Recommended settings and troubleshooting

This chapter does not prescribe one set of values for every problem. Instead, it helps you use the physics, mesh, and convergence state to choose a safe starting point and the next checks to make. The input reference is authoritative for allowed values of each key.

Pre-run checklist

Check the following before committing to a long run.

  • The analysis dimension matches the executable dimension (2D or 3D).
  • Mesh zone and marker names match domains.mesh_zones and boundaries.mesh_markers.
  • The physical model fits the flow problem (Euler or Navier–Stokes; laminar, SA, or SST).
  • Reference length, area, and density agree with the definition of force coefficients.
  • You have defined monitors and outputs for quantities of interest.
  • The feature is within the validated scope.
Warning

Some turbulence-model values other than sa and sst may pass input validation without producing a turbulence analysis. Always confirm the supported scope before evaluating turbulent-flow results.

Starting points by problem type

Problem Physics/time Numerical starting point Result assessment
Low-speed, smooth external flow DBNS, Euler or laminar NS, steady Use second-order spatial accuracy and a limiter; start with conservative CFL Stable force coefficients and pressure distribution
Compressible external flow with shocks DBNS, steady Shock-robust convective scheme and limiter; low initial CFL Shock location, force coefficients, and nonphysical oscillations
Attached turbulent boundary layer DBNS, RANS-SA or SST, steady First match the wall first-cell height to model or wall-function assumptions y+, skin-friction coefficient, and profiles
Unsteady flow, such as vortex shedding DBNS or FR, unsteady Decide time step and observation duration first, then configure BDF or dual time Frequency, mean/amplitude, and time history
Internal flow DBNS/PBNS appropriate for the boundary conditions First check mass-conservation monitors and inlet/outlet consistency Mass imbalance, pressure drop, and flow rate

This table provides starting points only. Before using results for design decisions, compare them with a Wake CFD Validation Reports case under similar conditions or an independent reference.

Schemes and CFL

Do not start with a high CFL or the most aggressive settings. This sequence improves reproducibility and root-cause analysis.

  1. Produce a baseline result with a validated case or conservative settings.
  2. Confirm that residuals and quantities of interest stabilize.
  3. Raise CFL incrementally to see whether the same result can be reached faster.
  4. Change schemes, preconditioners, and meshes one at a time.

Nonphysical oscillations near shocks, negative material properties, and early-iteration blowups usually first indicate excessive CFL, poor mesh quality, or inconsistent boundary conditions. Check those three before changing schemes alone.

Assessing convergence

For a steady analysis, “converged” means more than low residuals.

  1. Run completeness — Are there no input errors, NaN/Inf values, or abnormal termination?
  2. Residuals — Are their reduction from the initial values and downward trend sufficient?
  3. Quantities of interest — Are design quantities such as force coefficients, mass flow, and pressure drop stable?
  4. Conservation — For internal flow, is mass imbalance within tolerance?
  5. Spatial field — Do Mach, pressure, and velocity fields agree with the boundary conditions and physics?

Use the following response order when residuals stall.

Symptom Check first Next action
Divergence or NaN from the start Units, boundary conditions, negative mesh volumes, initial CFL Lower CFL and simplify initial and boundary conditions
Residuals fall but force coefficients keep changing Iteration count, monitor definition, actual flow unsteadiness Extend the run or consider an unsteady analysis
Residuals stall at a high level Mesh quality, scheme/limiter, linear solver Change one item at a time and compare with the baseline
Unstable internal-flow results Inlet/outlet conditions and mass balance Recheck the boundary-condition combination and domain markers

Mesh and wall treatment

Good solver settings cannot compensate for a poor mesh. In particular, for turbulent wall analyses, first-cell height and y+ must fit the model and wall-function choice.

  • Without wall functions: prepare a mesh that sufficiently resolves the near-wall region.
  • With wall functions: confirm the target y+ range and record wall-function use with the results.
  • Aspect ratio, skewness, and nonorthogonality can directly affect convergence and wall results.
  • Do not use a single-mesh result as a design basis without checking how mesh refinement changes it.

Output and restart operation

Writing large volume files at every iteration rapidly increases I/O and disk use.

  • Prioritize CSV monitors and residual output for convergence diagnostics.
  • Set volume and surface frequencies separately for intermediate checks and final results.
  • Enable restarts for long calculations and define a retention policy.
  • Keep configuration files, solver version, mesh, and run logs in the same folder as results.

For detailed formats, see File output and Monitoring.