Advisor(s)

Thomas P. Cullinane

Contributor(s)

James C. Benneyan, Emanuel S. Melachrinoudis

Date of Award

2009

Date Accepted

8-2009

Degree Grantor

Northeastern University

Degree Level

Ph.D.

Degree Name

Doctor of Philosophy

Department or Academic Unit

College of Engineering. Department of Mechanical and Industrial Engineering.

Keywords

Capacity planning, Golden section search, Heuristics, Reverse logistics, Simulation, Stochastic programming

Subject Categories

Warehouses--Management, Remanufacturing

Disciplines

Engineering | Industrial Engineering

Abstract

Manufacturing a product for the first time is generally quite different than remanufacturing the same item. The number of usable parts retrieved from returned products most often varies significantly, causing fluctuations in inventory capacity and configuration requirements. Consequently, remanufacturing requires storage designs that not only minimize warehousing space and inventory-holding costs, but also facilitate effective coordination of facilities planning and remanufacturing decisions. A well designed facility will affect the efficiency of operations and influence the operating costs and profits of a company.

The focus of this dissertation is on decision-making and modeling issues that arise in facility and warehouse designs in a remanufacturing context. Key components that decision-support systems need to address in such settings include uncertainty in yield rates and demand, reconfigurable and flexible designs, interdependencies between returned products, and type of inventory control system. In order to address the above important issues, a mixed-integer, multi-period and multi-component stochastic programming recourse (SPR) optimization model has been developed which identifies optimal schedules of internal, external, and reconfigured amounts of inventory space for a given time period. Extending the SPR model to two products with interdependent parts in a multi-period setting increases its size significantly due to the explosion of possible scenarios and the number of variables and constraints. It is recommended that heuristic methodologies be used to overcome the resulting problems of solving large combinatorial optimization models. In addition, results from these models are compared to the expected value formulations which result in a much higher minimum cost solution, underscoring the potential value in the SPR modeling approach.

In order to better emulate a generalized remanufacturing facility with random receiving patterns, component yields, and refurbished demand over multiple time periods, a Monte Carlo simulation model has been developed. Inventory storage space capacity is reconfigured as space needs change at a specified cost following a set of reconfiguration logic rules. Finally, a heuristic approach based on a multi-dimensional golden section search algorithm is implemented to identify the optimal storage capacities and reconfiguration decisions that minimize long-term expected total costs. The computation time with the heuristic approach is successfully reduced by 97% from 49.2 hours to 83.9 minutes with a higher number of inventory capacities. In several cases, total cost with this approach tends to be only 0.67% higher, which is sufficient in practical applications. Using the heuristic approach, the savings from reconfiguration are calculated under different yield rate and cost scenarios. The results demonstrate that reconfiguration becomes very important and can save a company substantial sums when the difference in yield rates among part types is high. In addition, experimental design analysis and response surface models are used to examine the impact of each inventory storage capacity on the total cost, and to develop useful heuristics for practitioners.

Document Type

Dissertation

Rights Holder

Aysegul Topcu


View Full Text

Click button above to open, or right-click to save.

Share

COinS