Based on earlier research on optimal disassembly sequence generation, notably graphically supported search methods, a method for solving general optimal disassembly sequence generation problems by linear programming has been developed and described. This method puts no serious restriction to the size of the model. It is adaptable to changes in model structures and constraints, and it is even not restricted to divergent operations, such as disassembly. This is demonstrated by solving the combined disassembly/clustering problem. Clustering is the combination of different disassembly products to categories that can be considered — and sold — as a single product, e.g. steel parts. The theoretical model is validated and clarified by its application to some standard problems from literature.