[Pablok]

Dear CFD Online community,

I'm a PhD student actually dealing with the reactingFoam solver, which I use for combustion simulation. I intend to modify it and add some functionalities (differential diffusion model, conjugate heat transfer, acoustic boundaries conditions ...).

First I'm trying to fully understand the reactingFoam solver and its implementation. It uses the rhoPimpleFoam solver, which is a pressure based solver for compressible flow.

I recently saw this video https://www.youtube.com/watch?v=HZwvtLgxJy4&t=5269s, which is a webinar of Prof. Hrvoje Jasak talking about reacting flow in OpenFoam.
At about 1:28:40 of the video someone says that: "rhoPimpleFoam isn't really compressible".
Prof. Hrvoje Jasak then gave an explanation that I didn't fully understand, isn't a solver considered compressible when it takes into account the density variations? Or this definition is incomplete?

Thanks for your help!

Pablo

[sharonyue]

For compressible flows, some are doing calculations with really high Mach number, some are doing heat transfer with weak compressibility. The student said "rhoPimpleFoam is not a really compressble solver" by admiting that "only density-based solver is a real compressible solver", like that Roe method or HLL method, and any other kind of approximate Riemann solver. In this case, rhoCentralFoam is a real compressible solver which employs central scheme and Roe/HLL scheme can be implemented similarly.

As Hrv said, rhoPimpleFoam is also a highly validated solver but its pressure-based solver instead of a density based solver. It does not have the entropy fix so for those who do reseach mainly on hyperbolic equations, rhoPimpleFoam is not perfect. Oh the otherhand, rhoPimpleFoam is good enough to capture good shock resolution and other numerical aspects, maybe not high enough resolution though.

[Pablok]

Hi Sharonyue,
Thank you very much for your quick answer. It is more clear now.

I compared "rhoCentralFoam" with "sonicFoam" which is the implicit pressure-based solver on the "forwardStep" case. For sonicFoam, the shock waves pattern after some time is smeared and does not have high resolution (see attachment to illustrate: rhoCentralFoam left vs sonicFoam right). I guess that in this case a density based solver is more suited.

For my case, at the moment I mostly do combustion simulation but at low Mach numbers in openfoam (using "reactingFoam"). So implicit pressure-based solver should be fine for now.

Pablo