Convergence Accelerator
Some flow problems—for example, cases with a wide range of speeds in the domain (including super- or hypersonic flow)—require a lower CFL number for stability. This can happen when using both constant or flow-dependent material properties.
Typically, the mass-flow convergence is slow for these cases and these simulations can take many iterations to converge. To speed up these computations, Simcenter STAR-CCM+ features a mass convergence acceleration tool within the Coupled Implicit solver.
The Continuity Convergence Accelerator (CCA) formulates and solves a pressure-correction equation using the Density-based/Riemann flux discretization. The solution of this additional equation provides updates for pressure, velocity, and other flow field variables in such a way that the overall and individual cell mass imbalances are minimized at each iteration.
Continuity Convergence Accelerator Guidelines
The Continuity Convergence Accelerator does not modify the final converged solution; it only modifies the path to convergence.
The Continuity Convergence Accelerator has a positive effect on convergence for cases with slow mass balance convergence due to strong non-linearity of the flow and the CFL limitations of the solver, for instance when the Coupled solver is used with an Outlet boundary type. However, the Continuity Convergence Accelerator is not a universal remedy to all convergence or robustness issues in the Coupled Implicit solver. Use the Continuity Convergence Accelerator only when mass-balance convergence is slow.
If convergence issues occur with the Continuity Convergence Accelerator, reduce the under-relaxation factor. For highly rotational and/or high speed flows, lower under-relaxation values and careful under-relaxation ramping (from 0.01 or lower) may be required.
Once activated, the Continuity Convergence Accelerator can be deactivated at any point in the simulation without any negative effect on the Coupled Implicit solver.
For maximum convergence efficiency of the Coupled Implicit solver with convergence acceleration, tune the value for the CCA Under-Relaxation Factor for the specific flow problem. For difficult problems with stiff numerics, values as low as 0.01 or lower can be used to allow CCA to smoothly correct or reduce the remainder of the continuity equation residual.
If the values for the CCA under-relaxation factor are too large for a model, the result can be strong solution oscillations, lack of convergence, or even divergence. In these cases, reduce the CCA Under-Relaxation Factor tenfold and restart the simulation. In these cases, you are advised to use the Expert Driver for the Coupled Implicit solver, where Corrections Control: Minimum and Maximum Explicit Relaxation need to be reduced with respect to default values, for example using minima of 0.1 and 0.2 and maxima of 0.3 and 0.5.
Use Enhanced Stability Treatment for those cases where the Coupled Implicit solver with CCA shows poor convergence. When it is selected, both Bounded Velocity Corrections and Enhanced Mass-imbalance Calculations are activated and the AMG Acceleration Method is set to None. Any of these properties can then be activated or deactivated independently. However, when Enhanced Stability Treatment is deactivated and activated again, Bounded Velocity Corrections and Enhanced Mass-Imbalance Calculations are also re-activated.