PLANNING SECTORAL INVESTMENTS WITH ECONOMIES OF SCALE: A COMPUTATIONALLY EFFICIENT MIXED BIVALENT PROGRAMMING APPROACH

Özet

In this paper, a sectoral investment planning model is examined from computaüonal point of view. It is a well known fact that conventional mixed bivalent programming models with complex combinatorial structures are generally intractable. The paper exploits such a model to determine what capacity, if any, should be maintained at the various geographical regions during the planning period in order to meet regional demand and minimise total cost of the entire system. A Lagrangean relaxation-basedprocedure is developed to decompose the model into submodels by each geographical region and planning year. Following that, computationally efficient algorithms to solve the submodels are presented and computational results are discussed