[stupidlyme]

Hi everyone,

I'm simulating flow around a fin and tube heat exchanger in 3D, looking specifically at 2 tube rows.

Problem I'm having is simply that my solutions are not converging enough to give a more accurate result (residuals only to the power of 1e-3). At first, I had to apply a 1st Order UDS to the momentum equations in order to stabilise the convergence. However, I needed more accurate results and refined my mesh further. Yet, comparing my unrefined and my refined mesh (2x more nodes at 300k nodes) yielded only a 0.69% difference in the Nusselt number.

Physically, this could be because of vortex shedding and unsteadiness behind the cylinders (I'm simulating for steady, laminar and incompressible as per literature).

I've toyed around with altering solution methods, under-relaxation factors, mesh refinement, to little avail. Any advice on how I can enable a better convergence would be much appreciated.

Many thanks in advance.
Brian