Modeling And Research On Disassembly Planning Of Returned Products With Uncertain Yield

With the increase of population and industrial development,environmental-ecological and resource crisis are getting serious. Recycling andremanufacturing became a process to get ready for environmental friendlyinstructions, as well as a way to attract consumers by “green” image and createeconomic values by the low-cost remanufactured products. End of use productsare taken back from users and disassembled into parts. Disassembling is aprocess where the input is a single product while the output is multiple parts.Disassembly amount of parts are predefined by the product, while demand doesnot follow this relation. And thus, inventory in disassembling is unavoidable.The quality of returned products are unpredictable, resulting uncertainty in theamount of recoverable parts.This thesis developed two disassembly planning models for multipleproducts in multi-periods, taking yield rate and capacity restrictions intoconsideration. First model studies disassembly lot sizing problem, where yieldfollows a given stochastic distribution with stochastic demand. A Lagrangianheuristic algorithm is used to solve the problem. The study shows that theoverall cost quasi exponentially increases when variance of demand increases,and quasi logarithmically increases with inventory holding cost of single items.Second model studies disassembly lot sizing and returned product pricingproblem, where the yield is a given function of returned product price. ParticleSwarm Optimization (PSO) method is adopted to set a proper price for thereturned products and then a Lagrangian heuristic similar to that in the firstmodel solves disassembly lot sizes. Computational test on the randomlygenerated problem shows that PSO gives the result in acceptable time period.The study shows that PSO results are better for short period problems with moreproduct types, and suggests higher returned product prices when the inventoryholding cost or disposal cost of unqualified parts are higher.