Computational Issues

Computational Issues


For simple flows, single precision runs (32 bit arithmetic) are usually adequate for convergence. 

In some more difficult cases, where there may be extremes of scales in the problem, or very fine meshes, it can be advantageous to use double precision (64 bit arithmetic). 

This will require more memory, but may not add a huge overhead on computational times, depending on the nature of the hardware being used.