Hello,

generall opinion seems to be that:

FEM based solvers are less efficient (speed) and less accurate (balances of conservation quantities)

what do you seasoned guys say?

Question arises in context of coupled vs segregated (one DOF at a time) solvers:

I am not sure if CFX has a coupled solver setting, e.g. for FSI problems. As far as I do know it hasn't.

Instead FLUENT (seems to) have the option to uttilize a coupled solver,

this appears being the case with ADINA also (which in turn is a FEM based solver, as far as I understand).

Thank you for directions, flex

Note: this isn't to fire up any religious wars, I am interested in unbiased rational statements, that's it.

Dear flex,

CFX does not have a "same matrix" coupled solver for FSI, nor does Fluent (where did you get that kind of information)

ADINA does have (based on their website) a full matrix coupled solver for FSI; I have never used though..

Opaque..

Dear Flex,

it is probably best to go back to the basic components of a flow solver and/or structural solver and understand this first...make a list and then do a sort of "map" of how the various commercial solvers fit into it.

For example you could break it into:

Discretization methods:

FEM, Finite Volume, higher order Discontinuous Galerkin (of which Finite Volume is the first level)

Equation coupling methods:

Segregated, Coupled Mass and momentum, fully coupled

Equation Formulation methods:

pressure based and density based fluids solvers

Equation solution methods:

etc.

Regards,

Bak_Flow

what do you say about fluent.IS it fully coupled solver>? or just mass and momentum coupled?

Hi Sam,

Fluent has 2 coupled solvers:

There is a pressure based solver which is mass and momentum coupled. ie energy, turbulence, species, radiation, etc are all outside the coupled loop and solved segregated.

There also is a density based solver which is coupled mass-momentum-energy-species but segregated for other scalars and turbulence.

Although there is an overlap in applicablility of flow solvers, pressure based solvers are very natural candidates for low Mach number flows and density based solvers are very natural candidates for high Mach number flows. There are of-course tricks which extend the solver out of its "natural range"....preconditioning for density based solvers, etc.

Hope this helps.

Bak_Flow

could you explain the preconditioning.

Why we put the zero as opereting pressure for coupled compressible flows.

Hello Bak, opaque,

thank's Bak! for the breackdown of aspects, in there the equation coupling methods. I weren't aware of the existence of coupled mass and momentum coupled option, as opposed to a fully coupled solve. I thought there were ony segregated or fully as a option, blind spot. That answers the question of opaque too. I did hear about coupled FLUENT solves. These were coupled mass and momentum solves then, I assume.

Thanks too, for the info on equation formulation methods. Neat and instructive.

flex

Hello+,

a opinion on if 'FEM based solvers are less efficient (speed) and less accurate (balances of conservation quantities)

then FVM based ones' would be desirable still, hazarding.

flex

still happy to receive opinions relating to the original question, from seasoned guys, generalizations aren't the target.

Scope of problems for which reporting of experience is seeked (of primary interest):

in/external, sub/trans/supersonic flow, in/compressible, single phase, w and w/o CHT, under two-equation turbolence models.

flex

with original question, from my side I intendet a opinion on:

if 'FEM based solvers are less efficient (speed) and less accurate (balances of conservation quantities)

then FVM based ones'

regards, flex

thanks again to Bak_Flow, 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.

flex

[RPereira]

Very, very, very, very, later answer I know…but, take a look at this paper:

http://www.optimal-tech.com/download/publications/FEM-FDM_eng.pdf

Cheers!