Developing most numerical solvers involves making certain assumptions which may neglect certain physics but with an acceptable loss of accuracy. Once the numerical solver is created, its quality is measured with
Verification: A process to determine that the model equations are solved correctly. This frequently includes comparisons to known solutions for particular cases or to results from previous work. Also, conducting grid convergence studies is essential in any numerical study to verify that the solution is insensitive to the grid resolution.
Validation: A process to determine that the model equations replicate the true behavior of the physical system. This often involves comparisons to experimental data.