**7. Conclusion**

In this chapter, we have demonstrated that Benders' decomposition algorithm for solving the capacitated plant location problem can be accelerated substantially when the master problem is solved heuristically. The hybrid Benders/GA algorithm is a variation of Benders' algorithm in which, instead of using a costly branch-and-bound method, a genetic algorithm is used to obtain "good" subproblem solutions to the master problem. The numerical example shows that the hybrid algorithm is effective to solve the capacitated plant location problem. The results imply that the hybrid algorithm is much more practical when only near-optimal solutions are required. Future work could extend the proposed algorithm to other location problems.
