Hi all,

i'm running a low Re centrifugal pump simulation on SST. my maximum y+ is 6.8. how great an effect will it have on the accuracy compared to having a y+ below 2?

also, should i start with a larger or smaller timescale and should i change it at some point? i've one simulation where i start with auto timescale for 600 iterations (reaching about 5e-4) and after i increase the timescale factor by 10 the convergence plot drops steeply to below 1e-4. for another simulation where i start with a timescale factor of 10. the solution almost reaches 1e-4 in about 150 iterations but then starts fluctuating and does not go below 1e-4.

Well, Y+ below 2 is recommended. How different will the solution be is a question that no one can really answer without proper experimental results.

Since we use pseudotime anyway to solve for steady state, it doesn't matter what timestep you take as long as get to your residual target values. Try to use monitor points with variables of interest to see how they change with increasing time.

Regards

Omer

Hi,

Omer answered the y+ question.

For the time step, sometimes you need a small timestep to simulation going without diverging, but once you reach the region where the residuals are reducing nicely with iterations you can often increase the step size. A larger step size often converges better as it does a better job of "averaging out" the small instabilities and vortex shedding.

Glenn Horrocks