About the divergence when using unstructured mesh
 

HI ALL,

I AM DOING A PROBLEM OF EXTERNAL FLOW ON A CYLINDER AT M=2 AND 10 DEGREE ANGLE OF ATTACK. I HAVE MADE A COMPLETE TET/HYBRID MESH. I KEPT THE CFL NO LOW TO 0.2 BUT AFTER 3000 ITERATIONS IT DIVERGED.

THEN I RAN THE PROBLEM AGAIN AND INCREASED THE CFL NO SO THE DIVERGENCE CAME AT 1600 ITERATIONS.

SECONDLY, BEFORE DIVERGENCE IT GAVE ME SO MANY WARNINGS LIKE TIME-STEP REDUCED IN 3 CELLS TEMPERATURE LIMITED TO 1 IN 2 CELLS IN ZONE 2 (DEFAULT-INTERIOR) ABS PRESSURE LIMITED TO 1 IN 3 CELLS IN ZONE 2

THE PROBLEM IS OF 2 MILLINON CELLS SO I RUN IT ON 3 COMPUTERS ON PARALLEL TELL ME HOW THE DIVERGENCE PROBLEM CAN BE SOLVED?

Re: About the divergence when using unstructured m
 

keep CFL low. gradually ramp up flow rate gradually with a number of increasingly higher flow rates. Also adjust multigrid settings. Ensure grid quality is good and not too coarse. Use node based gradients.

Re: About the divergence when using unstructured m
 

Yeah the grid quality is approx 77%? Is it OK

Re: About the divergence when using unstructured m
 

Normally grid quality is quoted by max skew. What are you quoting and where from (as a %)? Max skew should be about <0.96/0.97 in Gambit/Fluent terms or >0.05 in ICEM.

Re: About the divergence when using unstructured m
 

Sorry Lemon, Actually I wrote it in % but Gambit showed the quality as 0.77. I thought that u would get it. Anyways, I didnt get ur ppoint that " gradually ramp up flow rate gradually with a number of increasingly higher flow rates"

I don't monitor Mass flow rate and by the way how can we increase it.

Also tell me that what will happen if I use Multi-grid option