Hi Glenn and Robin,

I am simulating overexpanded flow through a bell nozzle. I get massive flow separation. I am using transient run with SST model for turbulence.

I am using adaptive timestepping. When i had set target RMS residual as 5e-5, the MAX residual was >1e-3. The p-mass imbalance was plus-minus 0.05% and H-energy imbalance was plus-minus 1%. Since the MAX residual was >1e-3, I set target MAX residual as 1e-3. Now the timestep reduced. But p-mass imbalance became plus-minus 1% and H-energy imbalance became plus-minus 2%.

I had expected the imbalances to reduce with the decrease in MAX residual!?

thanking you,

Hi,

A converged solution will have both small residuals and small imbalances. A non-converged solution will have either or both residuals and imbalances too big. Generally improved residuals gives improved imbalances but not always.

Also the imbalances may or may not be significant in your simulation. Do a sensitivity study and establish whether you need it or not.

Glenn Horrocks

Hi Prabhu,

As Glenn has suggested, check whether your solution is actually converged. It may also be the case that 1e-3 MAX was insufficient and you just "got lucky" in having the mass imbalance lower on your earlier run. If you plot the imbalances in the solver manager you can see that they often oscillate between positive and negative values.

The residual is essentially the normalized imbalance across a control volume. In a steady state simulation this imbalance is calculated entirely based on the fluxes and source terms but leaves out the transient term, since the accumulation (which may be positive or negative) of mass, momentum, etc. within a control volume should be zero at steady state. In a transient simulation the residual includes the transient term, as the accumulation is an important transient feature of the flow, hence the difference in residual values people often report when they switch to transient.

Regards, Robin