The Compartmentalized Knapsack Problem is a variation of the classical knapsack problem and can be stated considering the following hypothetical situation: an alpinist must load his knapsack with items. Two numbers, a weight and an utility value are associated with each item (so far, the problem coincides with the classical integer Knapsack Problem). However, the items are of different classes (food, medicine, utensil, etc.) and they have to be packed into separate compartments inside the knapsack. The compartments are flexible and have limited capacities. Each compartment has a fixed cost that depends on the class of the items and, in addition, each new compartment introduces a loss on the capacity of the knapsack. The Compartmentalized Knapsack Problem consists in determining suitable capacities for each compartment as well as their packing. The objective is to maximize the total utility value paid off the cost of the compartments. In this work, we model the problem as an integer non-linear optimization problem and we design some heuristic methods for its solution. A practical application of this problem arises in steel rolls cutting subject to lamination, which is discussed in detail in the appendix.
knapsack problems; cutting and packing problems; integer and combinatorial optimization
