Periodic Baffle Minor Loss: models, wall treatment, convergence
 

Hello,

I am interested in determining the minor loss associated with steady flow through a channel around alternating baffles for different baffle lengths and spacing (ideally within 2-3 significant figures). Using ANSYS Fluent 2021 R2, I have set up a periodic inlet/outlet condition targeting a specific flow rate, such that Re > 2e3 (based on hydraulic diameter).

My primary concerns are: (1) which viscous model is best suited for the flow described (2) whether I have the correct near-wall treatment (3) how to know if my results are sufficiently converged (4) if there is a preferred solver for the given model and flow conditions.

I was thinking to use realizable k-epsilon or k-omega SST, but I was wondering if anyone had an intuitive sense of which model would be best suited to this flow.

In the event that I use a k-epsilon model, I was also wondering whether standard wall functions or enhance wall treatment is more suitable. I am thinking that I do not need to resolve the viscous sub-layer, since my interest is more in the overall periodic pressure drop across the inlet/outlet, so I was thinking to use standard wall functions. However, I don't really know how the bulk flow condition and measured pressure drop might depend on the wall treatment.

From simulations that I have run with realizable k-epsilon and standard wall treatment, the solution seems to converge rapidly, with scaled residuals below 1e-6 for the shorter baffle lengths (fixed spacing). However, for longer baffles I can only get scaled residuals below 1e-3 (largest residual representing continuity). In both cases, a monitor of average wall shear and inlet pressure do reach steady values. Nevertheless, I am uncertain as to whether the reported pressure drop in these cases can be trusted. There are some values in the literature as well as experimental values that I can compare against, but only for a select range of geometries. I tend not to think of this as a meshing issue, since it is a fairly simple geometry, I was able to obtain good mesh quality for all hex elements within the recommended y+ for the standard wall functions.

In any case, I appreciate any insights on the above points. To illustrate the flow geometry, I attaching an illustration of the flow domain. Thanks for the help, and let me know if you have any questions.

Sincerely,
Andrew