Solution controls
Solution control parameters have a strong influence on the progress of the analysis, so it is important to have a basic understanding of their significance and effect during a run. You are therefore advised to refer toChapter 7 in the Methodology volume for a detailed discussion of under-relaxation and other solution control topics.
STAR-CD offers two alternatives for solving steady-state problems:
1. An iterative method employing under- relaxation factors
2. A pseudo-transient time marching to the steady-state solution with a
fixed-length time step. Note, however, that an under-relaxation factor (default value 0.2) is still used on the pressure correction equation.
Incompressible, non-reacting and low Mach number flows usually converge smoothly and fast in a combination with the inertial under-relaxation shown in equation (7-14) of the Methodology volume.
If fluid flows which are characterized by travelling waves (e.g. pressure waves in compressible fluids, or gravity waves in free surface flows) can reach a steady state, the convergence process is typically much more robust and stable if one can resolve to some extent the waves travelling during the iteration process. For this, it is important that waves travel in all cells with the same pseudo time step, and the pseudo-transient mode is usually better suited to this class of problems.
In other problems (e.g. inviscid flows and where the initial velocity field is zero), we can have very small or zero values of the central coefficients at an early stage of the iteration process. In such cases, local pseudo time steps become very large or infinite (seeequation (7-19)), which again has an impact on the convergence and stability of the solution. In flows which exhibit this kind of problem, use of the pseudo-transient mode is recommended.
The main advantage of the iterative method employing under-relaxation factors compared with the pseudo-transient mode is that, in the former, the under-relaxation
AP
Analysis Controls for Steady-State Problems
factors vary between 0 and 1 and a large number of cases run nearly optimally with default values (e.g. 0.7 for momentum and 0.2 for pressure correction) that are based on considerable past experience. However, in the pseudo transient approach, the time step varies between zero and infinity and a suitable value is not always easy to find. The optimum value can be determined only by numerical experiments. As a guideline, one should choose a time step such that the Courant number based on the characteristic velocity and the characteristic mesh size is between 1 and 8. Note that, as with under-relaxation, lower values of time step are more likely to promote convergence, while larger ones lead to a faster solution.
The task of setting up solution controls for either of these methods can be divided into the following steps:
Step 1
Start up the STAR GUIde system and then define the type of problem you are solving by selecting Steady State from the Time Domain pop-up menu in the
“Select Analysis Features” panel Step 2
Go to the Solution Controls folder and open the“Solution Method”panel. From the pop-up menu at the top of the panel select:
• Steady State for conventional steady-state runs. Also choose the numerical algorithm to be used (see topic“Steady-State Solution”). In every case, specify the maximum residual error tolerance (i.e. maximum acceptable level of remaining error in the solution), plus any additional parameters required by the algorithm you have chosen.
• Pseudo-Transient for pseudo-transient runs (see topic“Pseudo-Transient Solution”). The maximum residual error tolerance (i.e. maximum acceptable level of remaining error in the solution) should be specified; the normalised residuals are displayed on the screen and also saved on file case.run, as in ordinary steady-state runs.
Step 3
In the“Primary Variables”panel, inspect the solution status for flow variables and material properties (see topic“Equation Status”) to confirm that the right variables will be solved for.
Step 4
Check the“Solver Parameters” (under-relaxation factors, number of calculation sweeps and residual error tolerances for each solution variable).
Step 5
Choose one of the available“Differencing Schemes”. It is suggested that
higher-order differencing schemes such as LUD or MARS should be used if high spatial discretisation accuracy is required.
Output controls
Having set the solution control parameters, the next task is to choose the type and volume of output from the forthcoming STAR run. The bulk of this output consists of solution variable values at cell centroids. Output controls can be applied by going to the Output Controls folder in the STAR GUIde system and following the steps
Analysis Controls for Steady-State Problems
below:
Step 6
Consider whether detailed printout on the solution progress is required and if necessary specify the appropriate settings in the“Monitor Numeric Behaviour”
panel.
Step 7
Decide whether you want to follow the progress of the analysis by generating various types of monitoring data at every iteration. If so, go to the Monitor Engineering Behaviour sub-folder and use one or both of the following panels:
• “Monitor Boundary Behaviour” — select one or more boundary regions and the type of monitoring information to be generated for them
• “Monitor Cell Behaviour” — select one or more sub-domains, defined in terms of cell sets, and the type of monitoring information to be generated for them
The requested data are stored in special files (case.erd and case.ecd for boundary and cell data, respectively), from where they may be displayed as pro-STAR graphs at the end of the analysis (see panel“Engineering Data” in the Post-Processing folder) or read by an external post-processing package.
Step 8
Specify the manner of saving mesh data for use in post-processing and/or restart runs via the“Analysis Output”panel (“Steady state problems”). If desired, go to the
“Additional Output Data” section to select any wall data to be included in the solution (.ccm) file. This is important, as these settings will affect the availability of data for post-processing. You can also select what wall data are to be ‘printed’
(i.e. displayed on your screen) and stored in the .run file at the end of the run.
For both post and print control parameters, it is up to you to check the default settings and change them, if necessary, according to the type of problem being analysed.
Other controls Step 9
Go to the Sources sub-folder and inspect the“Source Terms” panel to see if any additional information (such as extra source terms for flow variables) is needed to completely describe your problem. Note that STAR-CD provides special switches and constants for activating various beta-level features in the code, or for turning on calculation procedures designed for debugging purposes. These are found in the
“Switches and Real Constants”panel and are normally used only after consultation with CD-adapco. An alternative way of performing this function is to enter special debugging instructions into the Extended Data panel, accessible from the Utilities menu in the main window (or issue command EDATA).
Step 10
Go to the Analysis Preparation/Running folder and open the“Set Run Time Controls” panel:
• For conventional steady-state runs, enter the maximum number of iterations