How to handle time step control for proper chemistry computation?

Dear all,

I am struggling with some Toughreact basics and maybe some of you can explain me shortly how to handle the time step control in a proper way to avoid mistakes and miscalculations.

I am simulating impure CO2 storage in saline aquifers. I am using TOUGHREACT V3.0-OMP / ECO2N.

I always thought to ensure a sufficient accuracy in the mineral reactions and chemistry balances by applying a sufficiently low TOLCH value in the solute.inp file. In my latest simulations I used a value of TOLCH=1E-7. However, in these simulations the automatic time step control eventually led to time steps between 500s and 4000s. However, in the regions of my model with high mineral reactivity (low pH), significant local mass imbalances occur. In particular I am locally loosing Ca2+, which is dissolved from carbonate phases, but does not reprecipitate as sulfate. If I manually limit the maximum time step to 50s to 200s, the results look very much better and plausible. (Unfortunately, the overall computational time now takes 12-20 days instead of 1 day). In Addition, I checked also a simulation using TOLCH=1E-8, but that did not change much concerning the mass imbalances.

Can someone give me hint which parameters I have to check in addition to TOLCH for a better utilisation of the time step control?

Best regards, Lennard