Different results between OMEdit and FMU?

[Amuri44]

I have a model R in in OpenModelica, in which it is a user-defined block.

[ThummeTo]

Dear @Amuri44 ,

what I see looks fine so far, but the code is not complete, e.g. is dt=1e-6?
Further information on the FMI2-calling convention ("state-machine") can be found in the FMI2-specification.

Further you could check via fmiGetReal(comFMU, ["aV.Pm","aV.Vs","aV.Vic"]) if all parameters are set correctly after fmiExitInitializationMode(comFMU). How does the output differ? Could you provide a plot and/or further information? Do you log Pmm, Vff inside the loop? It is hard to debug without further information.

Please note, that this could be done (more efficient) with fmiSimulateCS as well.

Regards!

[Amuri44]

• Yes, dt=1e-6 (I have added it in the previous code)

[ThummeTo]

The provided code overwrites the variables Pmm, Vff in every simulation step. This leads to a flat curve.

What is the simulation result if you use

success, recordedValues = fmi2SimulateCS(comFMU, t_start, t_stop;
                        tolerance=mySimulationTolerance,
                        dt=1e-6,
                        recordValues=["aV.Pm","aV.Vf"],
                        setup=false,
                        reset=false,
                        inputValues=["aV.vd","aV.vq"],
                        inputFunction=myInputFunction)

with

function myInputFunction(t)
   return [5.0, 5.0]
end

instead of your custom loop? The results are saved in recordedValues.

Finally if you need a high simulation resolution you could try changing the solver tolerance via mySimulationTolerance.

[ThummeTo]

return [1.0,5.0, 5.0,1.0]

[ThummeTo]

The provided code overwrites the variables Pmm, Vff in every simulation step. This leads to a flat curve.

Ok, how to prevent the overwrites?

Don't overwrite it ;-)
For example pushing the current value into an array at every simulation step.

[ThummeTo]

Could you check another simulation tool, like e.g. fmpy? Another fast check would be to import the FMU again into OMEdit and check if the FMU simulates like the original system.

It is hard to say where the error is coming from. The first question is if this error is caused by the FMU (exported by OMEdit) or by the simulation tool (FMI.jl).