Crane scheduling with time windows in flow-shop environments
Abstract
An M-crane, N-tank scheduling problem in a flow-shop environment is studied. This kind of problem can be found in flexible electrochemical processing lines, where a line typically consists of one or two cranes, and 10 to 20 tanks. Such lines can process different job types with random job arrivals. In each tank every job has a minimum required processing time and a maximum allowable processing time. A job can be picked up from a tank at any time between the minimum and maximum processing times by one of the cranes. The objective is to maximize throughput subject to the processing time limits and the capacities of cranes and tanks. In this research, we analyze the problem and discuss the feasibility of system states. Then, based on the insights derived from the mathematical formulation, an effective heuristic approach is developed. That combines real-time scheduling and mathematical programming. It can handle multi-type job scheduling and overcomes the problem of cyclic scheduling (the existing cyclic scheduling approaches can only be applied to systems with one type of job). Moreover, while the existing real-time approaches that are knowledge based often result in the production of defective jobs, the proposed approach does not. We conducted an extensive computational study on one-crane systems and two-crane systems. Several crane selection rules and job selection methods we proposed have been tested. The results indicate a substantial (100%) improvement over the existing methods for maximizing throughput in the one-crane systems. Unfortunately, for systems with more than one crane there are no existing computational results to which our results can be compared.
Degree
Ph.D.
Advisors
Yih, Purdue University.
Subject Area
Industrial engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.