A comparative model of facility network design methodologiesPowell Robinson, E.; Swink, Morgan L.
doi: 10.1016/0272-6963(95)00027-Ppmid: N/A
Operations managers use a variety of decision making tools when designing facility networks. Intuition, scenario evaluation, heuristic and optimization procedures are commonly applied. This paper discusses the relative advantages and disadvantages of the alternative methodologies for network design and proposes a cost trade‐off model for choosing the best approach. Accessibility, precision and supplemental analysis are key factors to consider during the selection process. We also present the results of a laboratory experiment which measured the performance of analysts using intuition, scenario evaluation and enhanced scenario evaluation procedures to solve realistically‐sized network design problems. The test problems included several sizes of single‐echelon, multi‐activity and two‐echelon facility network design problems. The experimental results, when embedded into the cost trade‐off model, provide quantitative guidelines for selecting the appropriate decision support procedures. Implications for practitioners, researchers and decision support system developers are provided.
Alternative approaches for reducing schedule instability in multistage manufacturing under demand uncertaintyKadipasaoglu, Sukran N.; Sridharan, V.
doi: 10.1016/0272-6963(95)00023-Lpmid: N/A
Schedule nervousness caused by uncertainty in demand or supply or by dynamic lot‐sizing can be an obstacle to effective execution of material requirements planning (MRP) systems. Previous studies on the effectiveness of alternative approaches for dampening nervousness almost always considered single‐level product structures and deterministic demand conditions. This study evaluates the effectiveness of three strategies for reducing nervousness in multi‐level MRP systems under demand uncertainty: (1) freezing, (2) end‐item safety stock, and (3) lot‐for‐lot scheduling below level 0. The strategies are evaluated under a wide range of operating conditions characterized by variations in item cost structure, product structure, demand uncertainty, and lot‐sizing methods used. A multi‐level MRP system is simulated to study the relative performance of the three strategies. Results indicate that freezing the master production schedule is the most effective approach in terms of reducing both instability and cost as the freeze length covers the cumulative lead time. It is also the least sensitive to changes in operating conditions.
Evaluation of scheduling rules with commensurate customer priorities in job shopsJensen, John B.; Philipoom, Patrick R.; Malhotra, Manoj K.
doi: 10.1016/0272-6963(95)00028-Qpmid: N/A
Much of the existing job shop scheduling literature has focussed on scheduling rules and heuristics that typically do not incorporate customer specific characteristics into their scheduling decisions. This paper specifically considers the situation in which depending upon the identity of the customer, different jobs have different tardiness penalties associated with them. In addition, the tardiness penalties associated with different customers are comparable in magnitude with one another. We propose here and test four different types of scheduling rules which utilize the jobs' tardiness penalty information, and show that the performance of the shop is only mildly affected by the shape and dispersion of the weighting scheme used in determining the tardiness penalties. The performance of scheduling rules is however contingent on due‐date tightness. In general, apart from the traditionally well performing shortest processing time rule, the ATC rule works well on the mean flow time measure in loose due‐date situations, while COVERT performs well on mean flow time under tight due‐date conditions. Lesser differences are seen to exist between mean tardiness and root mean square tardiness (TRMS) performance of different scheduling rules. Weighted mean tardiness deteriorates with increasing due‐date tightness and coefficient of variation of job tardiness penalties. Finally, customer service performance worsens considerably with weighted dispatching rules when the performance criteria are not weighted by each job's tardiness penalty. Implications of our results for providing more effective management of job shops, and exploring future research directions in this area are also presented and discussed.
Optimal and heuristic policies for lot sizing with learning in setupsRachamadugu, Ram; Schriber, Thomas J.
doi: 10.1016/0272-6963(95)00024-Mpmid: N/A
We develop heuristic and optimal methods for determining lot sizes when setup cost reductions occur over time due to emphasis on continuous improvement, learning effects and incremental process changes. Our heuristic methods are intuitively appealing, easy to implement, require little information about setup costs, and have low computational burden. Computational studies show that choosing the appropriate heuristic yields nearly optimal solutions. Recommendations for choosing the appropriate heuristic are also provided. The optimal method developed in this paper is also useful to managers for evaluating investments in hardware and/or worker training for setup reduction. Concepts and methods are illustrated with numerical examples. Managerial implications of using our policies are discussed.