EOS7C poor convergence, e.g., in subroutine SOLIT
EOS7C v.1 models five components (water, brine, CO2 or N2, tracer, and CH4) and heat. Many users have noticed that time-step convergence can become poor during CO2 injection problems, especially when the other gas components (tracer and CH4) are not present. This poor convergence sometimes manifests itself as failure to converge in Subroutine SOLIT.
If you experience poor convergence, or simply want to avoid it preemptively, I suggest you try to ensure that there is always some of each of the gas-phase components present. For example, in a CO2 injection problem, you can inject one part in 10^4 or 10^5 of tracer and CH4. E.g., if injecting 1 kg/s of CO2, try also injecting 1.e-5 kg/s of CH4 and 1.e-5 kg/s of tracer. Similarly in a CH4 injection problem, try injecting a small amount of CO2 and tracer. This additional injection will not affect the physical system in any substantive way, but will improve the linear algebra and thereby improve overall convergence.