I normally log my batch runs (use option "-log fileName.err"). If the model doesn't converge, a note will be written to that file. You can manually inspect those log files or write a little code that reads and checks them (depending on the number of runs your dealing with) after your batch run.
As far as I know, that's the only way to check for convergence. It would be nice if the Error Norm History was saved in the DAC file, since that would provide more information.
Hope that helps, Chris