I'm modeling a 3-D heat exchanger that has opposite direction laminar water flows on either side of a thin aluminum plate. The remainder of the water is bounded by PVC.

I've found that at an Energy convergence criterion of 1e-6, the heat exchanger does not have balanced heat flows. But iterating Energy to 1e-7 and beyond greatly improves this situation. It also more closely agrees with experimental data.

But when I continue these iterations, I notice that the continuity residual oscillates between 1e-3 and 1e-4 (converged by Fluent's default criterion) while the X/Y/Z velocities and Energy continue a nice downward slope. I'd like to know what factors may cause the continuity residual to behave this way.

Also, I've found that decreasing the momentum underrelaxation factor causes the continuity residual to immediately drop by a noticeable amount. But it then continues its oscillatory behavior at the new lower value. Any explanations out there?

I'm using first order upwind discretization and simplec PV coupling. I've lowered the momentum UR factor to .2 and the Energy UR factor to .9. All others are default values.

Thank you!

The oscillating continuity residual could mean that the magnitude of your residual is on the same order as the numerical accuracy of the calculations. It doesn't necessarily mean that anything is wrong.

I agree with Evan. Have a look at your mass balance. Or track an other value during iteration using monitors.

Laika, still orbiting