SVP, Head of Energy Industries
In today’s industrial landscape, the scale and momentum of digitalization is compelling companies to reinvent themselves and strive for continuous improvement in their operations. Industry 4.0 is revolutionizing production and distribution processes by integrating smart technologies such as the Internet of Things (IoT), cloud computing and analytics, AI, and machine learning for better business outcomes. Data is the new gold for decision-makers, enabling them to optimize and accelerate their operations with confidence and reliability
While the industry is certainly embracing digitalization – particularly in terms of an enhanced ability to collect data – knowing how best to utilize it is key. Analyst studies suggest that industrial companies typically are able to use only 20% of the data generated which limits their ability to apply data analytics meaningfully. The application of artificial intelligence on data produces meaningful insights for prediction and optimization that improve business performance. AI is proving to be very effective for implementing predictive maintenance for companies and saving expenses. According to McKinsey & Company, AI-based predictive maintenance can boost availability by up to 20% while reducing annual maintenance costs by up to 10%.
AI can address issues before they impact productivity
When it comes to maintaining process equipment, there is no one-size-fits-all solution. Different strategies — from proactive, predictive maintenance to reactive maintenance — have their own benefits and drawbacks in terms of costs and time-savings. The key is to choose the right strategy for the right situation. This is especially important for rotating equipment (compressors, pumps, turbines, and others) that are essential in most industrial plants. However, getting a 360-degree view of the condition of rotating equipment can be challenging and time-consuming. Without it, industrial operators may miss the chance to optimize their maintenance plans and increase their operational efficiencies.
Condition monitoring is a critical aspect of asset management and maintenance. It enables the detection and diagnosis of abnormal activities or faults in equipment and processes to optimize maintenance and performance. However, traditional condition monitoring relies on the assumption that equipment failure is random and unpredictable and that the best way to prevent it is to perform regular inspections and repairs. It focuses on scheduled maintenance activities, such as lubrication, cleaning, calibration, and replacement of parts, regardless of the actual condition of the equipment.
A more advanced approach to condition monitoring is to leverage data analytics, AI, and ML. Enterprises can use data to understand how an asset performs and when it will degrade. This can improve maintenance and prevent failures. To achieve this level of predictive maintenance, the enterprise needs to incorporate more data sources. For example, sensors can capture data points from key components. Other valuable data sources include ERP and procurement data, historical maintenance and repair data, production data, and field reports from employees. AI can augment and enhance traditional condition monitoring by creating an expert system that delivers timely and actionable insights for asset management. It can analyze multiple sensor signals in combination and provide holistic and accurate assessments of equipment health. AI can also generate prescriptive recommendations and predictive estimates of future health and the remaining life of assets. This will allow operators to adopt reliability-focused maintenance strategies that reduce costs and improve uptime by minimizing unnecessary or late interventions.
Top of FormFor instance, if a plant operator wants to make sure their equipment is working well and avoid any breakdowns or accidents. What’s the best way to monitor the condition of the assets? The operator could use the traditional method of performing planned maintenance activities. But there are higher possibilities of missing some early signs of trouble or it could alert the operator too late. This method also ignores the bigger picture of how the overall assets are performing. Or the operator could use the smarter method of using AI. Through this, they can detect problems earlier and more accurately by looking at multiple sensors’ signals together. AI can also tell what to do to fix the problem and how long the asset will last. This way, the operator can save money and time by performing maintenance only when needed and not too often or too late. AI can help improve the plant’s overall reliability and safety by offering expert advice and predictions based on data.
Taking a transformative step in operations
The cost of unplanned interruptions, the impact of unforeseen failures, and the effect of unexpected breakdowns can result in significant business losses. Early detection of anomalies can provide critical information which can help prevent potential system failures and reduce downtime.
By using advanced technology like the ABB Ability™ Genix APM, plant operators can bring together condition information from disparate systems into one dashboard view, accessible via a web browser. It gives users who are both inside and outside the organization instant and secure access to the equipment data they need, so they can make decisions faster and prioritize actions that help optimize operations and reduce maintenance and operating costs. The suite also includes tools for analyzing historical data, which can be used to identify trends and optimize equipment performance over time. Some of its key benefits include aggregated equipment health overview, highlighting assets with the degraded condition, a 14-day failure prediction AI algorithm, a report generator, and a dashboard for raw data diagnostics.
System Anomaly detection using AI
Avoiding a trip in the plant is one of the major objectives of operations and maintenance teams. Any process upset or plant trips imply inherent hazards along with loss of production. Most often, operations and maintenance teams, as part of root cause analysis post-plant trip, notice changes in critical parameter patterns that caused the trip. Often, these go unnoticed due to limited resources and the massive amounts of data involved. The System Anomaly Detection App, which is part of the ABB Ability™ Genix Industrial Analytics and AI Suite, is designed to detect unusual, anomalous behaviour from process streaming time series data. It then uses Artificial Intelligence / Machine Learning (AI/ML) methods to support dynamic decision-making in all types of process-driven industries (including oil & gas, refinery, petrochemicals, metals, cement, and the like). Typical predictive maintenance solutions focus on asset health checks. The app, however, focuses on integrated assets (or systems) based on the function it performs. Potential system anomalies are highlighted using AI/ML for plant operator review, enhancing response time for process upsets. In essence, functionalities in the app are focused on reducing unscheduled trips, increasing plant availability, avoiding process upset conditions, and increasing operator responsibility. It also has the capability to identify factors that are responsible for the anomalous state of the system, allowing for immediate action and problem resolution before it affects the business process. The system anomaly detection solution can reduce unscheduled trips by up to 50%.
Unlocking the potential of AI and analytics in maintenance and reliability is not easy, but leading players in different industries have enjoyed significant rewards for their efforts. With product maturity and technology architecture in place, now is the optimal time to invest in technologies such as AI-based predictive maintenance, machine monitoring, and asset management. These systems can help industrial operators reduce cost and risk, improve reliability and efficiency, and become more sustainable and competitive.