Date on Master's Thesis/Doctoral Dissertation

12-2008

Document Type

Master's Thesis

Degree Name

M.S.

Department

Industrial Engineering

Committee Chair

Heragu, Sunderesh S.

Subject

Materials management--United States--Data processing; Production control--Data processing; Inventory control--Data processing

Abstract

Material Requirements Planning (MRP) is one of the earliest production scheduling approaches that utilizes computers. MRP is still regarded as one of the most widely used systems for production scheduling. Even though MRP has made contributions, there are some fundamental problems (i.e. the assumption of infinite capacity and fixed lead times) which make the MRP system vulnerable to effects of uncertainty. To overcome this fundamental flaw, there was a trend towards the development of detailed finite-capacity scheduling systems (i.e. MRP II, ERP, and APO). All these MRP-based systems still ignore variability and randomness and are inherently push systems. Instead of creating a detailed schedule based on forecast, Factory Physics Inc. developed Dynamic Risk-Based Scheduling (DRS), which creates a set of policy parameters (e.g. WIP level, lot sizes, reorder point, and reorder quantity) that work for a range of situations to calculate the production schedule. This thesis compares the key performance measures of DRS and MRP-based scheduling systems. We begin with a single-machine problem and develop simulation models for varying levels of uncertainty in forecast demand (i.e. base demand scenario, under-estimated scenario and over-estimated scenario) and two levels of variability in the system (i.e. moderate variability and no variability). Then the experiment is extended to multiple-machine problems. We also introduce more constraints into the DRS and MRP models to improve their performance. We also test the performance of MRP models for different planning horizons. We find that the DRS strategy is more robust to forecast error than MRP-based strategies. DRS also usually obtains better performance than MRP-based models in terms of higher fill rate and lower inventory.

Share

COinS