A new optimization model of integrated manufacturing assembly and disassembly based on petri nets | |
| Author | Roongrat Seeluangsawat |
| Call Number | AIT Diss. no.ISE-04-04 |
| Subject(s) | Manufacturing processes Assembly-line methods Flexible manufacturing systems |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering, School of Advanced Technologies |
| Publisher | Asian Institute of Technology |
| Series Statement | Dissertation ; no. ISE-04-04 |
| Abstract | Today, there are a lot of separate models of manufacturing, assembly and even disassembly systems. However models that integrate these three systems do not exist. This research formulates first a new black token timed Petri Net (PN) based model of integrated manufacturing, assembly and disassembly systems as a pull system. The architectural design of the model is derived from a generic bill of materials in the Flexible Assembly System (FAS), a reverse bill of material in the Disassembly System (DAS) and the process plan in the case of Flexible Manufacturing System (FMS). Secondly, a new performance evaluation model based on Petri Net invariant analysis is proposed. Based on the invariants, the performance is optimized using mixed integer programming (LINGO). To compute P-invariant, the incidence matrix is required as input data. To be able to compute all invariants for the complex system, this study proposes new algorithms for computation of the invariants which cannot be determined from existing methods such as Fourier-Motzkin method and INA software. A new method to break the incidence matrix in smaller sub-matrices and refine its sub-incidence matrices for computing Petri Net invariants, faster are also introduced. At present, there is no comprehensive theory that provides a categorization of the complexity to identify all invariants. Also, the main problem when applying Petri Net to real cases is how to divide a large sized problem into small sized problems that can be solved in feasible computation time with a computer having reasonable memory storage capacity. This research categorizes the physical interpretation of the Petri Net model and identifies an approach to compute all invariants in each category. The algorithms are encoded in MATLAB. The categories are sharing resources between workstations inside a production line (Category 1) and multi-products on multi-machines sharing resources amongst production lines (Category 2). Finally, a technique to combine the Petri Net of each category is introduced to solve all p-invariants. The algorithms are illustrated on Thairung Union Car as a case study. The Petri Net is a strongly connected event graph with 610 places and 319 transitions. The number of invariants of the whole system with kanban is 380916 invariants, computed with the usage of 12 hours CPU time on a mobile computer (CPU: Pentium III 1 GHz). Finally, analysis and performance evaluation of the new model based on invariant analysis and linear programming are also introduced. LINGO is used as tool to solve the optimize model by using mixed-integer programming. From 380864 invariants, 245360 invariants resulted in constraints after initial marking stage. Then, the best kanban allocation of maximum throughput with minimum work in process is determined. The solution of the mixed integer model with LINGO used 45 minutes CPU time. |
| Year | 2004 |
| Corresponding Series Added Entry | Asian Institute of Technology. Dissertation ; no. ISE-04-04 |
| Type | Dissertation |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Industrial Systems Engineering (DISE) |
| Academic Program/FoS | Industrial Systems Engineering (ISE) |
| Chairperson(s) | Bohez, Erik L. J.; |
| Examination Committee(s) | Huynh Trung Luong; Vilas Wuwongse;Janssens, Gerrit Karel Eugeen ; |
| Scholarship Donor(s) | Kasetsart University;Asian Institute of Technology Fellowship; |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2004 |