Simulink solver type: fixed-step or variable-step

Usually, a variable-step solver can significantly shorten the simulation time. Indeed for a given level of accuracy, it can adjust the step size dynamically and thus reduce the total number of steps.

A fixed-step solver uses a single step size throughout the simulation, which explains why a very small step can sometimes be required for a given desired level of accuracy. Especially for systems in which dynamics appear and disappear locally (e.g. if a hydraulic valve opens), this can be very constraining and inefficient in terms of CPU time.

In Simulink, solver type selection (fixed-step or variable-step) depends on the dynamics of the model - as in other tools -, but it also depends on the way the model will be deployed. See the following table.

Interface mode	Simulink solver type
Fixed-step solver	Variable-step solver
SL2AMECosim		Not supported (*)
AME2SLCosim
AME2SL
(*) Using the variable-step solver is currently not permitted by Simulink. This is always the case each time code is generated from a Simulink model.

If the goal is to import into Simulink a Simcenter Amesim model without the Simcenter Amesim solver embedded into (AME2SL), then the choice between a variable-step and a fixed-step solver for the Simulink master only depends on the dynamics of the resulting Simcenter Amesim-Simulink model.

Note

Most Simulink controllers do not require variable step integration due to the presence of low dynamics or sampled signals in the model.

Note

The fact that the Simulink solver is always fixed-step if the Simulink control model is used in Simcenter Amesim is not a limitation for users who want to export the resulting model to a real-time target for which a "direct" export workflow exists (i.e. with no need for the Simulink Coder between Simcenter Amesim and the target).

A fixed-step solver is indeed necessary for real-time.