needed for the lift to equal the weight for the given alpha and finally the thrust needed for the drag for the given alpha.

Yes, you are right, it should work, isn't it?

X8 has blended wing design with pusher configuration, shall I change location of the thruster to be at the end of fuselage?

One minor thing you haven't taken into account in your calculation is the thrust angle.

I have taken those equations from the paper mentioned in the reference in x8.xml file.
"Aerodynamic modeling of the Skywalker X8 Fixed-Wing Unmanned Aerial Vehicle"

The thruster is oriented at 0,0,0 deg.

[seanmcleod]

The thruster is oriented at 0,0,0 deg.

Yes it is, but if you're trimming for level flight gamma = 0 and say an AoA of 10 deg, then the pitch attitude, theta = 10 deg nd then the thrust line is oriented at 10 deg now relative to the ground. So some of the thrust vector is now providing a force in the lift direction, so you would need slightly less lift, but also slightly more thrust since the drag is parallel to the flight path etc. A trim solution needs to take this into account.

[seanmcleod]

One thing I noticed is that the aircraft seems to be very overpowered with an 800W electric engine, so trim wise it's trying to trim around throttle commands on the order of 0.00001.

So I dropped the power down to 4W.

So now you can see that the trim solution is looking at a throttle command of around 0.23. But still failing.

  Full Trim
  Sorry, qdot doesn't appear to be trimmable

  Trim failed
  Trim Results: 
       Angle of Attack:   2.97  wdot: -2.57e-01 Tolerance: 1e-03  Failed
              Throttle:   0.23  udot: -7.11e-02 Tolerance: 1e-03  Failed
            Pitch Trim:   0.00  qdot: -6.66e-05 Tolerance: 1e-04  Passed
            Roll Angle:   0.00  vdot: -3.79e-19 Tolerance: 1e-03  Passed
              Ailerons:   0.00  pdot:  0.00e+00 Tolerance: 1e-04  Passed
                Rudder:   0.00  rdot:  0.00e+00 Tolerance: 1e-04  Passed

I used the following python code to get a feel for what is happening, which is how I noticed how over-powered the aircraft is.

You'll see I've used your 3 deg AoA trim point to choose an airspeed and elevator angle, and then run it for 14 time steps to see how much deviation there is.

import jsbsim

fdm = jsbsim.FGFDMExec('..\\') # The path supplied to FGFDMExec is the location of the folders "aircraft", "engines" and "systems"

fdm.load_model('x8') 

# Set engines running
fdm['propulsion/engine/set-running'] = 1

fdm['ic/h-sl-ft'] = 1
fdm['ic/vc-kts'] = 30.31882
fdm['ic/gamma-deg'] = 0
fdm['ic/alpha-deg'] = 3
fdm['ic/theta-deg'] = 3

fdm['fcs/throttle-cmd-norm'] = 0.225
fdm['fcs/elevator-cmd-norm'] = -0.012810666

fdm.run_ic() # Initialize the aircraft with initial conditions

#fdm['simulation/do_simple_trim'] = 1

elevator_pos = fdm['fcs/elevator-pos-rad']

cm0 = fdm['aero/coefficient/Cmo']
cm_alpha = fdm['aero/coefficient/Cmalpha']
cm_q = fdm['aero/coefficient/Cmq']
cm_de = fdm['aero/coefficient/CmDe']

alpha = fdm['aero/alpha-deg']

for i in range(14):
    alpha = fdm['aero/alpha-deg']
    gamma = fdm['flight-path/gamma-deg']
    cm0 = fdm['aero/coefficient/Cmo']
    cm_alpha = fdm['aero/coefficient/Cmalpha']
    cm_q = fdm['aero/coefficient/Cmq']
    cm_de = fdm['aero/coefficient/CmDe']
    
    M = fdm['moments/m-total-lbsft']
    aero_M = fdm['moments/m-aero-lbsft']
    propulsion_M = fdm['moments/m-prop-lbsft']
    fdm.run()

# Trim
try:
    fdm['simulation/do_simple_trim'] = 1

except RuntimeError as e:
    # The trim cannot succeed. Just make sure that the raised exception
    # is due to the trim failure otherwise rethrow.
    if e.args[0] != 'Trim Failed':
        raise
    
print('Done')

You can then also take a look at the JSBoutX8.csv file to look at a number of other states to see how they change over 0.1s.

You'll notice that there is a slight pitch up, a combination I think of slightly too much thrust giving a slight positive climb angle, which decreases AoA slightly, which means Cm-alpha decreases and so Cm_de results in a slight pitch rate etc.

[rishi1906]

@seanmcleod sorry for the late reply, yes what you mentioned is correct, it's indeed overpowered.

I rechecked the configuration file and found some errors, CG locations, AERORP loc, locations of engine and thruster, and in controls as well some inputs were missing. JSBSim can do full trim successfully for updated configuration wrt the initial conditions provided as a solution from Matlab code.

