[randolph]

hi everyone,

I think OpenFOAM is not really designed for the turbulence-resolving business (DNS/LES). I used to struggle a lot with even getting some accurate turbulence stress for simple channel flow. An example is the report that the fox posted.

If you only care about the mean flow profile, you will find you do not even need a subgrid model, an under-resolved DNS can give you pretty accurate mean flow. Or even a well-calibrated RANS can give you the mean profile in 5s.

I would suggest taking a look at the high-order codes. For incompressible flow/low march flow, take a look at NEK5000. It just like OpenFOAM has some learning curves. But my experience is that with those high-order codes, even a highly under-resolved DNS can give better the Reynolds stress more accrate. With some filtering, turbulence stress for the channel is just a piece of cake.


Thanks,
Rdf

[Santiago]

Originally Posted by randolph

I would suggest taking a look at the high-order codes.

Low-order incompressible codes are also suitable for LES, as long two discrete properties are guaranteed: (1) mass conservation, and (2) energy conservation. The standard incompressible solvers in OF do not guarantee the second property, due to the "inconsistent" treatment of the non-linear term. This is also dependent on the numerical schemes you use. Perot, Mahesh, and others cut right through the "myth" that only high-order codes are suitable for LES/DNS.

There is considerable work done in OF with energy-preserving solvers, also using LES. I have published some work related to this as well.

Flux reconstruction methods are becoming increasingly popular nowadays, but given the upwinded nature of these codes, there is always the argument of whether is convenient to use explicit LES vs it's more esoteric cousin (implicit)....

[randolph]

Santiago,

Thank you for potentially correct my misunderstanding. Would you mind recommending some reading material (perhaps your papers)?

It sounds like the native OF LES solvers are not energy preserved. Are these true for both the developing lines (OF org and ESI)?

Thanks,
Rdf

[Santiago]

Originally Posted by randolph

Santiago,

Thank you for potentially correct my misunderstanding. Would you mind recommending some reading material (perhaps your papers)?

It sounds like the native OF LES solvers are not energy preserved. Are these true for both the developing lines (OF org and ESI)?

Thanks,
Rdf

https://www.mdpi.com/2311-5521/4/3/171

... and search for authors like Mahesh, Verstappen, Constantinescu, Perot, etc