Master of Science (MS), Ohio University, 2019, Industrial and Systems Engineering (Engineering and Technology)

The objective of this thesis is to explore the problem of scheduling jobs on unrelated machine in the presence of ready times and deteriorating processing times. The objective of the schedule is to minimize one of five performance measures including average flow time, total tardiness, maximum tardiness, number of tardy jobs, and makespan. Two methodologies are proposed to solve the problem: a constraint programming model and a genetic algorithm. Eighty data sets are created using four generator parameters. The constraint programming model is tested using these data sets for the five performance measures and is in many cases able to find optimal solutions to the problems. The genetic algorithm is tested against sixteen of the eighty problems for every performance measure but with four additional genetic algorithm specific parameters, generations, population, crossover probability, and mutation probability. It was also able to find optimal solutions but not with the same frequency or speed of the constraint programming model. The two solution techniques are compared statistically and the constraint programming model is found to be definitively better at producing higher quality results. Three of the four genetic algorithm parameters are tested for their standardized effects on the result but not one parameter or interaction between any number of parameters is found to be consistently statistically significant across all performance measures.

Committee: Gursel Suer (Advisor); Tao Yuan (Committee Member); Dusan Sormaz (Committee Member); Ashley Metcalf (Committee Member) Subjects: Industrial Engineering