About the Reinitialization Process
Solver events make it possible to reinitialize dependent variables. This section clarifies how the provided reinitialization values are processed.
Summary of the Reinitialization Process
Triggered events are processed through the following steps:
1
Dependent variables are reinitialized sequentially. If there are N triggered events, then N reinitializations are carried out.
2
The reinitializations carried out in step 1 may have led to nonconsistent dependent variables (for example, boundary conditions that are not satisfied for a PDE, or a non-null residual for a DAE). By default, the solver will try to correct the reinitialized state to enforce consistency, using a small artificial time step with the backward Euler method. This can be bypassed by clearing the Enforce consistency after reinitialization check box (see Enforcement of Consistency After Reinitialization below for more details). The solver log indicates whether and how consistency has been enforced.
3
If some implicit events have been triggered by steps 1–2, then steps 1–2 are carried out again, to process these events. This is indicated in the solver log.
Even if several events (explicit or implicit) are triggered simultaneously, consistency is enforced at most once in step 2.
Enforcement of Consistency After Reinitialization
The Enforce consistency after reinitialization check box makes it possible to control step 2 of the reinitialization process.
If the Enforce consistency after reinitialization check box is deselected for all the triggered events, then consistency will not be enforced. Not enforcing consistency can yield a performance gain. However, this may lead to a solver failure if the dependent variables exhibit a significant nonconsistency after reinitialization.
If the Enforce consistency after reinitialization check box is selected for at least one of the triggered events, the method used to enforce consistency is controlled by the value of the Consistent initialization list associated with the active time-dependent solver node (see, under Time Stepping (BDF and Generalized Alpha), the section about Algebraic Variable Settings):
•
If set to Backward Euler, then dependent variables are corrected with the backward Euler method using a small artificial time step.
•
If set to Off, then consistency is not enforced after reinitialization.
•
If set to On and a Fully Coupled or Segregated solver subnode is active, then dependent variables are corrected with a backward Euler step.
•
If set to On and the built-in nonlinear IDAS solver is used, then consistency is enforced using either IDAS or a backward Euler step, depending on the equation solved. See The Implicit Time-Dependent Solver Algorithms for instructions on how to use the built-in IDAS solver and its applicability.