Many of the parameters listed in Table 2: Summary of Advanced CHEMKIN-CFD Parameters are relevant only for steady-state solution. Here a brief description of the steady-state solution algorithm is provided in order to clarify the role of these parameters and their impact on solution convergence.

For steady-state problems, the Ansys CHEMKIN-CFD solver is called by Ansys Fluent for each cell during each Fluent iteration towards a steady state. Within each cell, the chemistry solver will attempt to determine an appropriate steady-state using a combination of a modified Newton-iteration procedure and a “pseudo” time-stepping procedure. The Newton-iteration method is used to attempt to directly iterate towards the desired solution. However, at early times in the simulation, this procedure may not succeed, especially when the initial guess is far from the actual steady solution. In such cases, the CHEMKIN-CFD module will automatically switch to a time-stepping mode to advance the initial guess closer to the physical solution. After taking a certain number of time steps, the Newton-iteration algorithm tries again. In this way, the steady-state solution is achieved with a minimum amount of time-stepping. Since time-stepping is computationally expensive, this results in a computationally efficient but robust method for achieving the steady-state solution.

You can control the number of times the CHEMKIN-CFD module is allowed to use time-stepping to advance a solution towards steady-state before reporting a failure for any individual cell (see Maximum Pseudo Time-stepping Attempts in Table 2: Summary of Advanced CHEMKIN-CFD Parameters ).

In performing the pseudo time-stepping procedure, there are two solver options - a “basic” time-stepping procedure and a “robust” time-stepping procedure (see Pseudo Time-stepping Solver Option in Table 2: Summary of Advanced CHEMKIN-CFD Parameters ). The “robust” procedure (default) uses an integration method that is very accurate with error control enforced on all solution variables. This can sometimes be more time consuming but is generally more reliable in tracking transient states towards an eventual steady condition. The “basic” solver option is much more approximate in the time-tracking technique, but is often faster and sufficiently robust for some cases.

When the “robust” time-stepping procedure is selected, you can specify the time interval used for each attempt to advance the solution using time integration (See Pseudo Time-stepping Size in Table 2: Summary of Advanced CHEMKIN-CFD Parameters ). The actual time steps used during the time integration will be automatically determined based on the rate of change of the solution variables. The time interval, however, controls the total time advanced before returning the solution to be used as the next initial guess in the Newton-iteration.method.

In addition to the pseudo time-stepping controls, you may also control certain parameters that affect the convergence behavior of the Newton-iteration algorithm itself. In particular, you can modify the solution bounds imposed during the iterations (see Upper Bounds and Lower Bounds parameters in Table 2: Summary of Advanced CHEMKIN-CFD Parameters ). By extending the bounds slightly outside of the physical limits, the Newton iteration can more easily resolve species fractions that are very near to those limits. This can be especially helpful in early iterations where variable values may be changing rapidly. It is also helpful to try to “reset” any negative species fractions to very small values between iterations, so that these values do not have unphysical effects on reaction rates. For this purpose, you can specify a small positive value, which should not exceed the specified absolute tolerance, as a replacement value for negative fractions (see Small Positive Value to Replace Negative Species Fractions in Table 2: Summary of Advanced CHEMKIN-CFD Parameters ).