dear all, i m doing a simple crankshaft simulation using CFX and the BC's are static pressure at inlet-576 [pa] and static pressure at outlet- 0 [pa].

it has one inlet and four outlets. this is actually an oil flow simulation thru the crankshaft of a rotary compressor

and when i m giving it a "run" in the solver...i m getting an error message as "fatal overflow error in a linear solver"....what is this mean ?

i want to know the answer !!!

could anyone plz help me out in this issue

regards with thnx Rashmi V Rao

Hi,

As you can see from most previous posts on this topic, it can be due to a variety of reasons like mesh, BCs, timestep etc. In your case most probably due to the BCs, this is well documented in the help files. Try a differnt BC at inlet say massflow.

Santhosh

Hi,

Well your problem is diverging and has crashed. This is what the error means. If this error is encountered within first few iteration from the start of the simulation, there is a definite issue with the bc/physics setup/timestep in the case.

You have used static pressure at the inlet which is a unstable bc. Also the combination of pressure boundary at inlet and outlet makes the case very sensitive to intial guess.

I would suggest you first try giving mass flow at either inlet or outlet. Make sure you change the static pressure condition at the inlet to total pressure atleast.

Regards, test

dear all

thnx a lot for ur valuable suggestion !!!!!!!!! i will surely try out what all u told me to do.

thnx once again

regards Rashmi V Rao

DEAR ALL

AS test SAID I CHANGED THE INLET BC AS MASS FLOW RATE AND KEPT THE OUTLET BC AS STATIC PRESSURE....BUT STILL I M NOT ABLE TO "RUN" THAT SIMULATION IN THE CFX...THE SAME ERROR IS SHOWN.("FATAL OVERFLOW IN LINEAR SOLVER")

COULDNT FIND THE EXACT REASON. I EVEN REFINED THE MESH TOO.PREVIOUSLY IT WAS AROUND 5,00000 AND NOW IT IS AROUND 10,00000...WHAT SHALL I DO NOW...?

ANYBODY TO HELP IN THIS REGARD THNX WELL IN ADVANCE

RASHMI V RAO

Hi,

Try:

1) reducing the timesteps size for the first few timesteps/iterations

2) Using a simpler turbulence model (what turbulence model are you using?)

3) Using upwind differencing and first order time differencing (if transient) to get it started, and change over to a more accurate scheme after you have a better general flow field.

Glenn Horrocks