We are trying to run a 70 million cell model. The .msh can be read into fluent, after writing down the .cas it can not be read into fluent. Also we have seen that it takes extremely long time to read and write down large models (30 millions) in parallel. Has anybody some idea regarding these problems ?

Yup. Takes extremely long. My 3 million cell problems take between 1 and 2 hours to read onto a parallel machine. There is an rpsetvar command to speed up the communication but this will lead to a crash, tested it several times.

I guess that now people will go more and more to larger problems on parallel machines, Fluent will really have to address this issue!

I have run a 31 million cell simulation before. Read in and write out time will depend on the machines you are using, but my read in time is about 6 hours for case and data on two Sun Ultra Enterprise 6500 machines using 36 cpu's. As for the error on read in, I have no idea from the information provided. Maybe someone could help if you gave the error you are receiving? Good Luck, Casey