Doctoral Dissertations
Date of Award
3-1985
Degree Type
Dissertation
Degree Name
Doctor of Philosophy
Major
Engineering Science
Major Professor
William W. Claycombe
Committee Members
Eldon L. Deporter, Kenneth E. Kirby, Oscar S. Fowler, William T. Snyder, John M. Bailey
Abstract
The primary objective of this research was to develop a heuristic MRP model with emphasis on the multi-objective MRS model and capacity-constrained CRP model. The multi-objective MRS model is a unique approach in resource-constrained MRP system in terms of optimizing conflicting objectives in planning stage. The capacity-constrained CRP model is also a unique approach in capacity requirements planning in capacity-oriented MRP system in terms of finite loading over multiple work centers with three types of heuristics for loading process.
A multi-objective MRS model has been developed utilizing lexicographic, multiphase linear goal programming with generalized manufacturing objectives and constraints. The conflicting manufacturing goals and their constraints were itemized by three functions of manufacturing; production, marketing and finance. A generalized discrete, multiproduct job shop problem has been mathematically formulated over planning horizon. Two models were developed with the variations in decision variables and constraints.
Two FORTRAN computer programs were developed for the multi-objective MRS model: MPSINP and GOALPM. The MRS formulation program, MPSINP, generates goal programming input file based on the given manufacturing objectives, constraints, and model options. The second program, GOALPM, is extensively modified version of an existing goal programming computer code with the addition of a subroutine for parametric analysis.
The optimized MPS solution can be obtained through serial executions of two programs. The computational experiments showed that the solutions provided optimized prospective master production schedules based on the decision maker's priority structure and constraints. The additional parametric analysis confirmed that it can be used as a vehicle to search for a most feasible solution with possible trade-off among the objectives. The level of computer resources requirements for both program execution was very reasonable with minimal effort in data preparations.
The capacity-constrained CRP model has been developed to implement the result of multi-objective MPS with detailed production planning. This model assumed that the end-items in MPS compete for a group of work center. The load profiles were used for finite loading process with three different shifting methodologies: backward shifting, alternate backward/forward shifting, and partitioning back ward shifting.
A computer code for capacity-constrained CRP model, CRP3M, has been developed including three shifting algorithms as options for loading procedure. The program converts initial production planning to required capacity of each work center, then searches for the avail able capacity over planning periods. If the required capacity is not found on desired planning period, the program performs shifting algorithm searching for the available capacity within maximum shifting limit.
The computational experiments were performed based on three types of initial production planning which were established from disaggregation of a prospective master production schedule by given fixed batch size. Three types of measure of performances were adopted to compare these results: amount of required overloading, utilization rate, and extended lead time.
The capacity-constrained CRP model assumed flexibility in capacity adjustments through the searching for the optimum shifting of minimized overloading. This searching process made it possible that three shifting algorithms improve the initial loading, or infinite loading, with less overloading requirements and higher utilizations rate.
The results indicated that three shifting algorithms significantly improved initial loading schedule compared to the infinite loading through reduction of overloading requirements and higher level of utilization rate. If the initial loading had more fluctuated pattern, the shifting algorithm can achieve more stabilizations in capacity planning. The computer time required for the CRP3M program was less than one minute, and input preparation requires very minimal effort. The multi-objective MPS model and the capacity-constrained CRP model can be easily modified to accomodate any specific manufacturing planning with flexibility and applicability.
Recommended Citation
Kim, Jason Jungsun, "A Heuristic model for multiobjective resource planning and capacity-constrained scheduling : in material requirements planning systems. " PhD diss., University of Tennessee, 1985.
https://trace.tennessee.edu/utk_graddiss/12576