A model for more accurate maintenance decisions

Abstract

It is usual when using CM technology for assessing the state of a component and planning maintenance actions using predetermined levels for warnings and replacements. The replacement of a damaged component is usually done at lower or higher than the predetermined level, which both means losses. This is because the probability of doing replacements just at the predetermined level is negligibly small. The accuracy in the assessment of the condition of a component has big technical and economic impact on the output of the machine, production process and consequently company profitability and competitiveness. The higher the accuracy in assessing the condition of a component yields higher probability of avoiding failures and planning maintenance actions at low costs. In this paper, techniques for assessing the state of a component using both mechanistic and other statistical approaches are considered. This paper also applies Cumulative Sum (CUSUM) Chart for identifying the time of damage initiation and reducing false alarms. Techniques for assessing the probability of failure of a component and its residual life, and predicting the vibration level at the next planned measuring opportunity or planned stoppage are introduced, discussed, computerised and tested. The problem addressed is: How is it possible to increase the accuracy of assessing the condition of a component? The major result achieved is; development of a model for more accurate assessment of the condition of a component/equipment through combining different approaches. The main conclusion that can be drawn is; applying the model, it is possible to enhance the accuracy of assessment of the condition of a component/equipment and consequently maintenance decision since the integrated model provides comprehensive and relevant information in one platform.