When connecting FMUs in co-simulation mode in a loop an algebraic loop errors occurs, which causes termination of simulation. Therefore, I assume Simulink identifies the FMU-blocks as direct feedthrough. In the FMI 3.0 specification it is stated, that there cannot be a direct feedthough in Co-Simulation, as the step-size has to be greater that 0. Why does Simulink handle FMUs in co-simulation mode as direct feedthrough (and not like integrator blocks instead)?
Unfortunately I could not find anything regarding how the C-API functions of the FMUs are called by the Simulink discrete fixed-step solver.When connecting FMUs in co-simulation mode in a loop an algebraic loop errors occurs, which causes termination of simulation. Therefore, I assume Simulink identifies the FMU-blocks as direct feedthrough. In the FMI 3.0 specification it is stated, that there cannot be a direct feedthough in Co-Simulation, as the step-size has to be greater that 0. Why does Simulink handle FMUs in co-simulation mode as direct feedthrough (and not like integrator blocks instead)?
Unfortunately I could not find anything regarding how the C-API functions of the FMUs are called by the Simulink discrete fixed-step solver. When connecting FMUs in co-simulation mode in a loop an algebraic loop errors occurs, which causes termination of simulation. Therefore, I assume Simulink identifies the FMU-blocks as direct feedthrough. In the FMI 3.0 specification it is stated, that there cannot be a direct feedthough in Co-Simulation, as the step-size has to be greater that 0. Why does Simulink handle FMUs in co-simulation mode as direct feedthrough (and not like integrator blocks instead)?
Unfortunately I could not find anything regarding how the C-API functions of the FMUs are called by the Simulink discrete fixed-step solver. fmu, fmi, algebraic loops, discrete fixed-step solver, fmu-block MATLAB Answers — New Questions

​