AcuSolve and convergence … (Tips & Tricks)

Did You Know that AcuSolve relies on two separate conditions to establish the convergence of a steady state simulation? Before AcuSolve considers a simulation to be converged to a steady state, both conditions must be satisfied. The first condition that must be satisfied pertains to how well the current solution satisfies the underlying equations. The metric used to quantify this is referred to as the residual ratio and measures the relative imbalance of the left and right hand side of the equations being solved. The second condition ensures that the solution is not changing significantly between time steps and is measured by the solution ratio. This metric reflects how much the solution at each node in the model has changed when advancing the solution from one step to the next. In both cases, each of the metrics is normalized by an appropriate value. The normalization factor is computed separately for each stagger and is re-computed at each time step.
When comparing the convergence of AcuSolve against other solvers, it is important to consider the method used to determine convergence.
Most other solvers do not check the change in the solution at each step (i.e. solution ratio), but rely solely on the residuals to determine convergence. Additionally, they are often times not re-normalized at each step, which may lead to false indications of convergence. For more information regarding the computation of residuals and solution ratio, see the following article … (January 15, 2013)

Have a question or comment?

Post your comment/question on our user forum.