Did anybody tell me what's new in version 4 vs 3.26?

be very careful not to upgrade your .mdl file but make a copy in a new directory and only use v4 if really necessary.

It is probably better to stick with star-cd v3.26, which is well tested, unless you are desparate to use poly mesh. Tousers, it doesn't really have anything new (probably less compared to v3.26 such as engine calculation)

It somehow depends. Calculations with tet-meshes are about 3 times faster in v4. There are other advantages as well.

Where did you get these performance figure or are you talking about using "poly mesh" for underlying tet mesh?

i made similar test: using a large tetmesh (approx 2million cells) and used same mesh, same boundary conditions, etc. and made calculations with V3.26 and V4. for same convergence level V4 was 2-times faster than V3.26. The results were 99.999% the same with both versions.

thomas

I just run some cases to compare the codes.

Same tet-mesh with prism layer (4 mio cells) v4 was about 3 times faster. After converting the tets into polys the speedup was even more impressive.

Also the segregated solver in star4 seems to be very much better than the solver in ccm+. Using ccm+ I often have to ramp up the boundary values to get a solution. In star (3 and 4) it is no problem to start with the final boundary values and even with default relaxation factors.