You may need to spend more time looking at the output to figure this out, or try some simplifications. Have you determined exactly which replica(s) are causing trouble? Can you run the 1D systems using extreme values, e.g. sample pH at 550 K or sample T at a pH extreme? How are you populating the replicas? Do you have coordinates equilibrated at 550 K for that bath? How did you choose the T values? (They seem far apart for solvated REMD.)

Rick Venable
computational chemist