This dashboard tracks the numerical parity and structural compatibility of the ModelScript FMI exporter across FMI 2.0 & FMI 3.0 (ME and CS) against the official Modelica Reference-FMUs. The Reference-FMUs are compiled directly from their manual C source code via their provided CMake build system, serving as the definitive ground truth for these validation benchmarks.

• Native msc: The ModelScript Arena Simulator running the source .mo file directly. Uses the internal WASM CVODE solver by default.
• msc-FMU (fmusim): The exported ModelScript FMU (.fmu) simulated using the official FMI standard fmusim C-binary. We explicitly pass --output-interval to ensure exact parity with the Reference FMU execution.
  • Model Exchange (ME): The FMU only supplies state derivatives. The host (fmusim) integrates the equations. We explicitly configure fmusim to use the CVODE algorithm (--solver cvode).
  • Co-Simulation (CS): The FMU contains its own embedded solver. ModelScript natively compiles a custom 4th-order Runge-Kutta (RK4) integration algorithm directly into the FMU binary to advance the state.
• msc-FMU (omsim): The exported ModelScript FMU (.fmu) simulated using OpenModelica's OMSimulator binary (which also utilizes CVODE/KINSOL algorithms internally).

Values are reported as Max Absolute Error (NRMSE %). A simulation passes if the max absolute error is tiny (< 1.0) or the Normalized Root Mean Square Error (NRMSE) is under 5%.

• Done (No Ref): Indicates the simulation succeeded internally, but no Ground Truth Reference-FMU exists for this FMI version (e.g. StateSpace is not available in FMI 2.0 reference FMUs).
• OMSimulator FMI 3.0 Failures: OMSimulator v2.1.3 explicitly hardcodes schema validation to the FMI-2.0 standard. FMI 3.0 FMUs fail during ingestion because the fmiVersion="3.0" attribute does not match the hardcoded #FIXED value of '2.0'. This is a limitation of the current OMSimulator build.

Simulation Options: stopTime = 1, stepSize / outputInterval = 0.01, Native Solver = cvode

model BouncingBall     "The bouncing ball model"
  Real g(start = -9.81);
  parameter Real e = 0.7;   // Elasticity constant of ball
  parameter Real v_min = 0.1;
  Real h(start = 1);        // height above ground
  Real v(start = 0);        // Velocity of the ball
equation
  der(h) = v;
  der(v) = g;
  der(g) = 0;

  when h <= 0 and v < 0 and -e*pre(v) >= v_min then
    reinit(v, -e*pre(v));
  end when;

  when h <= 0 and v < 0 and -e*pre(v) < v_min then
    reinit(v, 0);
    reinit(g, 0);
  end when;


end BouncingBall;

Simulation Options: stopTime = 10, stepSize / outputInterval = 0.1, Native Solver = cvode

model VanDerPol  "Van der Pol oscillator model"
  Real x(start = 2);
  Real y(start = 0);
  parameter Real mu = 1.0;
equation
  der(x) = y;
  der(y) = - x + mu*(1 - x*x)*y;

end VanDerPol;

Simulation Options: stopTime = 10, stepSize / outputInterval = 0.1, Native Solver = cvode

model Dahlquist
  parameter Real k = 1.0;
  Real x(start=1.0);
equation
  der(x) = -k * x;
end Dahlquist;

Simulation Options: stopTime = 10, stepSize / outputInterval = 0.01, Native Solver = cvode

model Stair
  Integer counter(start=1);
equation
  when time >= counter then
    counter = pre(counter) + 1;
  end when;
end Stair;

Simulation Options: stopTime = 10, stepSize / outputInterval = 0.1, Native Solver = cvode

model StateSpace
  parameter Integer m = 3;
  parameter Integer n = 3;
  parameter Integer r = 3;

  parameter Real A[n, n] = [1, 0, 0; 0, 1, 0; 0, 0, 1];
  parameter Real B[n, m] = [1, 0, 0; 0, 1, 0; 0, 0, 1];
  parameter Real C[r, n] = [1, 0, 0; 0, 1, 0; 0, 0, 1];
  parameter Real D[r, m] = [1, 0, 0; 0, 1, 0; 0, 0, 1];
  parameter Real x0[n] = {0, 0, 0};
  Real u[m] = {1, 2, 3};
  output Real y[r];

  Real x[n](start=x0);
equation
  for i in 1:n loop
    der(x[i]) = A[i,1]*x[1] + A[i,2]*x[2] + A[i,3]*x[3] + B[i,1]*u[1] + B[i,2]*u[2] + B[i,3]*u[3];
    y[i] = C[i,1]*x[1] + C[i,2]*x[2] + C[i,3]*x[3] + D[i,1]*u[1] + D[i,2]*u[2] + D[i,3]*u[3];
  end for;
end StateSpace;