Performance evaluation of unreliable production lines using fuzzy linear programming and chance constrained programming

Cycle time is an important indicator for capacity evaluation. Not only discrete event simulation models but analytical models can also help elucidate the system behavior. Johri [1987] presented a linear programming model to estimate the capacity of a production line manufacturing multiple products with finite buffers. However, his analytical model didn't explore explicitly the effect of the machine's accidental stoppage and the repairing behavior on cycle time. The objective of my thesis is to expand this analytical model by taking into account the phenomena of random failure and repair of the various machines composing the production line.; A fuzzy linear programming model is first utilized to solve the capacity estimation problem in the manufacturing indicated by imprecise information of cycle time. Meanwhile, since machine failures are described as the problem of a stochastic program, chance constrained programming technique is applied to convert the probabilistic nature of the problem into an equivalent deterministic situation. This approach allows mathematical convenience in the numerical manipulations of down time approximation.; The numerical comparisons with discrete-event simulation on a two-machine production line indicate that this analytical model is an alternative solution for capacity estimation when machine performances are unreliable. Then, a numerical example of a five-machine production line case illustrates the application by the analytical approach on the longer production line. Later, through cycle time estimation in a case study, results show that a machine with frequent failures but short repair time is superior to one with infrequent failures but long repair time, if the machine availability is the same.