[samraj]

Hallo all,
I have been doing a flow simulation. the flow happens after combustion. In order to find which mesh size would help me obtain i need to do mesh convergence study. Like structural analysis, one can lock all parameters to a constant value but one with one variable and try it for different mesh sizes? Also with time convergence, since it deals with cfl criteria, i am very confused. I have two variables, cfl and time step. if i understood it correctly, the time step and cfl both influence each other over the flow process. how to approach this problem. does the process vary with the simulation tool? any help would be very much appreciated.Thanks all again

[FMDenaro]

Originally Posted by samraj

Hallo all,
I have been doing a flow simulation. the flow happens after combustion. In order to find which mesh size would help me obtain i need to do mesh convergence study. Like structural analysis, one can lock all parameters to a constant value but one with one variable and try it for different mesh sizes? Also with time convergence, since it deals with cfl criteria, i am very confused. I have two variables, cfl and time step. if i understood it correctly, the time step and cfl both influence each other over the flow process. how to approach this problem. does the process vary with the simulation tool? any help would be very much appreciated.Thanks all again

I suggest to do a convergence study by taking a fixed value for the cfl on all the grids you refine. That will take a constant ratio between dt and mesh sizes

[samraj]

Hallo, Thanks for your message. I understand that. does this mean i can do both spatial and temporal convergence at the same time? Thanks

[FMDenaro]

Originally Posted by samraj

Hallo, Thanks for your message. I understand that. does this mean i can do both spatial and temporal convergence at the same time? Thanks

yes, this way you are sure that the all terms of the local truncation error vanishes asymptotically

[samraj]

Hallo,
Thanks for your message. I will try this method now. Also does this method of eliminating truncation error vary with different tools ? i need to do this on LS Dyna. I am not sure if ls dyna related doubts can be asked here. However, i request you to share your knowledge and experience regarding this. Thanks

[FMDenaro]

the convergence study does not depend upon a specific code, it is a numerical methodology to assess the behaviour of a code on a known test.

[samraj]

hello,
Thanks. I understand that. Thanks much.

[samraj]

Hallo,
Would you also help me with time step value? For fixed cfl over different mesh size, how one could define the time step?should it also be a fixed one? Thanks.

[FMDenaro]

once the cfl value is fixed, you compute the time step for each grid, for example in a 1D case that would give dt= cfl*h/|u|