It has only been in the last 20 years that plant shutdowns have evolved into a detectable problem because of technological evolutions that have allowed for more accurate predictions. Before that, industry executives had no idea when a shutdown would occur. Once the inevitable had happened, maintenance was only undertaken as a necessity … and those subsequent days of unexpected downtime severely affected the bottom line.
Technological developments within the area of asset performance management have been notoriously slow and regarded as issues that couldn’t be dealt with by most of the industry. However, progress has been made in recent years that has shown real leaps forward in the areas of data scheduling techniques. However, with recent studies showing that more than 85% of shutdowns and failures are completely random in nature, there’s no way to correlate these failures with maintenance dates. The industry now needs to rely on something more robust, away from the calendar, that can drive these maintenance schedules.
To alleviate these issues and begin to address the relationship between maintenance and failures, it is of critical importance that the right working culture is in place. Maintenance teams and departments should work closely with operations to drive reliability from the very foundations, working alongside the capability to predict asset failures earlier and more accurately.
At present, many companies are relying on data scientists to build large numbers of asset models to enable the simulation of failure scenarios. This level of work is unsustainable and fraught with issues; outputs frequently arrive too late and are usually in need of further expert consultancy to predict and interpret the model to provide a correct course of action. These consultants are also in short supply. Thankfully, a solution has arisen in the form of low touch machine learning. This new technology is representative of a breakthrough in automated data collection, cleaning and analysis to provide maintenance protection. This integration is a transition from estimated engineering with statistical models to more measured asset behaviour patterns.
Low touch machine learning works by deploying accurate failure pattern recognition with high precision. This is then used to predict future equipment breakdowns, giving enough notice so that the appropriate avoidance action can be taken. If deployed alongside the appropriate automation protocols, this solution can enable greater flexibility and agility. As these systems increase in agility and adoption, they incorporate the nuances of asset behaviour.
Factoring in ongoing skills shortages, a low touch machine learning approach would eliminate the requirement for substantial resources and expertise to realise the value of the application. This approach to asset management needs process data to achieve more accurate and advance knowledge of asset breakdowns.
Most failures today are directly related to process operations, so early warning systems require more than just condition and maintenance data. Companies have gone as far as they can with condition-based monitoring (CBM), which is incapable of identifying the process-induced conditions causing the bulk of the breakdowns. Predictive maintenance requires looking upstream into process data. Low touch machine learning can deliver comprehensive monitoring of all the mechanical, upstream and downstream process conditions in a far more scalable way than data scientists or CBM. The result is hyper-accurate predictions of production degradation that ultimately leads to asset failure. Such insights are exactly what is needed to get the company’s maintenance and operations teams working together to drive enhanced reliability.
Positive prospects
The world of asset performance management is far from what it was just a few years ago. Previous maintenance practices have been completely reinvented, evolving to recognise all the issues that can affect asset degradation. The integrity of these operations can vastly improve when chemicals and pharma companies implement the correct strategies to detect root failure causes as early as possible. This provides more time for a plan to be deployed to avoid downtime. Low touch machine learning is ready to eliminate catastrophic failures on assets, which will improve overall reliability, avoiding downtime and negative effects on the bottom line.
