Help needed: Large Phase and Composition Change in GEM model

Hello Friends,

I am recently working on a project of CO2 storage in hydraulically fractured shale gas reservoir with EGR. However, the model have great numerical difficulties as the pressure goes up, changing from 80% gas saturation to 80% water saturation in one time-step and gas phase disappearing. Does anybody have experience how to deal with such a difficult situation? I have tried to loose the converge tolerence and use smaller time-step but not helping much.

Many thanks