Search

The problem considered in this work is a practical problem of a Croatian olive oil producer. In the Croatian region considered in this paper, Istra, Individual farmers harvest their olives room the middle October till the end of December. Traditionally, the harvesting of the olives is accomplished by hand. After this operation the olives are transported to the olive oil producer for the production called olive oil extraction process. The best case would be to process the olives immediately. But, in the practice could happen that the harvested olives should wait for the production or should be stored causing the additional cost that we call the storage overtime cost. If the producer decides not to store the olives but to leave them stored inappropriately, this way of storing the olives can last for a few (2-3) days. Otherwise the olives are not good enough for the production of extra virgin olive oil. Also, some olives are of good quality, some of them not causing the different qualities of the olive oil produced (extra virgin and virgin olive). So, the producer has to clean the machine very carefully every time when the machine switches from one kind of olives to another. This process requires sequence dependent setup times of the machine called decanter having a certain capacity. In order to keep customers loyal and to be competitive at the market the producer wants to satisfy the demand with the minimum overtime cost and minimum setup, storage, waiting and production cost. Thus, the producer has two objectives. Minimum overtime cost will satisfy the customer’ s wish to produce their olives as soon as possible while the minimum setup, storage, waiting and production cost will keep the producer's costs as low as possible. Since the priority of the producers is to keep customers loyal the problem is modeled as a bilevel mixed integer programming production planning problem where high level manager decides about minimizing the overtime while the low level manager decides about the production. The problem is an NP-hard problem and for solving it the heuristic based on taboo search, variable neighborhood method and Lagrangean relaxation is proposed.