5 Maintenance techniques
Selecting the best approach to sustaining a particular asset has been quite a challenge for the asset owner. The choice of the best maintenance strategy is a selection of the most suitable maintenance techniques for the policy to be operationalized. Maintenance techniques (MT) in this context are the methods used to forecast and predict the remaining useful life of an asset. MT enables the application of maintenance policies. The available data, the possibilities for data collection, and the required outcome determine what MT to select. Based on the classification, five different forms of MTs can be distinguished below.
I. Experience-based: In the experience-based technique, forecasts of failure times are based on knowledge and experience outside or inside the organization (e.g., OEM). Occasionally, little or sparse data supports them. Predictions are based on expert judgment (for example, facilitated by FMECA techniques). Such methods estimate the life span of an average variable operating under average historical conditions.
II. Reliability statistics: The prediction techniques for reliability statistics are based on historical (failure) records of comparable equipment without regard for specific component (use) variation. This method explains precisely the likelihood of population-wide failure. Such techniques also estimate the lifespan of an average item that operates under average conditions, for example, the distributions of Weibull.
III. Stressor-based: Stressor-based predictions are based on historical records with stressor data, for example, temperature, moisture, or speed, including environmental and operational variances, and produce predicted system life expectancy results in a specific environment. Pronouncements are based on a general direction derived from a physical model, built-in performance, or operating history.
IV. Degradation-based: Degradation-based prognosis is based on the extrapolation of a general path to a failure threshold from a forecast parameter, a degradation measure. The system can be diagnosed by measuring symptoms of initial failure, such as an increase in temperature or vibration. The prognostic parameter is also calculated by sensor measurements, i.e., often dependent on measurement. The forecast starts with the current deterioration situation and results in a lifetime required of a particular system in a certain environment.
V. Model-based: model-based projections give the predicted remainder of the lifetime of a certain system. Two distinct categories of models exist:
a. Physical model-based: The prognostic parameter is computed based on a degradation mechanism physical model based on direct load-sensing or usage, which governs individual components’ critical failure mechanisms.
b. Data model-based: Data analytics approach that uses sensed load variations, utilization of process data, or condition/health monitoring data as input to measure or extract the prognostic parameter. The algorithms are designed to extract or try to predict anomalies by comparing them with historical data. The aim is to generate value-adding preventive and corrective maintenance work orders to keep the equipment in basic condition.
#MaintenanceTechniques #PreventiveMaintenance #IndustrialMaintenance #MaintenanceBestPractices #AssetManagement #OperationalExcellence #MaintenanceStrategies #Efficiency #Reliability #MaintenanceOptimization #IndustrialEngineering #MaintenanceManagement #MaintenanceTips #OperationalEfficiency
