This section describes the ways coupling steps and coupling iterations can be combined in a steady coupled analysis (that is, an analysis that has Analysis Type set to Steady if set up in the GUI or CLI, or to General if set up in Workbench). In a steady analysis, the specific combination of coupling steps and coupling iterations:
determines when restart data can be written,
allows you to balance the required file storage space and the need for analysis restarts, and
determines how you can use System Coupling's under-relaxation factor and ramping, as these only apply to coupling iterations and cannot be applied over coupling steps.
Steady coupled analyses are generally iteration-based. An iteration-based analysis can have multiple coupling iterations. Multiple coupling steps cannot be defined.
When an iteration-based coupled analysis is solved, the analysis continues executing until either the solution converges, or the specified maximum number of coupling iterations is completed.
Restart points are written at the end of coupling iterations according to the iteration-based Output Control settings in the data model. The creation of intermediate restart points helps to ensure restart coverage, so that restarts are possible even when a simulation with a lengthy runtime is terminated either abnormally (for example, due to an error or power interruption) or via an Abort operation. (An Interrupt operation allows the current coupling iteration to finish, creates a restart point, and then cleanly stops the coupled analysis.)
For any coupled analysis, the creation of restart points is a trade-off to be weighed against available resources. Minimizing the number of restart points created minimizes file storage space at the expense of the ability to restart the analysis, while maximizing the number of restart points can save time and computational effort if a restart is required, but at the expense of file storage space.
Ramping and under-relaxation can be applied across coupling iterations.
Note: The exception is the steady analysis which is step-based because it includes a participant that can create restart points only for coupling steps — that is, with a Participant Type of AEDT (only for 2D participants), CFX, MAPDL, or EXTERNAL DATA). This type of analysis can have one or more coupling steps, with multiple iterations permitted for each.
When a step-based coupled analysis is solved, the analysis continues executing until the specified number of coupling steps is completed. The transition from one coupling step to the next occurs when either the solution converges, or the specified maximum number of coupling iterations is completed.
Restart points are written only at the end of coupling steps, according to the step-based Output Control settings in the data model. This limits restart capabilities, particularly for single-step analyses (for which a restart point is written only at the end of a solution) or particularly complex or long-running multi-step analyses which might experience abnormal terminations.
Ramping and under-relaxation can be applied across coupling iterations, but not across coupling steps. For multi-step steady analyses, the full data transfer value is transferred at the end of each coupling step. Participant solvers may ramp data received from System Coupling at the coupling steps.
For best practices, see Improving the Accuracy of Steady Analyses.