[dave13]

Hi,

I have a question regarding my choice of the turbulence model for my two-way FSI simulation as it seems to be not entirely correct.

I'm simulating flow through a channel (6x2.5x40mm) with a circular membrane (d=5mm) on the bottom which is inflated to give 15% strain. It's a steady state simulation with flowrates from 0.1 - 0.5 m/s.
At the inlet, I input a velocity profile from a previous solution with a longer geometry to have fully developed channel flow at the inlet. Additionally I specify the turbulent kinetic energy and dissipation at the inlet from this boundary condition profile (previous simulation with medium turbulence option). The flow is turbulent along the entire channel.

What we are interested in is the wall shear distribution on the membrane, so it is important that we get a correct result at the 'wall' (membrane) and apply the best turbulence model for that. I am though not sure if I am using the correct turbulence model as it seems to predict a too high wall shear compared to experimental data.

I used the SST model which to me seems most suitable for the case. I used it with the production limiter option (clip factor 10) because of stagnation points we get on the membrane and curvature correction (production correction factor 1.0). It does give similar results as with SST only (no advanced options), but slightly lower WSS. It predicts separation behind the membrane for all flowrates, unlike the k-e model (which does not give good near-wall results as expected though). I also tried to check the results with reynolds stress models, the BSL model confirmed the SST results whereas the SSG model gave similar results as the k-e model.

I am confused though as the SST model still seems to give too high wall shear values. Any inputs/suggestions what I did wrong, or which model you would apply?

Thanks a lot!

[bharath]

Hi dave
You are doing it right.Each turbulence model have its own advantages and also disadvantages.Which one you think right after your detailed analysis?

[ghorrocks]

I would go back to basics here. I would ensure that your turbulence models can predict wall shear along a straight duct first. Then find a benchmark result which has a turbulent duct with a similar obstruction in it - hopefully you can find a case. You could use an airfoil blade boundary layer if you have to, but hopefully you can find something closer to your case. Once you see how the various turbulence models go on these benchmarks cases you can then approach your real case with more confidence.

[dave13]

Thanks for your input! I will look into it.