Abstract
We propose a method for multiple exchanges of job operation orders to cancel a violation of prohibition by mathematical programming. The method supports the local search for scheduling problems under no-buffer constraint. This constraint results in a blocking state which prevents any new operation after a machine completes job, and this blocking complicates a partial change of a schedule once made. When the exchange of two consecutive operations causes a violation of prohibition, the violation can be cancelled by exchanges of job operation orders on the other machine. The proposed method is for minimizing the number of exchanges. The method is also adopted to simulated annealing for improvement of a schedule planned by a dispatching rule.