Full Trim

 Trim successful
 Trim Results: 
      Angle of Attack:   0.48  wdot:  3.80e-04 Tolerance: 1e-03  Passed
             Throttle:   0.23  udot: -8.97e-05 Tolerance: 1e-03  Passed
           Pitch Trim:  -0.02  qdot:  6.39e-16 Tolerance: 1e-04  Passed
           Roll Angle:   0.00  vdot: -4.36e-17 Tolerance: 1e-03  Passed
             Ailerons:  -0.00  pdot:  1.58e-17 Tolerance: 1e-04  Passed
               Rudder:   0.00  rdot: -4.37e-13 Tolerance: 1e-04  Passed

Even upon getting trimmed, it doesn't maintain level flight in simulation (Flightgear), and altitude keeps dropping.
PFA updated XML and CSV files here. Could you check and let me know?

x8.zip
JSBout.csv

[seanmcleod]

Take a look at the last column Direct Thrust (engine 0 in lbs) in JSBout.csv it's 0.

What does your script/process look like in terms of performing the trim etc.?

It looks like the trim routine is figuring out how much thrust is required but then when you continue running the engine isn't running and/or the throttle has been reset to 0.

[rishi1906]

When I run the same script with standalone JSBSim code (without SITL) it gives a non-zero Direct Thrust value and maintains the height which indicates the trim is working correctly.

But same initial conditions with same configuration file when used in SITL behaves differently, what I has shared before. Any idea regarding Arduplane SITL ?

[seanmcleod]

Who is controlling the throttle during SITL? Log the throttle command and engine state during SITL.

[rishi1906]

That's what I need to figure out. Will update you once I get figure it out.

[seanmcleod]

Maybe you're getting caught out with ArduPilot's motor arming?

https://ardupilot.org/copter/docs/arming_the_motors.html

[rishi1906]

Yes, I figured it out today, it's due to arming only, as before arming Ardupilot doesn't give control command to JSBSim. It's working now (I had to tune the aircraft though using AutoTune mode in Ardupilot).
I am facing a strange issue now. When I do full trim for X8 wrt following initial (trim) conditions for level flight, it works fine.

<?xml version="1.0"?>
<initialize name="basic_ic">
  <!--
    Basic set of initial conditions at 1000 ft with engines running.
  -->
  <ubody unit="M/SEC">     25.076 </ubody>
    <!-- <ubody unit="M/SEC">     50.00 </ubody> -->
  <vbody unit="FT/SEC">     0.0  </vbody>
  <wbody unit="FT/SEC">     0.0  </wbody>
  <!-- <longitude unit="DEG">    -122.372586 </longitude> 
  <latitude unit="DEG">     37.620397  </latitude> -->
  <longitude unit="DEG">    0.0 </longitude> 
  <latitude unit="DEG">     0.0  </latitude>
  <phi unit="DEG">          0.0  </phi>  <!-- roll -->
  <theta unit="DEG">        0.0  </theta>  <!-- pitch -->
  <psi unit="DEG">          0.0  </psi>  <!-- yaw -->
  <alpha unit="DEG">        0.0  </alpha>  <!-- alpha -->
  <altitude unit="FT">      1000.0 </altitude>
  <elevation unit="FT">     0.00000001 </elevation> <!-- bug in JSBSim makes 0 cause NaNs -->
  <winddir unit ="DEG">     90.0 </winddir> <!-- wind direction -->
  <vwind unit ="FT/SEC">    0.0 </vwind> <!-- wind speed -->
  <running>                  1 </running> <!-- starts all engines -->
</initialize>

But if I change lat long (say KSFO), trim fails. Trim shouldn't fail for different lat, and long, right?

[seanmcleod]

Welcome to the non-flat world 😉

Graph what the trim routine computes for fcs/aileron-cmd-norm as you slowly increase the latitude from 0 deg to just over 7 deg when the trim starts failing.

The trim routine then fails with:

rdot doesn't appear to be trimmable

I haven't looked in detail, but what I'm guessing is that the increase in aileron deflection is generating a yaw moment which the trim routine can't cancel out with rudder sucessfully. So double-check the lateral and directional stability coefficients are 'correct'.

[seanmcleod]

Ah, I just took a quick glance at the X8 FDM, and I noticed that there is no yaw channel. So as soon as a yaw moment develops, e.g. due to aileron deflection, then the trim routine can't cancel it out with rudder, because there is no rudder/yaw control!

            <function name="aero/coefficient/Cnda">
                <description>Yaw moment due to aileron</description>
                <product>
                    <property>aero/qbar-area</property>
                    <property>metrics/bw-ft</property>
                    <property>fcs/effective-aileron-pos</property>
                    <value>0.1202</value>
                </product>
            </function>

[agodemar]

@seanmcleod good catch!

[rishi1906]

How does yaw moment generated due to aileron deflection depends on lat long?

How come with no rudder control for (0,0)lat,long, its able to trim rdot ?

X8 has blended wing config with no rudder control. What could be the way around to this problem ?

[seanmcleod]

X8 has blended wing config with no rudder control. What could be the way around to this problem ?

Take a look at the B-2, it uses split elevons/flaperons for yaw control.

Or the B-21, comments from Rodney Robles on his LinkedIn post.

