Scope and limitations

• The co-simulation performed by the Generic Co-simulation is of explicit type, which is the "minimal" but most general and least constraining requirement for coupling companion solvers. This is why, in theory, any third-party or external tool can be coupled to Simcenter Amesim via co-simulation, provided that this tool is also "open" to co-simulation.

• For now, the Generic Co-simulation is limited to bipartite solver communication i.e. between Simcenter Amesim and Simcenter Amesim itself, or between Simcenter Amesim and another tool. In other words, a maximum of two companion solvers can be coupled: 1 master and 1 slave. However, specific cases can be envisaged by associating Hybrid Co-simulation (in Simcenter Amesim) with "standard" co-simulation (see section Simcenter Amesim co-simulation types and features for definitions): for instance by importing a Simulink model (including solver) into Simcenter Amesim and simultaneously connecting Simcenter Amesim with another specialized code, thanks to co-simulation feature.

• In terms of performance, co-simulation especially suits well to situations where:

the dynamics involved are weakly coupled,

the communication time step is greater than largest time constant in the system,

the "low frequency" part of the model performs lots of CPU time consuming operations.

Only if all of the three criteria mentioned above are fulfilled, significant CPU speedups can sometimes be observed, thanks to a more optimal use of the solvers. However, in most cases, co-simulation usually slows down the simulations and, since each model involved is seen as discrete by the other one, it can also lead to stability and/or accuracy issues, which is usual when coupling solvers together.

• Batch run simulations and other advanced features (parallel or serial co-simulation selection) for the fine-tuning of the co-simulation scheme (see Generalities) only comply with Hybrid Co-simulation.