“Don’t try to fix, which is not broken” is a famous saying. Production machines come with a lot of prescriptive maintenance schedules like changing belts, oils, and controllers at a fixed interval or after they run for a specific duration. The standard prescriptive/preventive maintenances are proper but not the best; reason being - the maintenance involves downtime in production and cost of replacing/replenishing parts unnecessarily, which can run a few more miles. Thus Predicting the right time for maintenance increases the resource utilization and reduction in repair frequency.
Part of quality standards there could be many maintenance calendars like 52 PPM, checklists, etc., to avoid any unplanned downtime or reactive maintenances. As mentioned, proactive or prescriptive maintenances involve scheduled but unnecessary downtime – Industry 4.0 focusses on data-driven based Predictive maintenances with the analytics on data points from machinery.
But, there will be tonnes of data points received, which need to be fault detection and classification with an analysis for possible failure events, the predicted time of happening, etc., require AI, and ML trained Big data engines. Predictive maintenance is too attractive but requires experienced SMEs to implement due to
• Historic and Failure data availability
• Production uncertainties
• Data Pre-processing and Quality
• Domain expertise and life of machines
• Framing the problem correctly
• Prediction models evaluation
• Flagging abnormal behaviors
• Deriving asset value for varying condition and age
Predictive maintenance improves the equipment’s life span, reduces unscheduled, and scheduled assembly line downtime, improves productivity.
Thus Predictive Maintenance is a crucial Business Driver for all Manufacturing plants!!
Read the case study – Predictive Maintenance