[sarrazs]

Hello
I'm simulating a 3D bubble column with Euler-Euler approach. For the outlet Boundary condition, I used the same boundary described in the Ansys workshop1. The outlet at the top of the geometry is a pressure specified opening through which either air or water could flow. To preserve the
initial liquid loading set in the initial guess, an air headspace above the liquid is specified in the initial guess for the volume fraction field.
So above the liquid phase, I add an air headspace.
My first question: the air headspace is not a dispersed phase. how does cfx solve this problem at the interface where the void fraction gradient is huge. and in the headspace zone, the Navier Stokes equation for liquid is 0=0. how cfx handle this numerical problem.

My second question: when I ran the simulation in serial mode it works but it doesn't work in parallel: I don't understand why
+--------------------------------------------------------------------+
| ERROR #004100018 has occurred in subroutine FINMES. |
| Message: |
| Fatal overflow in linear solver. |
+--------------------------------------------------------------------+
map size mismatch; abort
: File exists

[ghorrocks]

Q1: The numerics have been developed to be able to handle large gradients in volume fraction. Many cases have large gradients in volume fraction, free surface modelling being an obvious example.

Q2: FAQ: https://www.cfd-online.com/Wiki/Ansy...do_about_it.3F