Is it possible to simulate transitional flow in OpenFOAM? Or, alternatively, does it make sense to use k-epsilon or k-omega model with low Reynolds number? If not, why? What kind of errors would appear?
$\begingroup$
$\endgroup$
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
2
There seems to be a misunderstanding of the purpose of a turbulence model. Turbulence models attempt to solve a homogenized problem involving time- and space-averaged quantities. When simulating transitional flows, however, you are typically interested in exactly those finest scales that Reynolds Averaged Navier-Stokes (RANS) would homogenize. Also note that RANS is a steady-state model averaging over all scales of turbulence (because turbulence is inherently transient). "Unsteady" RANS (URANS) allows transient features, but still does not attempt to resolve turbulent structure.
OpenFOAM can do LES and (to some extent) DNS, which is more appropriate for transitional flow.
answered Aug 26, 2012 at 21:01
-
$\begingroup$ Thanks for the clarification. Is there a solver you could recommend? Perhaps PIMPLE? $\endgroup$– Igor F.Sep 2, 2012 at 21:01
-
1$\begingroup$ PIMPLE has nothing to do with turbulence modeling, it's just the algorithm for pressure-velocity coupling. As for LES, I guess dynamic Smagorinsky (openfoam.com/features/LES.php) would be a common choice, but I can't say if it's the best choice. $\endgroup$– akidSep 3, 2012 at 8:05
$\begingroup$
$\endgroup$
RANS models, more precisely those based on transport equations of $\omega$ may be used as a rudimentary transition models. Transition due to growing instabilities though, cannot be predicted by RANS models. By-pass transitions (with turbulence intensity $Tu > 1$ % )can be predicted with Low-Re RANS models owing to action of diffusion terms in transport equations which bring turbulence from free stream to near wall region, leading to production of more turbulence and then to transition.
The method however is considered unreliable.
More advanced models are those like Menter and Langtry (based on SST model) which solve additional transport equation for intermitency $\gamma$. When gamma increases it multiplies $P_k$, the production of turbulence kinetic energy. Intermitency factor itself tells you what persentage of measured time fluid flow is turbulent at certain fixed position.
If you still want to use RANS models for predictions of transition in OpenFOAM, either search for implementation of Menter&Langtry model or implement it yourself. Word goes arround that implementing new transport equation in this code is easy.
