Hello all,

anyone out there to comment on the following generalized statement, it's truth's or biasedness:

FEM based CFD is less accurate then FVM based CFD, because of it's limitations in satisfying conservation quantities like impuls, mass, thermal energy etc.

FEM based CFD is less computationally efficient in terms of computational speed, solution time for a given problem, compaired to FVM based solvers.

Thought a general yes/no may be hazardous, I'd be grateful if the experienced (numerically/application wise) pronounced there weighted opinion.

flex

hi heres my view, FEM involves numerical integration and is more efficient when it comes to solution which have different modes. so FVM is better except for specialized cases. which are difficult to model otherwise.. prapanj

prapanj,

can you clarify what you intend with different modes?

It probably depends on the implementation. You have a wide range of accuracy in the FVM codes - just like you would with FEM codes. Check out AcuSolve for a highly accurate FEM code.

This is an interesting question. I hope some experienced people can elaborate.

hi Flex, I was working with a noncomercial a FVM CFD (self developed), a FVM (Numeca code) and a FEM CFD for Modelling turbulent flames in a 3D burner, my point of view: I think that generally FV methods (depending on the scheme used in the particular case) using FBC is more accurate than FEM CFD in problems mainly governed by diffusion and also could be faster and required less computational memory (potentially, off course) although usually, Period(time step) and grid independence of a FVM CFD without using FBC could be a bit less satisfactory in cases with more complicated geometries...in problems governed by convection I think FEM CFD and FVM are similar in almost every performance aspect. !interesting topic¡ bye

hi Flex, I was working with a noncomercial a FVM CFD (self developed), a FVM (Numeca code) and a FEM CFD for Modelling turbulent flames (CMC methods validation) in a 3D burner, my point of view: I think that generally FV methods (depending on the scheme used in the particular case) using FBC is more accurate than FEM CFD in problems mainly governed by diffusion and also could be faster and required less computational memory (potentially, off course) although usually, Period(time step) and grid independence of a FVM CFD without using FBC could be a bit less satisfactory in cases with more complicated geometries...in problems governed by convection I think FEM CFD and FVM are similar in almost every performance aspects. !interesting topic¡ bye

thanks anyway to those who provided some opinion.

Seems as nobody wants to touch that dial, or perhaps because few have the nessesary cross-over experience to tell hard facts.

In FEM you can have superconvergence, which you cannot in FVM. So I wouldn't say FEM is less accurate than FVM.