What if you could stop unplanned downtime altogether? Plant operators and engineers could actually fall out of their chairs. why? Unplanned downtime causes significant production loss, reduces profits, and causes frustration and headaches for everyone involved.
Unfortunately, not all equipment failures are related to aging, so time-based maintenance strategies do not work in all scenarios. It is estimated that 82% of failures exhibit random patterns, and these random patterns are difficult to predict and prevent.
This is where predictive maintenance (PdM) comes in. Predictive maintenance using machine learning (ML) is not limited to looking only at aging-related patterns or predefined threshold limits. Predictive maintenance using machine learning uses a risk and severity approach to detect anomalies and predict failures.
Machine learning, a subset of artificial intelligence (AI), continuously learns and studies patterns in data. Used for predictive maintenance, it identifies deviations early to determine if they will lead to future outages or incidents. When an outage is predicted, the maintenance team is alerted and can schedule and plan maintenance work at a convenient time, avoiding unplanned downtime.
Predictive maintenance takes traditional maintenance techniques such as preventive and reactive maintenance a step further. Preventive maintenance involves scheduling maintenance work on a regular basis, regardless of the condition of the equipment. Corrective maintenance focuses on repairing equipment after a failure. Both approaches can be costly and time consuming. PdM, on the other hand, is designed to optimize maintenance efforts by focusing on the specific needs of each piece of equipment.
did you know?
Predictive maintenance is very cost-effective, “saving about 12% over preventive maintenance and up to 40% over reactive maintenance.”
Source: Webinar Care
Predictive maintenance, especially when combined with machine learning, can be used across a wide range of equipment and is especially useful for rotating equipment that continuously generates data such as pumps, turbogenerators, gas compressors, industrial fans and rotary kilns.
Unlike other predictive maintenance techniques that analyze data from a single piece of equipment or isolated processes, PdM techniques that utilize machine learning models analyze data from the entire facility.
Applying machine learning to operational data across facilities can identify bottlenecks caused by seemingly unrelated processes.
Predictive maintenance includes several steps such as data collection, analysis and action. Here’s a step-by-step overview of the process:
1. Data collection:
The first step is to collect historical and real-time data from sensors and other sources such as distributed control systems (DCS), programmable logic controllers (PLC), and supervisory control and data acquisition (SCADA). This data may include information about equipment usage, temperature, vibration, etc. The data is then stored in an enterprise data warehouse or process data historian for analysis.
2. ML modeling:
A machine learning model is then trained and the data is analyzed. Models are trained on past failures and can be built to detect patterns in the data that indicate potential equipment failures. For example, if a machine has increased vibration levels over time, this could be a sign of impending failure.
3. Equipment diagnostics:
ML models can provide factor and root cause analysis to support decision making. Once a potential equipment failure is detected, the maintenance team can use diagnostic tools to determine the root cause of the problem. This may involve taking the equipment apart to inspect individual components or using non-destructive testing techniques.
Four. Maintenance plan:
Based on diagnostic results, maintenance teams can plan and schedule maintenance activities. This includes repairing or replacing damaged components, performing cleaning and lubricating operations, or adjusting equipment settings.
Five. Continuous monitoring:
PdM is an ongoing process and requires continuous monitoring of equipment performance. ML models can run continuously, learning as operational changes are made, monitoring and predicting future outcomes. Maintenance teams can use these insights to identify potential problems early and take action to prevent them from becoming serious problems.
Below are a few examples of companies using ML PdM to avoid unplanned downtime and improve equipment reliability.
Avoid gas leaks: A major oil and gas company used ML PdM to prevent loss of primary containment (LOPC) events that could lead to severe environmental incidents. The model detected that the problem was caused by another process, the gas well. The company was able to intervene quickly and take steps to prevent platform outages and environmental disasters.
Ventilation fan: A mining company received a call that it had a problem with its exhaust fan, resulting in moderate to severe drop in outlet pressure. This prediction allowed the engineering team to plan repairs and confirm the exact root cause predicted by the ML model. Planned activities resulted in minimal downtime and avoidance of $700,000 in costs that would have been incurred if the failure had not been discovered.
Compressor Reliability: An oil and gas company had recurring compressor reliability issues, resulting in unplanned downtime and significant costs. ML PdM was used to validate the reliability engineer’s conclusions and identify additional factors. Using the insights provided, the company was able to extend gas compressor uptime to $21.7 million in less than four months.
Predictive maintenance using machine learning provides a solution to random patterns of equipment failures that traditional maintenance strategies cannot address. By continuously learning and analyzing patterns in data, machine learning models can detect anomalies and predict potential failures, allowing maintenance teams to schedule maintenance work before unplanned downtime occurs. and can be planned.
PdM is an ongoing process that requires continuous monitoring of equipment performance and is applicable to a wide range of equipment. Companies that have successfully implemented PdM have been able to avoid environmental incidents, save costs, and increase equipment reliability. Overall, his PdM with ML is a game-changer for plant operators and engineers in terms of preventing unplanned downtime and increasing profitability.
