Development-Scheme Selection for River Valleys

This paper presents a new approach to the problem of selecting a development scheme for a river valley from preliminary surveys on candidate sites. Rather than proceeding by elimination, this approach identifies the scheme which minimizes the investment and operating costs. This minimization problem is decomposed using the Benders technique into a master problem covering the site selection and sizing aspects and a subproblem covering the production aspect. The former is solved by mixed-integer linear programming whereas the latter is transformed by a Lagrangian relaxation into a nonlinear network flow problem which is solved by a variant of the Frank-Wolfe method. Numerical results are reported.