Design Mixd Model Assembly Lines With Adjustable Production Capacity

Mixed model assembly lines(MMALs)are widely used in many industries such as automobiles,white goods and consumer electronics as an efficient manner to produce a series of products.A core decision in designing an assembly line is the assignment of tasks among stations,which determines the cost and efficiency of the line and is commonly referred to as assembly line balancing problem(ALBP).ALBP has been stuided for over half a centenry,and most of previous studies design the assembly line based on constant demands for all products.However,the reality in many industries is that the production plan for the assembly line is frequently changed in order to cope with the flucuation of market demand.When the realized demands deviate from what are predicted,the MMAL may be not as efficient as designed,or even not be able to finish all jobs on time until some prodcuiton adjustments are taken,such as ovetime work,hiring temporary workers.The frenquency and cost of these adjustment depend on the initial balancing and the extent of demand flucutation,but these factors are seldom considered in previous studies.This disseration focus on the performance of the MMAL in an environment of unstable demands.The produciton capacity of the MMAL can be adjusted when demand change occurs.By incorporating the cost of capacity adjustments into balancing decision,this disseration aims to design a MMAL which is readily adaptable to demand changes,and to lower down the average cost in long run.Three methods to adjust production capacity are discusseed: hiring utility workers,allowing overtime work,or varying the number of stations by redefining boundaries,and their effects on balancing decision are seperately studied.The main results of this dissertation are summarized as follows.1.The MMALBP with utility workers is discussed.The utility workers are hired to help ordinary workers in case work overload occurs in any station.A utility worker is multi-skilled so that he can help different stations,and hence more expensive than an ordinary workers.Due to the complexity of their work,utility workers can't be fired or hired at will.The decision maker need to determine the number of ordinary workers(stations)and utility workers,as well as the task assignement for each station,in order to satisfy demands in all scenarios with minimal labor cost.Some useful properties of the problem are characterised,and a recursive method to estimate cost lower bound is presented.A heurisitc is proposed to quickly find a feasible solution,and the reuslt is used as an upper bound for the branch,bound and remember(BB&R)algorithm,which is developed to solve the problem exactly.Numerical experiments on 500 randomly generated instances show that BB&R algorithm finds and verifies optimal solutions for 406 instances.For the remaining 94 solutions whose optimality are not verified,the average gap from lower bound is 5.17%.2.The MMALBP with overtime work is studied.On the MMAL,each work shift is followed by a down period,where overtime work can be scheduled whenever demands cannot be fulfilled in normal shifts.Overtime is paid at a higher rate than normal shift,and the maxmal length of overtime is restricted by the law.The balancing problem concerns about the determining of the number of stations,the task assignment of each station and the amount of overtime in each scenario.The objective is to satisfy demands in all scenairos with the minimum labor cost,which is the wages paid for both normal shifts and overtime work.An iterative method is proposed to estimate the lower bound on the total labor cost.Based on the lower bound,a heuristic and a BB&R algorithm are developed to solve the problem.The proposed algorithms are tested on the same data sets which contain 500 instances.BB&R algorithm optimally solves 408 instances,and gives high-quality solutions for the rest within 60 seconds.For those solutions whose optimality are not verified,the average gap from lower bound is 2.67%.3.The MMALBP with fixed task sequence and changeable station boundaries is investigated.It follows the idea of Simaria et.al.(2009)[3]: the number of stations can be changed as demands change,but the task sequenc is fixed to reduce reconfiguration cost.The decision maker determines the task sequence,the number of staitons and the staion boundaries in each scenario.Differing from the research of Simaria et.al.(2009)[3],this problem is formuated with the objective of minimizing average number of stations used in a straight MMAL,and some structural results are presented.For a given partial task sequence,an optimal procedure to determine station boundaries in each scenario,and a method to estimate lower bound are presented.A swap heuristic is proposed and incorporated into a single-pass heuristic,which can quickly find a feasible solution.The solution is used as an upper bound for the BB&R algorithm which is designed to optimally solve the problem.The proposed algorithm is tested on the same data set of 500 instances.308 instances are optimally solved within 60 seconds,and the optimality gaps of the others are low.The gap between the not-verified solutions and lower bound is 4.05% on average.