How to Winterize Your Power Plant with AI

If you live and work in Texas, you probably thought the 2011 Groundhog Day blizzard was just another black swan event—an anomaly. Something that would most likely never happen again. After all, Texas winters are generally uneventful — a respite from the blistering 100-degree temperatures. Just a short decade later, the Lone Star state faced reality once again: single-digit temperatures, statewide blackouts, and concerned citizens wondering why the state’s power grid failed yet again. 

At the peak of the 2021 Texas freeze in February, nearly half of the state’s power generation (53 out of nearly 110 gigawatts) went offline at some point as demand for gas and electric-generated heat spiked across the state. If that isn’t eye-opening enough, the Electric Reliability Council of Texas (ERCOT), which operates the state’s independent grid, reported they were within minutes of a total collapse of the statewide grid. Reports show that it would’ve taken weeks — or even months — to restore. The thought of energy-rich Texas going months without power is truly inconceivable. 

Lessons From the 2021 Texas Freeze

Texas freeze 2021

A few short weeks after the crisis, our Founder and CEO, Amir Husain, shared his perspective on the events of the storm in an article on Forbes.com. He spoke at length about the lack of preparation, that equipment at gas, nuclear, and wind facilities had not been made weatherproof. Was the 2021 Texas freeze foreseeable? “In truth, yes,” he said, acknowledging the state’s failure to learn from the 2011 snowstorm. How is it that we could not adequately anticipate and prepare for such a disaster when meteorologists at ERCOT and the National Oceanic and Atmospheric Administration (NOAA) issued warnings of the storm’s severity over a week in advance?

In his article, Husain asserts that there will be other unforeseen events in the future. He further explains that the unthinkable now happens often enough that designating such natural disasters as black swan events is unfounded and the time for meaningful action from policymakers is long overdue. He says, “If there is one lesson our regulators, operators, and politicians in Texas take to heart, it should be this: ‘Start planning for what you’ve traditionally considered outlier events. And leverage the best technology to protect the state and its people. Now. There is no excuse to remain a deer in the headlights in the age of artificial intelligence.’”

As a leading AI company, SparkCognition’s mission is to build AI platforms that enable the world’s most creative problem solvers to ignite lasting impact in their organizations and throughout the world. Thus, we couldn’t turn a blind eye to climate disasters like the Texas freeze and not offer a solution with lasting, meaningful impact. One important facet of our business is leveraging AI-based predictive analytics and decision support capabilities that provide a significant opportunity for industrial leaders to better prepare for the unexpected, and just as importantly, operate more effectively and efficiently when the unthinkable black swan disaster strikes again. 

A variation of this solution has been delivered successfully across a variety of industry verticals including oil and gas, manufacturing, and aviation. For example, in 2019, an exploration and production energy company commissioned us to develop and deploy our predictive analytics for anomaly detection solution to its unmanned platform, specifically to improve the stability of a multi-phase pump in order to meet production rate targets. By leveraging predictive analytics on this pump and performing maintenance before failure becomes imminent, the E&P company stands to gain up to $500K of production for each day of downtime avoided.

Our solution extends upon these capabilities to provide a form of “soft winterization” of pumps, valves, and other critical pieces of equipment in a power plant. The goal is to ensure uptime, especially during the most critical moments when a trip or outage is possible, so that power producers can provide customers the electricity they need to survive during extreme weather events — whether hot or cold. While various winterizing retrofit solutions are available today, they are relatively expensive to deploy. These hard winterization solutions include building insulated enclosures around critical components, insulating instrumentation cabinets, heat tracing water, air, sensing and drain lines, and much more. Even worse, many of these solutions will take months or years to be completely deployed. Given the Lone Star state’s track record, months or years is simply unacceptable; crises that threaten our state’s power grid may be just around the corner. 

SparkCognition’s “Soft Winterization” Solution

winterizing power plant

While hard winterization is desirable to protect critical infrastructure in the long run, operators should consider adding additional layers of protection that increase operational resilience even further. Our soft winterization solution utilizes AI technologies like machine learning-based predictive analytics and digital twin technology to provide asset performance management and decision intelligence, and is a complementary force multiplier to hard winterization efforts.

In essence, AI can help leaders de-risk their operations by giving greater visibility into their operations as well as ample time to anticipate and proactively respond to high-impact, low-probability events. With our AI-based solution, soft winterization can be implemented within weeks or months, providing an invaluable layer of protection against extreme weather conditions in the future.

Navigating Power Plant Operations with Our “Soft Winterization” Solution

Today’s power plants generate an abundance of data that can lead to increased awareness and improved handling of operational impacts in real time. However, data by itself is insufficient to deliver substantial value; there’s simply too much data for human analysts to sift through without error. Artificial intelligence takes the guesswork out of extracting meaningful insights from data in times of great uncertainty, such as during a winter storm. However, this is not all that AI can do. Even more importantly, this AI-based solution enables operators in a power plant to execute necessary process changes, such as switching to a backup pump or valve, to keep operations up and running. 

Our solution is comprised of three key elements:

  • A reliability/engineering assessment
  • An AI-enabled intelligent advisor
  • Digital twin technology

Let’s break down each element.

Reliability Assessment

Before a power plant operator can apply an AI solution, experts must perform a comprehensive evaluation of the plant’s risk associated with extreme weather events. This assessment uses Reliability-Centered Maintenance (RCM) analysis to evaluate failure modes and identify the most vulnerable components and sensors during extreme conditions, such as below-freezing temperatures. 

AI-Enabled Intelligent Advisor

Once the reliability assessment is complete, operators can leverage an AI-enabled intelligent advisor that not only augments their capabilities but also improves the speed and quality of their decisions. For example, the intelligent advisor may recommend an inspection to test all vulnerable components or assets ahead of an incoming storm. 

Let’s say it’s another winter storm with sub-freezing temperatures with the potential to freeze critical assets. The intelligent advisor would issue temporary operating instructions to the operator to start monitoring weather enclosure heating systems and direct them to begin cycling through backup pumps and valves as temperatures continue to drop. It could also trigger an alarm to drain or purge water from the systems that are currently down or tripped, or monitor ice formations in cooling towers. All in all, the AI-enabled intelligent advisor enhances an operator’s ability to monitor process execution and make better decisions faster. As events such as winter storms tend to unfold in unexpected ways, improved decision-making is central to the soft winterization solution.

Digital Twin Technology

The final element, digital twin technology, improves upon data analytics to eliminate potential error when instrumentation, sensors, and sensing lines are the most vulnerable components in a power plant during extreme weather conditions. Digital twin technology alleviates, or potentially even eliminates, these vulnerabilities by providing imputations of missing or inaccurate data values using statistical techniques applied to the uncompromised inputs and overall model of the system. If there is a likelihood of instrument malfunction due to severe weather, the expected calculated value could be used temporarily as the operation shifts into an emergency mode of operations.

In Conclusion, What Does the Soft Winterization Solution Ensure?

Increased operational resilience! This is crucial to prevent blackouts from occurring during extreme events like the recent Texas power crisis. Governments, power generation companies, and other leading organizations have a significant opportunity to prepare and build operational resilience for low-probability, high-impact events. This added layer of operational resilience enhances the benefits of hard winterization initiatives once they’re finally deployed. But in the meantime, our solution can be deployed immediately to de-risk operations before those efforts have been completed.

Learn more about our soft winterization solution and how it can protect your power gen operations from extreme weather conditions.

SparkCognition is committed to compliance with applicable privacy laws, including GDPR, and we provide related assurances in our contractual commitments. Click here to review our Cookie & Privacy Policy.