I think I’ve already answered this in another post but:
- A steady-state simulation cannot capture an unsteady wake from a non-“aerodynamic” object.
- Therefore, velocity residuals will never reach an acceptable value (i.e 1e-4 at least). You may notice this in your residual plot.
- Because of this, field solutions, specially those close to the surface and in the wake are not accurate and hence your drag calculations are affected.
- In addition, capturing the boundary layer requires special grid size conditions, which are related to the y+ value. Are you resolving or modelling the boundary layer? At first glance, your grid looks very very coarse around the dimples.
I hope this helps,
J.A. Gutiérrez.