A GA-Benders algorithm for lot sizing and scheduling in a multistage production system with sequence dependent setup | |
| Author | Terasak Jitngarmkusol |
| Call Number | AIT Thesis no.ISE-05-10 |
| Subject(s) | Production scheduling Genetic algorithms |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Advanced Technologies |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. ISE-05-10 |
| Abstract | In this study, lot sizing and scheduling in a multistage production system with dependent setup problem is focused. Setup cost and setup time are depending on the preceding production. The problem is a mixed integer problem (MIP) with the objective function to minimize total cost. The total objective function cost consists of integer programming part (IP) and linear programming part (LP). These parts are related to setup cost, and production, inventory and purchasing cost, respectively. Inventory balancing, lot-sizing production and limitation of capacity are main constraints. In fact, Genetic Algorithm and Benders Decomposition are parts of GA-Benders algorithm. The problem is separated into IP part and LP part using the Benders Decomposition concept. A GA is applied to solve the IP part. In contrast, the LP part is computed by CPLEX. The fitness value is determined by the master problem. Bender cuts are added to the master problem at the end of each generation in order to make its feasible region tighter so that the solution can converged sooner. GA parameter experiments are conducted in three sizes of problems. Nocut_all algorithm and Nocut algorithm are comparable algorithms that do not use Bender cuts. Convergence index and gap index are used to measure the quality of the algorithm evaluation. The results show that the GA-Benders algorithm yields the effective lot size and schedule. In most of the cases, the quality of the purposed algorithm are worse than Nocut_all algorithm less than 7%; however, the computational time is much shorter. The algorithm is more efficient when dealing with large-scale problems especially in terms of computational time. |
| Year | 2005 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. ISE-05-10 |
| Type | Thesis |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Industrial Systems Engineering (DISE) |
| Academic Program/FoS | Industrial Systems Engineering (ISE) |
| Chairperson(s) | Anulark Techanitisawad; |
| Examination Committee(s) | Huynh Trung Luong ;Bohez, Erik L. J. ; |
| Scholarship Donor(s) | Royal Thai Government Fellowship; |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 2005 |