Dynamic resource allocation in flow lines with station-level flexible operation sequences and alternative resources

In this thesis, the impact of station selection rules for resource allocation, resource selection rules among alternative resources for stations, and number of resource duplicates available in a manufacturing flow line is investigated in terms of their effect on system throughput, manufacturing lead time, resource utilization levels, and process wait time for resources.

A hypothetical flow line with ten stations is modeled using the Arena TM Simulation package and experimental results are analyzed via the Analysis of Variance (ANOVA) using the SAS® statistical analysis software.

Similar to typical maintenance, repair, and overhaul (MRO) type of manufacturing flow lines, each station in the simulated model consists of operation sequence alternatives, as well as resource alternatives for each operation. A resource represents certain technical capability required by an operation; it can be an operator or a set of tools or both. Such flow lines typically run under limited number of resources in addition to lack of real-time visibility in terms of completion time of operations and actual whereabouts of resources. Therefore, utilizing the resources efficiently while achieving the highest possible throughput for a flow line is a challenge.

This thesis presents a statistical analysis of various decision factors, their main effects and interaction effects on the performance of such flow lines by facilitating Dynamic Resource Allocation (DRA) and real-time resource sharing.

The effect of dynamic resource allocation is investigated by simulating a combination of three station selection rules (Last Station First, Remaining Work Load, and Shortest Processing Time) for resource allocation, three resource selection rules (Largest Quantity Available, Least Utilized Available, and Preferred Alternative Available) among alternative resources for stations, and a variety of quantity (2, 3, 4, 6, 8, and 10) of resource duplicates available in the flow line.

This study first shows that DRA greatly enhances system performance when compared against a typical flow line operated without DRA. Second, the station selection rules and varying quantity of duplicate resources have the greatest effect on system throughput, and manufacturing lead time. While, the station selection rules, resource selection rules and varying quantity of duplicate resources have the greatest effect on system throughput, manufacturing lead time, process wait time for resources and the resource utilization rate. Third, the proposed approach can be used as a system design tool to determine the number of resources necessary to maximize the system throughput without wasting resources. Forth, this study also demonstrates via simulation how DRA can be facilitated using Radio Frequency Identification (RFID) technology, which is used as an enabler to facilitate a look-ahead approach for dynamic resource allocation and provides additional benefits in terms of all performance measures.