Trade-offs in Defining Time IncrementationYou must carefully define how time progresses to balance the quality of your results with the run time of your simulation. The equations that run your simulation produce a data set that describes the evolution of the flow as a result of the boundary conditions and the flow regimen. A successful transient flow simulation defines time increments that are small enough to resolve the unsteady fluid dynamics of interest. However, this same simulation defines time increments that are large enough to reduce the computational time required to run the simulation. Since each equation uses the same time increment to solve your simulation, changing your incrementation settings can drastically change the simulation runtime. You can balance the trade-off between the completeness of your results and the simulation run time by using the following methods to control how time progresses:
If you choose to create your transient flow step with fixed time increments, you can specify an amplitude to allow for variation in the fixed time increments. If you specify an amplitude, ensure the time increment evolution can resolve all the time scales of interest in the problem. Otherwise, the simulation might not accurately resolve some important unsteady phenomena. Specifying an amplitude provides additional control for cases where using the automatic time step option can produce time increments that are too small but a fixed time step cannot resolve all the time scales of interest throughout the simulation. Time Increment ValidationThe change in size of your time increments, from the first increment to the last increment of the simulation, can become unstable if it is not regulated. For example, if your time increments are too large, the simulation can lose information transferring across the smallest cells of your mesh. As a result, the solution would be inaccurate. You can validate the size of your time increment using the Courant-Friedrichs-Lewy (CFL) number, which indicates how fast information propagates across the smallest cell of your mesh. The amount of time required to propagate the information is the maximum time increment allowed to ensure stable conditions for advection. Increasing or decreasing the CFL number, therefore, affects the duration of the simulation. For large time steps (that is, CFL > 1.5), the equation solvers can perform a specified number of fixed point iterations on each cell to ensure accurate solutions. In other words, the equation solver calculates the solution multiple times until the time step's solution is within an acceptable error margin, such that the solver can advance to the next time step. You can control both the maximum number of iterations used for large time steps and the acceptable error margin. Both options have an effect on the time spent running the simulation. When the estimated error for every solver falls below the error threshold specified for the transient flow step, the step stops automatically and the simulation proceeds to the next time step. If the estimated error for one or more solvers remains equal to or greater than the error threshold, the transient step ends when it reaches the maximum number of fixed point iterations for all the iterations in the step. | |||||||