Why do you only need one solution for the whole problem?

Suggested Answer

One solution to the whole problem is sufficient. You will simply extract the part of it that you wish to work on. This is possible for two reasons. First, operations such as store and restore on solutions ignore parts of the solution that are not extracted. Therefore, a solution can cover more variables than have been extracted with no problems. Second, Dispatcher's neighborhoods (such as IloRelocate) ignore nonextracted parts of the solution and only generate neighbors for extracted parts. The result is that you can perform local search over an extracted part of the routing solution, while the remainder remains unchanged.