We notice jumps in the solution residual(especially Potential when using Joule Heating) and sometimes monitor points as well when we try to restart a solution from the last stopped iteration. Why does this happen? Does it impact solution accuracy?
One of the main reasons for this behavior is that the results fields are always written as single precision (even when the double precision solver is used). Rounding errors can occur when the solve is continued and this can occasionally cause a noticeable jump in residuals. But, the accuracy of the converged results is not compromised. Please try to do a mass-energy balance calculation or use the known data points to monitor for steady monitor points and use your engineering judgement to conclude whether the results are reliable.