Research On Service Rate Optimization And Load Balancing Of Mixed-model Manufacturing System Based On Queuing Theory

With the increasing of severity of market competition as well as customer demand forpersonalized products, mixed-model manufacturing systems are generally adopted inmanufacturing enterprises. They can product a variety of different classes or differentnumbers of products within the same manufacturing system without changing the facilitylayout and so have a high flexibility. But during the manufacturing process of multitypeproducts, whether a manufacturing system can produce synchronously and proportionallydepends on its group technology of products, its organization of manufactured productsqueues, its service rate and load balancing level. Besides, a manufacturing enterpriseemploying a better service rate level and balanced production capacity can hold an advantageposition in competition on the price, quality and the ability of rapidly responding to themarket. As a result, the mixed-model manufacturing system service rate optimization and loadbalancing problems should be studied.Since queuing theory can be applied to all the service system and the studies on themixed-model manufacturing system optimization problem based on queuing theory arelooked inadequate by far, in this thesis, the mixed-model manufacturing system service rateoptimization and balancing problems are studied based on queuing theory. Considering thatmixed-model manufacturing systems have the characteristics of high flexibility, productsqueues sequencing and so on, an open type queuing network model of a mixed-modelmanufacturing system can be built base on queuing network model. Since there are multipleclasses of products in the system, they are aggregated on the principle of products grouping.As a result, the whole mixed-model queuing network is simplified to be a single productmodel queuing network in which every work unit can be modeled as a G/G/1queuing system.Then the mathematical model of the mixed-model manufacturing system service rateoptimization problem is developed. To develop the expected cost model of every work unit,the sum per unit time of the products staying cost and the work unit service cost is consideredas the objective function and the traffic intensities when work unit is producing every class ofproduct as the constraint condition. Since the model is difficult to be solved, by importing thepunishment function, the initial constrained problem is transformed to be non-constrainedproblem. The transformed objective function is solved by the optimization methods such ascyclic coordinate method, golden section method, advance and retreat method and theMATLAB procedure to obtain the optimal service rate for every class product.Next, the mathematical model of the mixed-model manufacturing system balancingproblem is developed according to the transition probability matrix of every class of product.Based on the results of mixed-model manufacturing system service rate optimization, the mean processing time of the aggregated product at every work unit can be obtained byanalyzing the product transition probability matrix and the typical queuing model of a singlework unit. The combination of the work stations number and the load smoothing index amongthose stations is considered as the objective function and the product processing order as theconstraint condition. Then a genetic algorithm is presented and a C program is designed tosolve this problem.At last, a numerical example of toy car mixed-model assembly line is researched toverify that the optimization methods based on queuing theory are feasible and practical. Thenumerical example shows that when every work unit of the mixed-model manufacturingsystem is in a superior service rate level and the whole system is in good load balancingcondition, the mixed-model manufacturing system can keep working smoothly and utilizingresources fully.