We care about your data, and we'd use cookies only to improve your experience. By using this website, you accept out Cookie Policy.
It has been a long journey of predictive maintenance since it was introduced to the industry. But moving toward predictive maintenance might be a painful effort for some maintenance teams because it comes with higher cost and labor-intensive work (during the implementation). Most of the time preventive maintenance is considered to be “just enough” for those equipment lines up in the plant. However, critical equipment requires special care, and it is essential to prevent it from being unplanned shutdown. Condition-based maintenance (i.e. vibration assessment) is introduced in conjunction with preventive maintenance to detect an early failure of machine. Unfortunately, this practice consume man-hour and it does capture only mechanical characteristic not the operating parameters; meaning that an abnormal sign (of mechanical component) will be detected when the component failure is starting to occur and/or propagate but cannot directly identify what operating parameters leading to the failure without having a detail analysis.
The predictive maintenance strategy takes this role helping user to detect suspicious behavior of the equipment and help gather the necessary parameters to identify what corrective actions are needed to bring this equipment back to its “normal” condition. It can also help maintenance teams to plan a repair/overhaul job ahead of the predicted failure. This allows them to find a long lead spare part which in some cases might take up to 12 months to be delivered for the more severe case. Still, this comes with a drawback since the abnormal event might randomly appear and detect so it can interrupt a lot of planned schedules if those corrective tasks are not well organized.
There are several advantages of predictive maintenance over others approaches. A modern technique is utilizing AI/Machine learning for data analytic and machine behavior recognition to spot the anomaly. Here are some key benefits.
Reduce the number of unplanned downtimes – with preventive maintenance task only, it might not enough to cover all failure prevention because some failures can happen right before or after the PM task. Some failures can be induced by external conditions or begun to develop after the PM. This kind of hidden failure can be captured by continuously monitoring and predicting the trend of operating conditions with an appropriate number of sensors. With machine learning analytics, it can pinpoint abnormal behavior and identify the potential problem which may occur so that maintenance team can prepare for the corrective action at the right time.
Save maintenance cost and production loss – Spotting the hidden failure can help maintenance team to tackle and rectify the problem before it propagates to the more catastrophic case. Manhour for the task can be less as well as the spare part cost can be reduced. For the critical equipment, catastrophic failure shall be minimized and prevented because it leads to the loss of production opportunities (LPO). Altering from unplanned downtime to planned downtime make a lot different in term of production loss, because all the critical spare part can be ensured to be available (lead time of some parts can take longer time to deliver) and the manpower and the outage can be scheduled to handle the task effectively.
Extend asset lifespan – Predictive maintenance practice provides ability to assess the equipment condition continuously. It helps improve equipment reliability and also efficiency with a proper and appropriate maintenance task throughout its service life. With equipment running at higher reliability and efficiency, it gains better machine health in the long run. Some equipment has a specified target TBO (time between overhaul) advised by OEM, however, at different running conditions either low load or higher efficiency point, the equipment can experience less degradation than OEM expecting trend. In this case, condition assessment offers a useful set of information for the user to evaluate whether this operating condition is acceptable and can be prolonged. In some predictive maintenance platforms, the degradation trend and the predicted asset health have been analyzed and visualized to support the user’s decision in terms of service life extension.
Initial investment – An implementation of predictive maintenance can be expensive since it requires data integration to existing systems for all relevant sensors. In some cases, additional sensors are required to be installed to support condition monitoring to assess equipment health.
Compatibility – It will require a lot more work for the older plants which have only sensing gauges/indicators available on site since predictive maintenance technique utilizes the real-time (or almost real-time) data to analyze and predict. New transmitters installation, sensors upgrade, and control retrofit are needed to support the project.
Data Integration – Integration into the existing operating system is one of the big challenges. Some predictive maintenance platforms are not compatible with the system which will require an additional cost for licensing and or data integration. Many platforms require data historian for data acquisition and occasionally the user does not have data historian in place so user will be asked to upgrade the local data server. This also raises concerns about data security.
Staff Requirement - Staffs require a comprehensive training for software/platform operation and data analytic (if required). Additional skills are required when the data training for the model is needed to validate the model improvement. Specialized staff are needed to interpret prediction models and give recommendations for corrective action needed. Some vendors offer data scientists, technical staff or service engineers to validate models, monitor and give recommendations for users which come with extra subscription cost.
Predictive maintenance solution offers a proactive approach by utilizing data and analytic models to detect potential equipment failures and signs of deterioration or abnormal behavior. Overall, predictive maintenance is a powerful tool to optimize maintenance activities, improve efficiency, and enhance asset performance. Implementing predictive maintenance approach comes with additional costs that require solid business justification. Comparing to the benefits of predictive maintenance e.g. reduce number of unplanned downtimes, minimize process disruption, save production loss, increase equipment reliability, etc. implementation of predictive maintenance can become cost effective for the equipment with higher criticality ranking where their unplanned downtime has direct impact to the production. A hybrid approach combining preventive, condition-based and predictive maintenance might sometimes be used for the more critical equipment to cover benefits in all aspects.
User will have to apply different maintenance strategies based on equipment criticality or based on the worst performers in a plant whether it shall apply a run-to-fail practice, preventive maintenance, condition-based monitoring, predictive maintenance or even a hybrid approach.
