[andrew-c]

I am trying to simulate a distillation process between two columns of water at different temperatures which are connected by a partition/valve. The system takes advantage of hydrostatic boiling, where tubes of water are raised up quite high (~10 m) to lower the pressures and produce phase change at close-to-ambient temperatures. My geometry represents the top ~0.1 m of the system to minimize complexity.

Gravity is enabled, and I have open pressure boundaries at the bottom of each tube. I've attached a schematic for clarity.

At the start, I initialize the system with only liquid water. Based on the low operating pressure, the effects of gravity, and my multiphase model (Lee model for boiling/condensation and species model for evaporation/concentration-driven mass transfer, with temperature-dependent/pressure-dependent properties), the water boils on both sides. After about 15 seconds, I get a layer of vapor at the top of each column. The pressure on each side corresponds to the saturation pressure of water at the given temperatures (which I believe means they are in the saturated vapor state).

Up to this point I believe the model is correct, but after I open the mesh interface to start the diffusion process, I get lost. From theory and experiments on these systems, vapor should diffuse from the hot side to the cold side because the vapor pressure of the hot side is greater, creating a concentration gradient. This is only possible due to the saturated vapor conditions on each side.

However, when I open the interface, the pressures equalize instantaneously. Their value is about the average of the two temperatures, but the temperature of the vapor region does not support this.

I'm hoping to get some ideas about how to treat saturated vapor in Fluent multiphase. I know I start with saturated vapor, but after I open the interface, the pressures and temperatures no longer correspond to saturation conditions. Any advice on why the pressures are equalizing, or how to preserve the saturation conditions, would be greatly appreciated.

Images of density and pressure state before partition opens are attached.

Model details on model:
Eulerian multiphase; Lee model for boiling/condensation (pressure-dependent saturation temperature via UDF); species model for evaporation/concentration-driven mass transfer (temperature-dependent saturation pressure via UDF); Pressure-based solver; dt = 0.001 s; All URFs default besides energy (0.1);