Research

year 1997 
author Hosub Shin 
Keyword part type selection, FMS configuration, system performance,design justification 
Abstract The estimated advantages of introducing FMS are improved productivity, reduced work-in-process inventory and lead time, faster response to customer request, etc. Despite these advantages, however, the implementation of FMS has not been popular in Korea. There are two main reasons for such phenomenon. One is the difficulties encountered in operating the installed FMS, and the other is the lack of structured approach that can help top management to decide whether an FMS is a viable alternative to their production environment or not. This thesis deals with the latter subject.

When we plan to introduce an FMS, a preliminary feasibility study should be performed before the actual introduction of FMS. At this stage the basic decision support requirements are selection of part types, determination of FMS configuration and economic justification. Since the selection of part types and determination of FMS configuration are not independent and have some interactions between them, various design alternatives can be generated. And these design alternatives are also evaluated from both the system performance and economic justification perspectives.

In this thesis, a mathematical model is proposed for these basic decision support requirements. The model adopts a closed queueing network model for the performance evaluation of FMS design alternatives and then the well-known payback period method is reflected as an economic justification criterion. Since the resulting model is a nonlinear integer programming model, a structured solution procedure is presented and implemented. For the concurrent evaluation of system performance and economic justification, the proposed FMS design justification system consists of three modules: performance evaluation module, configuration search module and part type selection module.

An analytical model for the performance evaluation of an FMS is implemented in performance evaluation module, where starving delay of a station with limited local buffer in FMS occurs when the next workpiece to be processed is not transported to the station that as just finished working. When a starving delay occurs a possible extra is considered through a revised process plan of part types to be processed. In this revised process plan, we can include central buffer storage as one of the stations of classical closed queueing network model. Now, we can analyze the overload of transportation vehicle imposed by extra trips between central buffer storage and stations, which was not accounted for explicitly in the traditional closed queueing network model. The duration of extra trip and the speed of transportation vehicle that influence the unwilling starving time of each station are analytically approximated and reflected. The simulation study shows that the proposed model makes the performance evaluation of FMS more realistic than classical models can do.

For the efficient search of the solution space which spans the space of candidate part types and FMS configurations, the proposed solution procedure divides the entire solution space according to each FMS configuration. In configuration search module, two different search strategies are implemented. The one is for expanding processing capacities and the other is for modeling. For the capacity expanding strategy, this thesis shows that we don´t have to consider all the feasible FMS configurations and only expanding the capacity of bottleneck station is enough. When modeling an FMS using classical queueing network model, the processing equipment of the same resource type is generally modeled as a multi-server single station. Though this modeling approach can give a good system performance, it is not always true in view of economic perspective. Thus, in the modeling strategy, a multi-server processing station is divided into individual single-server stations and a heuristic is proposed to balance workload between them for the improvement of system performance. For the same part types to be processed in FMS, the modeling strategy can generate more cost-effective FMS configuration than the traditional modeling approaches.

Part type selection module is for the exploration of the solution space consisting various candidate part types that can be processed in a given FMS configuration. For a given combination of FMS configuration and candidate part types, the total benefit/cost factors are analyzed from an economic justification perspective and system throughput is examined from a system performance perspective. A greedy heuristic is proposed for the selection of part types to be processed in FMS by adding a part type considering these economic factors and system performance.

This thesis proposed a design justification approach for a series of decision support problems that can be raised at the initial design stage of FMS and show that the proposed solution procedure can be applied to a real world problem through a case study. 
c PhD 

Download :