[LiamH]

Hey CFD Experts,

I have a question for you guys concerning flash modeling through nozzles. The details of the model are below:

- liquid water enters the domain as a saturated liquid
- The liquid enters a 4mm nozzle throat which creates the onset of vapor formation (flashing)
- The fluid leaves the outlet with about 1% quality

I have experimental data to benchmark to (I am targeting about ~1.5psi) but the problem I am having is convergence since this model is extremely unstable (non-equilibrium phase change model).

The steps I have taken so far are:
1) Converge a single-phase (liquid) model at the target pressure drop (~1.5psi)
2) Using step (1), run the multiphase flash model. If unable to converge, start at a lower dP and slowly raise it.

Whats happening at step (2) is I am getting 100% reversed flow at my outlet so the model is putting a virtual wall at the outlet which is causing the model to crash. Even after changing the BC's to "opening", it still crashes. This occurs because after the 2nd iteration, the mass flow at the inlet is 100X the magnitude of the inlet AND is going backwards.

You guys have any thoughts as to why my single-phase model is not able to properly initialize the multi-phase model? I am now converging a 0.5psi single-phase model but I wonder how much effect this will have on the multiphase model.

Cheers,

Liam