Abstract
Formerly, Leachman and Carmon【1992】 have presented a novel approach for modeling the alternative machine capacitated loading problem for Linear Programming (LP) production planning. This approach makes modeling of complex processes with re-entry of alternative machines in large scale LP production planning possible. The feature of the alternative machines with a re-entry process is very common in today*s manufacturing environment, especially in semiconductor industry. However, there are two difficulties in their approach. First, in order to utilize the approach, a uniform assumption must be made. The uniform assumption states that the processing speeds on alternative machine types are proportional across all of the operations that the alternatives perform. However, in industrial applications this assumption may not be appropriate in certain cases. In addition, this approach creates new capacity constraints by forming unions of alternative machine sets with common machine types. Performing union operations will increase the number of capacity constraints, which may increase the solution time of the LP formulation. To ease these two difficulties, this thesis presents a modified hybrid modeling approach while still utilizing the concepts proposed by Leachman and Carmon. This study use computer experiments to compare Leachman and Carmon's model and the modified hybrid model. The modified hybrid model need less cpu time and memory when it generates capacity constraints than Leachman and Carmon's model. But the Leahman and Carmon's model takes less cpu time while solving the model than the modified hybrid model.