CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Fatal over flow in solver (http://www.cfd-online.com/Forums/cfx/19155-fatal-over-flow-solver.html)

Nandu September 4, 2002 11:47

Fatal over flow in solver
 
Hi,

I'm modelling supersonic jet flows into still air. I have managed to get convergence of M = 1.41 (ideally expanded) flow into still air using a local timestep factor for the solver control in CFX 5.5

When i change the inlet pressure conditions in the results file from the previous simulation so that i can get an underexpanded or over expanded jet, the solver crashes with the error being a 'fatal over flow in the liner solver'.

This error also comes when i use a new definition file with the old results file as an intial guess. Changing the local time step factor has not helped. The failure of the simulation is the at the 1st timestep. Any help to rectify this situation is most appreciated.

regards N Menon

Robin September 4, 2002 14:56

Re: Fatal over flow in solver
 
Hi Menon,

Your change in the boundary condition is probably too dramatic. Try making a series of small changes, or starting from scratch with the new condition.

Furthermore, you should limit the use of a local timestep factor to getting initial convergence for your simulation. You should switch to a Physical or Automatic timescale and run to convergence before interpreting the results.

Regards, Robin

Nandu September 5, 2002 06:43

Re: Fatal over flow in solver
 
Well heres the scenario. I ran the simulation for a M=1.41 jet (nozzle exit pressure equals ambient pressure, ideally expanded jet) using a local timestep factor of 10. the flow was convergerd to RMS 1e-04

Then i changed the timestep control to auto timescale in the results file from the previous simulation. the intermediate results show a totally different flow field from the semi-converged result (with local timestepping). The flow field shows the simulation (with auto timestepping) to be starting with the intial guess (not the results) for the local timestep simulation. Is there something i have missed with the intialization for the second simulation?

Is there a numerical issue with using extreamly small timesteps in regions where the flow is many time sllwer than in other regions of the flow field, apart from the obvious that the flow field will take ages to converge?

cheers nandu

Robin September 5, 2002 10:02

Re: Fatal over flow in solver
 
Hi Nandu,

Make sure you selected the Automatic with Value option for initial guess and not the Value option.

The Automatic with Value option tells the solver to use the existing solution if available, or the value you specified if there is no current solution. If you chose Value or Default, the solver will re-initialize the variables on restart.

See the on-line help for more details.

Regards, Robin

Nandu September 5, 2002 10:33

Re: Fatal over flow in solver
 
thanks Robin, u've been a huge help

cheers nandu

Dave September 16, 2002 07:19

Re: Fatal over flow in solver
 
Hi Nandu could you explain to me what the Local Time stepping Factor is ? I have read the on line manual and have managed to only confuse myself. Cheers Dave


All times are GMT -4. The time now is 23:48.