Thesis (M.S.)--The University of Texas at San Antonio, 2009
In this thesis, a dynamic resource allocation (DRA) methodology for manufacturing flow lines is developed, and its impact on system throughput is studied via simulation. The proposed methodology uses the Statistical Throughput Control (STC) concept (Hopp and Spearman, 1989) to monitor production levels in real-time and facilitate resource sharing among processes by dynamically allocating them in response to variations in production levels due to breakdowns and process time fluctuations. The methodology uses STC's alpha criterion to monitor the probability of missing the desired production level, i.e. quota
A hypothetical flow line with three productions areas, each with a limited number of resources and buffers in between, is modeled using the Arena(TM) Simulation package, and experimental results are analyzed via Analysis of Variance (ANOVA) using the SASRTM Statistical Analysis Software. Two of the three production areas are arranged in parallel and feed into the third area; each production area includes three parallel processes. Resource sharing is facilitated via STC-based dynamic resource allocation within production areas among the three parallel processes in each area. Control factors are confidence intervals, similar to control limits in statistical process control, number of resources exchanged (DeltaR), and time frequency of production level checks (Deltat) with each factor having three levels. Performance measures are system throughput, overage (deviation from desired production level), and resource allocation frequency
Twenty-seven scenarios of all combinations of factors/levels are simulated using the STC-based DRA methodology to investigate the effect of each level on all three performance measures. In addition, a fixed resource allocation (FRA) model that uses the same layout but does not allow for resource sharing is simulated as a reference. Overall, the proposed DRA method outperforms the FRA model. ANOVA results for DRA show that the number of resources exchanged (DeltaR) and the time frequency of production level checks (Deltat) are both significant in terms of throughput, overage, and resource allocation frequency. However, the confidence intervals have no statistical significant impact on throughput or overage but have a significant effect on resource allocation frequency. Overall, DRA facilitates resource sharing as a real-time response to variations in production levels and increases system efficiency in the presence of disturbances