I was delighted to see the maiden flight of the B-21 hashtag#Rider yesterday‼️

In a photo posted by The Avionist, there was something that got my hashtag#controlgeek attention, the two inboard flaperons deflected downwards and the outboard flaperon deflected upwards.

With the B-21 control distribution, if you want a high lift configuration, you have to deflect the flaperons down, those which are in the inboard that can generate more lift, and to compensate for the pitch down induced by the previous, you have to deflect the outboard (the one having more pitch control power) up, having a total increase of lift, meanwhile maintaining a trimmed condition.

However, there's a hidden good side effect here. This outer up, inner down flap distribution can generate hashtag#differential hashtag#drag and also can increase hashtag#directional hashtag#static hashtag#stability (CnBeta>0), just like the B-2 did with the drag rudder (split flaps).

I love hashtag#engineering ingenuity!

Now taking a quick look at the X-8 it looks like it only has two elevons, one on the port side and one on the starboard side. So it would be trickier to balance the pitch, roll and yaw moments required via differential movement.

[rishi1906]

Thanks for sharing, I agree using split elevons/flaperons is a good choice for yaw correction in blended wing design.

[seanmcleod]

Except that the actual X8 aircraft doesn't have split elevons/flaperons.

[rishi1906]

@seanmcleod could you guide me in deciding the parameters for the initialization file for the purpose of ground trim?
JSBSim is not able to trim X8 using Ground Trim even if I give everything zero basically no movement at all (it is supposed to be just standing on the runway).

<?xml version="1.0"?>
<initialize name="Start up at Canberra Model Aircraft Club">
  <latitude unit="DEG"> 0.0 </latitude>
  <longitude unit="DEG"> 0.0 </longitude>
  <altitude unit="M"> 0.0 </altitude>
  <vt unit="FT/SEC">      0.0 </vt>
  <gamma unit="DEG">      0.0 </gamma>
  <phi unit="DEG">        0.0 </phi>
  <theta unit="DEG">      0.0 </theta>
  <psi unit="DEG">        0.0 </psi>
</initialize>

It gives wdot not trimmable error

 Sorry, wdot doesn't appear to be trimmable
  Trim failed
  Trim Results: 
          Altitude AGL:   0.00  wdot:  3.21e+01 Tolerance: 1e-03  Failed
           Pitch Angle:   0.00  qdot:  2.92e-08 Tolerance: 1e-04  Passed
            Roll Angle:   0.00  pdot:  0.00e+00 Tolerance: 1e-04  Passed

Could it be due to AERORP and CG locations ?

[rishi1906]

Just resolved this, by giving a ground reaction to X8. but getting the following error while pre-arm check is performed in SITL.

AP: PreArm: AHRS: EKF3 Roll/Pitch inconsistent 41 deg. Wait or rebo
AP: PreArm: Check mag field (xy diff:260>100)

[seanmcleod]

FYI in terms of the trim routine in JSBSim.

jsbsim/src/initialization/FGTrim.h

Lines 74 to 94 in 5c31c25

	/** The trimming routine for JSBSim.
	FGTrim finds the aircraft attitude and control settings needed to maintain
	the steady state described by the FGInitialCondition object . It does this
	iteratively by assigning a control to each state and adjusting that control
	until the state is within a specified tolerance of zero. States include the
	recti-linear accelerations udot, vdot, and wdot, the angular accelerations
	qdot, pdot, and rdot, and the difference between heading and ground track.
	Controls include the usual flight deck controls available to the pilot plus
	angle of attack (alpha), sideslip angle(beta), flight path angle (gamma),
	pitch attitude(theta), roll attitude(phi), and altitude above ground. The
	last three are used for on-ground trimming. The state-control pairs used in
	a given trim are completely user configurable and several pre-defined modes
	are provided as well. They are:
	- tLongitudinal: Trim wdot with alpha, udot with thrust, qdot with elevator
	- tFull: tLongitudinal + vdot with phi, pdot with aileron, rdot with rudder
	and heading minus ground track (hmgt) with beta
	- tPullup: tLongitudinal but adjust alpha to achieve load factor input
	with SetTargetNlf()
	- tGround: wdot with altitude, qdot with theta, and pdot with phi
	The remaining modes include <b>tCustom</b>, which is completely user defined and

Now in terms of ground trim what is going to come into play are the ground reactions, typically between the gear and the ground. Now it looks like there is only as single ground reaction in the X8 at the moment, so maybe it can't balance, and/or the spring effect is too great etc.

    <ground_reactions>
      <contact type="STRUCTURE" name="Belly">
        <location unit="IN">
         <x> 0.0 </x>
         <y> 0.0 </y>
         <z> 0.0 </z>
        </location>
        <static_friction>  1.00 </static_friction>
        <dynamic_friction> 1.00 </dynamic_friction>
        <spring_coeff unit="LBS/FT">      10.00 </spring_coeff>
        <damping_coeff unit="LBS/FT/SEC"> 10.00 </damping_coeff>
      </contact>
    </ground_reactions>

[rishi1906]

Thanks for the detailed info, I updated Ground Reaction and Ground Trim is working now.