CFD Online Discussion Forums - Floating Point Error

inf = infinite

This could mean that there is a variable value (for example temperature) which at the beginning of the calculus (that has been initialised very far from reality somewhere in the flow domain, for example, the flow over a bluff body, initialised using far-field conditions, resulting in high initial velocities in front and behind the body) grows exponentially (in the given example, if the initial velocity is high enough, in the first few iterations, the temp will grow rapidly due to the high compression of the gas against the bluff body's flat nose).

This is especially favorised by an aggresive solver like density-based sover, when using high CFL numbers at start. The simplest way to avoid this problem is to lower the Courant number to let's say CFL=0.5 for about 50 iterations, then increase it to CFL=1 and iterate for another 50 times, save the .dat, and if the solution is stable, double the CFL every 25 iterations or so. But do not go over CFL=20-25, that should be more than enough in most cases.

All the best, Razvan