Learning Forum

[navidvd]

Hi there,

I am working on a steady simulation of flow inside a pipe (representing a coronary artery model) at a low Reynolds number (the bulk Re is about 500). Some of the cases I am working on have a narrowing in cross-section (representing stenosis) which makes the the local Reynolds number higher (around 1500 art maximum). I initially used a laminar solver for these simulations as the flow is laminar and the laminar solver just solved the NS equations without adding any turbulent models to the simulations. However, for the higher blockages, the choice of laminar solver tends to diverge for some of the cases. I increased the number of mesh significantly and it somehow helps with the convergence of the simulations. 

My question is: Should I stick to the laminar solver for the cases with lower cross-section (higher local Re) and increase the number of elements, or it is better to use some of the RANS models?

PS: For one of those cases, I compared the results of the laminar solver and SST k-omega one and they showed completely different behavior and results. Not sure which one was the correct answer. For the SST k-omega simulation, the jet formed after the local blockage diminished very quickly; however, it continued at a longer distance for the laminar solver.

Any thought on this is appreciated. For these sorts of simulations which approach results in a more accurate simulation outcome?

Thanks

Navid

[SRP]

Hi,

The choice between using a laminar solver or a RANS (Reynolds-Averaged Navier-Stokes) model for simulating flow inside a pipe with stenosis largely depends on the specific physics of your problem.

For cases with stenosis where turbulence is expected, using a RANS model like SST k-omega with proper setup and mesh refinement is generally more accurate.

[navidvd]

Thanks for your response, 

But even at the location of stenosis where we expect the most turbulence, the Re number is within the laminar region, and the jet flow induced by stenosis disappears shortly after the stenosis. What if we choose the laminar solver and use a finer mesh to solve the NS equation without any turbulence modeling?

I am worried about using the SST k-omega model and introducing artifacts to the simulations.

Thanks

Navid