Hi all, I want to ask a problem... When should the "double precision" in Star-CD be used? I guess that it should be used when flows involve chemical reactions. And is my guess right? However, if I use single precision in above, is the solution different? Wu

Double precision is normally used for flows with small density differences. This means there is only an affect on the convergence rate not on the result.

James

Thank you very much. You mean that there is only an affect on the convergence rate. And if using single precision as analysis suits double precision, the convergence rate will be slower. Right?

Yes,

this means for flows with small density differences it is nearly impossible to get a converged solution with single precision.

Always use double precision, memory is relatively cheap so why even risk the introducing additional round off error into your solution.

Using double precision all the time when its not necessary is just a waste of time and resources. Many (not all) machines simply take longer to do double precision arithmetic. When you are trying to run problems with several million cells, the difference in size may mean the difference in whether you can run at all on a 32bit machine (ie cheap linux or windows box) or whether you need a 64 bit machine. The difference in cost, even for a pc between 4Gb of memory and 2Gb is several thousand dollars. Memory is cheaper than it used to be, but its not free and its not so cheap on bigger unix machines either.