This research is motivated by the needs of a general manufacturing plant, where several material handling technologies may be used to satisfy the material handling requirements. The problem of selecting and specifying material handling systems for manufacturing operations is addressed. This is a complex problem, because of the flexibility of material handling technologies: often there is more than one technical solution. In addition, material handling technologies exhibit considerable fixed costs. As a result, it is often better to select fewer technologies and assign some tasks to technologies that are not the “first choice” based on criteria that might be used in an expert systems approach.

In earlier work there was developed a specification framework that included the individual task specification, which reflects whether the task is a move, a storage, an inspection, a sequencing, or other operation. Here the physical attributes of the load, such as weight, size, fragility, etc., and the task such as vertical displacement, horizontal displacement, positioning accuracy, etc., are important. The main function in this step is to eliminate technologies that are not capable of satisfying the requirements of individual tasks and to match single-task resources with the needs. The earlier work also included fast analysis tools to obtain performance estimates for material handling systems so that system selection can be made on a reasonable basis. These fast analysis tools are tailored to the various topologies for material handling technology. They are at a level appropriate for implementation on a personal computer using personal productivity software.

To complete the four-step specification framework outlined in the earlier work, this research addresses the following steps: 1) task aggregation, and 2) subsystem selection. Task aggregation is performed using cluster analysis. Two general methods were evaluated for this purpose: statistical clustering algorithms and optimization-based algorithms. In post-aggregation analysis, the clusters are re-examined with a view toward splitting and combining clusters, and exchanging tasks among clusters. An intersection algorithm was developed for this purpose. Subsystem selection is performed using a set covering optimization technique. In addition, the fast analysis tools have been augmented to include cost models; the earlier versions only reflected performance. The four-step specification framework is demonstrated using a case study from Camille Motor Works.
 

Contact: gunter.sharp@isye.gatech.edu