| چکیده انگلیسی مقاله |
Purpose: Determining preventive maintenance intervals is critically significant from a variety of aspects, such as cost, reliability, and downtime. Currently, in many industries, such intervals are determined based on the viewpoints of experts and technicians or the manufacturer's recommendations in the manual. Determining an optimal interval, by the means of scientific methods, can have many effects in terms of costs, reliability, and downtime. For example, if the interval is assumed as very short, it can increase the cost of spare parts or manpower. Also, if the interval is too long, it increases the possibility of machine failure and downtime. For this purpose, organizations that apply the approach of time-based preventive maintenance should consider the above-mentioned criteria. However, such criteria are not compatible with each other, and each of them has its behavior. Therefore, given the existence of more than one criterion in this decision-making problem and the incompatibility of the criteria, Multi-Criteria Decision-Making (MCDM) methods seem appropriate for this purpose. Design/methodology/approach: A component of a machine that is currently replaced periodically (and not necessarily at equal intervals) is considered for investigation. It is assumed that the failure distribution function of the component follows the Weibull distribution. First, using the Maximum Likelihood Estimation (MLE) technique, new and more general equations are presented to estimate the parameters of the failure distribution function. Subsequently, three important criteria are taken into account, including reliability, cost, and downtime due to failure and preventive replacement.. Given the existence of several criteria with different behaviors, as well as the vague or linguistic nature of human judgments, a hybrid MCDM model including Fuzzy Analytic Hierarchy Process (FAHP) and Viekriterijumsko kompromisno rangiranje (VIKOR) is proposed. Findings: To indicate the applicability of the proposed model, a real case was investigated in the Isfahan Steel Company. For this purpose, a circuit breaker switch of a compressor in the oxygen workshop was studied. The component was replaced at about a 200 days interval. After implementing the FAHP method, it was found that the reliability criterion was more important than the other two criteria. This is due to the high sensitivity of this component and the consequences of its failure. After Performing the VIKOR method, the set of top alternatives included replacement intervals between 50 and 90 days. It confirms that the importance of the reliability criterion, and the replacement interval, should be reduced by at least a half. Based on the results, reliability was expected to improve by 2.6% and the cost and downtime criteria to improve by 33.7% and 28.6%, respectively. Research limitations/implications: To implement the proposed method for each component, the following data are required: Preventive maintenance and breakdown records of the component Cost and downtime of the machine due to breakdown and preventive maintenance Three pairwise comparisons to implement the FAHP method If information systems exist in the relevant plant, the data for items 1 and 2 are available and do not constitute a constraint on the proposed model. The only limitation of this model is the inputs of the FAHP method for which, implementing the model depends on the humans. Doing three pairwise comparisons for each component may seem a bit difficult and time-consuming. Consequently, the removal of this limitation can be proposed as a suggestion for the development of this study. Other calculations of this model can be done automatically. Also, the study of other probability distribution functions such as gamma and lognormal can be effective in improving the results of this study. Originality/value: To the best of our knowledge, the fuzzy AHP or VIKOR method has not been used to address the subject of this study, even independently. In addition, unlike previous studies, the proposed model was implemented for a component that was undergoing preventive maintenance. In other words, in most of the data collected the component was replaced before failure, which made it impossible to: Discover the component failure distribution function by data fitting. As a result, it can be assumed that the failure distribution follows the Weibull distribution. Applying the existing equations to estimate the parameters of the failure distribution function. As a result, new equations can be proposed by implementing the MLE technique. |