46 September 2011 Vol. 23 No. 3 Engineering Management Journal A Budget-Sensitive Approach to Scheduling Maintenance in a Total Productive Maintenance (TPM) Program Abhijit Gosavi, Missouri University of Science and Technology Susan L. Murray, Missouri University of Science and Technology V. Manojramam Tirumalasetty, Ungerboeck Systems International Shreerang Shewade, Analytical Mechanics Associates Unexpected failures can reduce throughput—especially if the failure afects the process bottleneck. Furthermore, when a failure occurs, it usually takes longer to correct than a scheduled maintenance activity would, resulting in signiicantly higher costs. It has been empirically shown that preventive maintenance can reduce the frequency of unexpected failures and, if done at appropriate time intervals, can reduce the overall costs (Askin and Goldberg, 2002). TPM is now viewed as an integral part of regular operations in production irms. With the advent of computers and the cheap availability of personal computers in the last couple of decades, computerized maintenance systems have become increasingly popular in industry (Westerkamp, 2006). Such systems make it easy to collect and maintain historical data of machine failures, their frequencies, and down-times due to repairs and maintenance. hese databases can be used to determine parameters (such as distribution type, mean, etc.) of system failures, providing an excellent basis to model and improve the maintenance process. It is no exaggeration to state that production systems cannot remain healthy and productive without a good TPM program; however, developing efective operational strategies for TPM can be quite challenging because of numerous complicating factors, such as random failures of the diferent machines and pieces of equipment in a system, randomness in repair/maintenance times due to variability in the availability of spare parts and repairpersons, and the complex stochastic dynamics of production systems. he manager has to analyze the underlying stochastic processes, costs, and revenues, and a host of other factors in order to develop an efective TPM program. Problem Statement Production managers must balance the need to reduce lost production costs due to equipment failures with preventive maintenance costs. Managers do this by developing preventative maintenance schedules. Most statistical models in reliability and preventive maintenance (PM) attempt to optimize the system over the long-run (i.e., an ininite time horizon) to minimize long-run and average costs. While this leads to good long-run performance, in a inite time horizon such a scheduling policy can exceed short- run maintenance department budgets. Our goal in this research is to develop maintenance schedules that minimize the long-run average costs over the ininite time horizon, but at the same time minimize the chances of costs exceeding a predetermined daily or weekly budget. To accomplish this we employ a relatively less known measure of risk called semi-variance that measures variability of costs rising above a predetermined threshold (target or ceiling). he advantage of this metric is that when suitably combined with the long-run average cost metric, it can produce solutions that seek to keep the long-run average cost under Refereed Research Manuscript. Accepted by Editor Doolen. Abstract: Scheduling planned maintenance activities is key to the success of Total Productive Maintenance (TPM) in reducing the mean and variability of production lead time. Most existing maintenance-scheduling models are risk-neutral, striving to control long-run costs. Some are variance-penalizing, addressing both average cost and cost variance. Neither addresses budget constraints. We present two models that use semi-variance combined with the mean to simultaneously optimize maintenance with respect to long-run costs and short-term budgets. he irst model, geared for individual pieces of equipment (e.g., a pump or dryer), uses renewal theory. he second model presented is based on Markov decision processes and is appropriate for manufacturing systems composed of several units, any one of which can fail. An application of each model is presented. Beneicial operational costs variance is not penalized in this approach, which is more appropriate. Keywords: Preventative Maintenance Scheduling, Semi- variance, Budget-based Scheduling EMJ Focus Areas: Quantitative Methods and Models T otal Productive Maintenance (TPM) is a program that began in the 1970s in Japan. Seiichi Nakajima popularized TPM throughout Japan and is widely recognized as a pioneering practitioner of this ield. One of the main goals of TPM is to maximize equipment efectiveness and availability. In the 1980s, TPM started gaining popularity in the United States. he initial interest may have been due to competition from Japan and the need for cutting costs. It quickly became clear that a well- designed TPM program would lead to less lead time variability, which in turn, signiicantly lessened the need to carry inished goods inventory. Since the late 1990s, many manufacturers have transitioned from make-to-stock (push) to either make-to-order (pull) or delayed diferentiation strategies. To be competitive, make-to-order requires signiicantly reduced lead time and is less tolerant of unexpected machine failures disrupting production. Consequently, the need for TPM has increased in recent years. TPM is no longer viewed just as a continuous improvement tool to cut costs but is also considered a critical tool in keeping lead time in check, which is absolutely essential for survival in a world where customers demand low prices and rapid delivery. TPM is otentimes implemented in multiple phases, the irst phase being linked to the philosophy of being pro-active in order to improve equipment performance. A modiication of a popular slogan in industry captures the TPM philosophy: “If it ain’t broke, ix it anyway!” TPM’s overall goal is to prevent the unexpected failure that can disrupt a production schedule.