European Journal of Operational Research 33 (1988) 245-256 245 North-Holland Case Study Allocating work between an FMS and a conventional jobshop: A case study Ludwig H. AVONTS, Ludo F. GELDERS and Luk N. Van WASSENHOVE K.U. Leuuen, Afdeling Industrieel Beleid, Celestijnenlaan 30014, 3030 Leuven-Heoerlee, Belgium Abstract: This paper discusses a real-life production planning problem in a Flexible Manufacturing System (FMS) environment. The problem is to decide which products and in what quantities to produce on the FMS. The remainder of the production requirements can be produced in a conventional jobshop. It is shown how a rather simple linear programming model on microcomputer can be a valuable decision tool. It allows for easy evaluation of various objective functions and decision alternatives. Insights gathered using the LP model have led to the introduction of new part routings which have increased FMS throughput considerably. It is also shown how the LP model is used as a screening device to select strategies to be tested in a detailed simulation of the system. Keywords: Resource allocation, linear programming, job shop, FMS 1. Introduction This paper describes how a specific production planning problem related to an FMS was solved in practice. The problem occurred in a manufactur- ing company, which at the time of the study was installing an FMS. Figure 1 gives a schematic view of the system which consists of 6 NC-machines. Machines 3 and 4 are identical as well as machines 5 and 6. This means e.g. that each operation on machine 3 can also be done on machine 4 and processing times will be identical. Selection between two identical machines is done during the operation of the system. If one of the two machines is busy, the other one is selected. Each machine has an input/output shuttle with limited capacity. Ample buffer space must prevent the system from block- ing. Three carriers take care of transporting the workpieces through the system. Received May 1987 Ten different product types (numbered 1-10) can be produced on the FMS as well as on the conventional jobshop system (henceforth denoted by CS). One new part (product type 11) is espe- cially designed for the FMS and cannot be processed on the CS. In the Appendix, the part related data are given. The production planning on the parts is done by an MRP system. For the CS demand is aggregated into bi-weekly produc- tion lots, whereas on the FMS batches are planned on weekly basis. The problem is to decide which products to produce on the new FMS and which products to produce on the conventional system. This problem has not been treated extensively in literature. Some research has dealt with the long term (strategic) problem of choosing between classical manufac- turing methods (e.g. transfer lines) and new tech- nologies. For example, Hutchinson and Holland [5] attempt to provide a measure for the value of flexibility by comparing the costs of a typical fixed automation approach with those of a trans- 0377-2217/88/$3.50 © 1988, Elsevier Science Publishers B.V. (North-Holland)