2. Code and Solution Verification

Code verification and solution (or calculation) verification both chase numerical error, but they answer different questions: the former checks that the software actually solves the governing equations as written, while the latter measures how much numerical error remains in a specific simulation run.

Code verification is a testing exercise (typically using vendor-supplied benchmarks or Method-of-Manufactured-Solutions cases) that users should rerun whenever they upgrade the solver or operating system to be sure nothing in the implementation has regressed.

Solution verification is an estimation task that belongs to the analyst; by performing systematic mesh-refinement or Grid-Convergence-Index studies it quantifies discretisation, iterative and sampling errors until key quantities of interest stabilise within an acceptable tolerance.

Because exhaustive refinement can be prohibitively expensive on large 3-D or transient problems, practitioners rely on built-in error estimators; when vendors cannot supply local estimators for stresses, pressures or heat fluxes, organisations often develop their own add-on tools.

Frameworks such as ASME V&V 10 and 20 codify these practices, requiring demonstrated code verification before any validation work and prescribing uncertainty-quantification steps for solution verification.

Requesting the vendor’s verification data set and running in-house error checks therefore provides hard numerical evidence of credibility, protecting engineers from costly decisions based on unconverged or inaccurate simulations.