Cost of computing k-kl-omega
I shall analyse low-Re flow over a streamlined 3D body (velomobile at 10 m/s). For this specific task I have been recommended k-kl-omega as suitable model since the problem involves laminar-turbulence transitions where this new model seems to excel.
Q1: Will k-kl-omega work fine for this problem or should I consider simpler models as well? I plan on using Caedium and the target is to dial in on a practical and cost efficient body as part of a master thesis.
I have the impression K-kl-omega need very fine mesh - found a suggestion for 0.3mm boundary cells with 1.15 growth per layer for the 30mm near moving body. I assume that this fine mesh also will require a timestep small enough to let a particle move not more than one cell during a time step. That would require a time step of 0.3 mm / 10 000 mm / s = 30 uS.
Q2: Is the timestep too small?
Assuming Velomobile is 800 mm wide, 1000 mm high and 2500 mm long I plan on making a mesh of approx 6-12 million cells consisting of a rectangular box - "wind tunnel" - LxWxH of 5.0 m x 2.4 m x 2.0 m
Q3: Will the massive computational task simply be overwhelming for a single CPU - quad core with plenty RAM - to be able to converge within 8-16 hours?