Hello, I am running a simulation of (premixed) combustion in the cylinder of a gasoline engine.When combustion is turned off,the simulation works fine.With combustion on,however, it fails immediately after combustion begins with this message: *** Scratch unit 78, Failed to find scratch data CELL 1 56 1 1 *** Proc: 1 Error encountered, number: : 549

*** Error 549: Scratch/RAM file failed reading/writing

CELL 1 56 1 1

Most likely cause: full disk or not enough RAM memory

on the system.

If the problem persists contact User Support.

RUN STOPPED !!!

I have 8GB of RAM.Is the scratch space the tmpfs in Linux?How can I increase it? Any help would be much appreciated Thanks

if you use bash then you can set the tmpdir

ie export TMPDIR='path to temp directory'

I did and it still fails with the same message.Could it be related to tmpfs and RAM instead of the tmp folder?or my setup? Thanks! Emmanuel

Do you have a 64 bit version of linux and also the 64 bit version of star installed on your machine?

yes and may I add that this model works fine for cold flow.

Yesterday I tried setting TMPDIR=a dir with no limit for sure+adding 14 gigs of swap memory (default was 2G and my RAM is 8G) and I restarted the the calc from a couple of time steps before failing...... Still the same error! I am running it now restarting from 200 time steps before failing and monitoring RAM usage closely.It is using almost all of the RAM even before combustion starts...

What is your grid size, and turbulence model you are using?

did you follow the installation notes and set the swap space at least greater or equal than the RAM?

Initially the swap space was 2G (RAM is 8.2G).But I increased the swap space to 14G using a swap file and restarted the simulation.Same problem...I was actually monitoring RAM and swap space usage before and at the point it crushes (using the free command) and RAM usage was 8 out of 8.2G but swap space usage was only 48 Mb and didn't change as it was crushing, i.e. swap space use didn't suddenly increase...

around 700.000 cells.K-epsilon turb model with these values: TURBULENCE MODEL ............ => HIGH RE K-EPS MODEL |

| CONSTANTS ................. => C_mu=0.09, C_1=1.44, C_2=1.92, C_3=1.44, C_4=-0.33 |

| => cappa=0.419, Pr_k=1.00, Pr_eps=1.219, Pr=0.90

It turned out to be a bug in the CFM combustion model!!!!!!!!!! They are fixing it in a patch coming out next week!