With the data explosion happening in the past decade, manufacturing enterprises today have more data at their disposal than ever. With technologies like AI & Industrial IoT coming into play, massive amounts of sensor data are processed rapidly. But finding meaningful insights from the data that can affect real value in your daily processes or bottom plan is as difficult as finding a needle in a farm full of hay.
Plant Monitoring/Maintenance has been one aspect that is mission-critical to manufacturers. The cost of machine downtime is huge: according to the International Society of Automation, $647 billion is lost globally each year. Enterprises have been revamping their maintenance processes over the years to alleviate downtime and improve effectiveness. The ability to foresee machine failure & subsequent downtime and prevent them in advance using AI predictive maintenance manufacturing can be transformative to manufacturers in terms of their productivity, OEE & ROI. A McKinsey Global Institute report suggested that manufacturers’ savings from predictive maintenance could globally total between $240 and $630bn by 2025.
While the move from Reactive Maintenance to Planned Maintenance and subsequently to Predictive Maintenance has been paradigm-shifting in terms of being agile & proactive for manufacturers, there is still a lot that it leaves to be desired. With the advent of IoT data being monitored via sensors continuously, It has been the norm to set an alarm/alert when the machine data exceeds a safe threshold, giving the plant maintenance teams an advance notice about when a breakdown might be imminent. However, since these thresholds tend to be static in nature, the flaws in this approach are multi-pronged-
- These are set either too low, flooding your team with false alarms repeatedly, leading to an ‘alert fatigue’ that can cause your team to ignore them. Alternately, if the thresholds are too high, by the time the issue is diagnosed, it is too late to prevent damage or failure to the asset.
- Static thresholds tend to not be very effective when the environment in itself is dynamic, where conditions are constantly changing. Finding a threshold that alerts you early enough to take action without triggering false alarms in such a fluctuating environment has been a big challenge.
- Static thresholds also do not take into account the interdependency of the various systems upon each other that end up influencing the overall performance & health. Especially prominent in the process industry, where the production process is continuous and the output of one system influences all the other systems down the chain, these thresholds need to consider the impact of dependent & auxiliary systems or else leading to a flood of noise alarms.
- Static thresholds fail to include the learnings from the past and can easily get out of date. They have to be analyzed manually & reconfigured, involving significant effort & time by engineers.
So how do you then make your predictive maintenance systems more accurate with lesser alarms & more accuracy? Enter UptimeAI.
With UptimeAI application monitoring your plant, data is continuously gathered, processed & analyzed to record behavior & spot anomaly patterns, triggering alerts after compensating for all the dynamic parameters. It can allow the plant managers to see & analyze the system as a whole, accounting for the external environment, different systems & their interlinking effects & only then triggering an alert. On top of this, it is also capable of learning from new experiences & events, and then apply the learning to a future relevant scenario, thereby evolving like a virtual expert.
So what does this result in? Upto 5X reduction in the number of false alarms, saving the manual effort & time taken to diagnose & mitigate these. With an inferencing engine at work with pre-set 100+ equipment types to diagnose failure modes and suggest mitigation, the onus is not on the operator alone to diagnose a failure in time. In today’s competitive arena, a virtual plant expert on your side can make a wealth of difference, reducing costs & downtime, taking your OEE & performance to the next level.
All in all, 2021 promises to be an interesting year for Manufacturing, as the industry evolves into one that is innovative, agile & responsive both to the needs of consumers and the market, that will take it to newer heights.