I have implimentated a front tracking algorithm using a level-set technique. I am using it to track the contact surface between water and gas. The code maintains a pre and post front state in the cell containing the cell. Fluxes across the pre and post-front faces are combined with a flux across the front and a subvolume-change term to update the solution on each step. 1D planar problems give excellent results but I have trouble with axisymmetric problems. The front speed is taken as the average flow speed on each side of the contact surface. Unfortunately, this procedure seems to underestimate the front speed since I obtain a pressure dip in the water ahead of the front when simulating expansion of a gas bubble in water. I know that the front speed is underestimated because multiplying the computed front speed by 1.1 causes a pressure spike in the water ahead of the front.

Does anyone know a better way of computing the front speed for contact between water and gas. I am also solving 2D problems with similar difficulties.

Thanks for any advice you can give.

Don

I guess you are working in Eulerian description. The best way to track contact discontinuties is to work in Lagrangian or material description. In your problem the distortion is not much and perhaps no re-zoning will be required.