How to Maximize the Value of Gas and Coal Plants with Industrial AI

Energy companies are overlooking viable ways to generate more value from their existing gas and coal plants. How can they lower costs and reduce unplanned downtime without increasing risk?

Today’s energy companies face the tough task of reducing their operations and maintenance (O&M) costs without risking any loss to their overall power production. To achieve just the right amount of asset maintenance — that is, without overspending and without incurring costly unplanned outages — is both a significant challenge and a constant balancing act.

But with new technologies like industrial AI and optimization, energy companies now have an effective and scalable way to create financially optimized maintenance strategies that lower their O&M spend and maintain or reduce their risk profile.

Why is this a problem?

The rate of change in the power and utility sector is accelerating. It’s being driven by a few key factors:

  • Decarbonization: Gas and coal assets are being retired and replaced with intermittent renewables.
  • Decentralization: As big gas and coal units are retired, the units they’re being replaced with aren’t as big. Generation is becoming more distributed.
  • Digitalization: More machines are coming online, the sector is becoming increasingly connected, and consumers have greater choice and control than ever before.

The above factors are intensifying the industry-wide challenges that are also on the rise, including: declining margins, increasing flexible generation, rising costs of unplanned downtime and the aging workforce.

So, even though the cost to operate and maintain gas and coal plants is rising — and budgets are falling — how can energy companies get as much value as possible from them?

Top three ways AI improves gas and coal plants:

1) Leveraging AI to clean operations data.

Industrial data is messy, inconsistent and prone to errors. AI solves the issue by automating the correction, standardization and accuracy of your data — so that it can then be transformed into actionable insights that drive financial outcomes.

For example, work order data is input manually by operators and varies widely in quality. To derive insights from this data, AI engines can use natural language processing and other techniques to identify errant entries and incomplete fields, and automatically suggest fixes to improve the accuracy and quality of that information.

AI can understand the manufacturer and model of the machine, the asset class, the specific parts that were used for the job, and the written descriptions and comments from technicians. Based on all of that, it then determines which fields need to be corrected.

With the correct data in place, this AI engine can go even further by examining asset fault codes and correlating underlying work orders to reveal which assets are contributing to the highest cost of maintenance.

2) Financially optimizing maintenance strategies.

With detailed visibility into work order data, you will now be able to answer questions such as: “Are my maintenance strategies effective in the way I expected?” and “Is there an opportunity to reduce my maintenance costs without incurring additional risk?”

By clearly understanding what’s going on with your assets, you can accurately measure the effectiveness of the maintenance tasks you’re performing, helping you answer critical key maintenance questions:

  • Measure availability: “What’s causing downtime in my assets?”
  • Gauge financial impact: “Which of my assets have the highest maintenance costs?”
  • Maintain based on actual conditions: “What can I learn about the current conditions of my assets?”
  • Continuously learn and execute: “How can I gauge the impact of my maintenance strategy changes?”

Once you understand which assets are problematic, optimization software can be applied to create the most financially responsible strategies, customized for your business objectives.

3) Capturing “shadow” capacity.

AI helps you recoup shadow capacity — the untapped power potential of your plant — by detecting anomalous behavior in your most critical assets, and by predicting and preventing machine failures before they have a chance to happen.

First, a Machine Learning Anomaly Detection (MLAD) Engine uses similarity-based modeling — a form of pattern recognition — to notify you of when assets aren’t performing as expected. It uses sensor data and ambient variables like temperature, vibration, flow rates and other factors to create a multi-dimensional pattern from which it can recognize deviations. By understanding the operating conditions and circumstances, the engine knows whether or not the exceeding of a threshold is acceptable, or if an alert should be triggered.

Second, Failure Prediction Engines can forecast the probability of future occurrences that you want to get advance notice of, such as machine breakdowns. By knowing the probability of a failure within a specified window of time, you can proactively make smaller and less expensive repairs in the near term — and on your own terms — to stop costlier and more critical unplanned outages from happening to you down the road.

AI informs you when immediate repairs are required and when they can be placed during planned maintenance windows. Combine this with optimized maintenance strategies based on real work order data, and the incremental value gets maximized for gas or coal-fired plants.

How much value are your plants leaving on the table?

AI was one of the most prominent themes of CERAWeek 2019, the energy industry’s premiere annual conference. Check out this educational session led by one of Uptake’s subject matter experts, Ted Miller, in which he provides a digital approach to driving more value from gas and coal plants.

Curious to learn more about how industrial AI can reverse the trend of rising O&M costs for energy companies? Check out our research to see the financial value that’s at stake: