AI Promises Broad Applications for Energy and Utility Companies


We’ve come a long way since Thomas Edison invented his first high resistance, incandescent electric light back in 1879. We now live in a world where our entire existence is dependent on electronic devices and equipment to survive and do mundane tasks.

As technology advances at light speed with Moore’s Law paving the way for more interconnected smart devices, some smaller than the eye can see, the energy industry is going through significant changes as it navigates rapid evolution in consumer behavior.

The globalized energy market has become complex and sophisticated, where companies can trade carbon credits for how much they pollute. As more energy markets deregulate, new players are entering the market.

Adoption of smart meter technology that can log usage in shorter intervals and report back to utility companies in real time through wireless networks has resulted in substantial new datasets. This data can be mined for key business insights to aid in critical decision-making. Energy companies are now turning to artificial intelligence (AI) and advanced machine learning solutions to leverage the vast data we now have at our disposal.

Below are five examples showing innovative ways that energy companies are using AI to harness the power of data, stay competitive, and offer improved energy solutions to society at large.

1 – Detecting & Reducing Non-Technical Losses (NTL), such as Energy Theft with AI

In some countries, as much as 40% of power distributed through the grid network is lost. According to one study, energy theft costs an estimated  €1.3 billion a year just in the European gas sector. NTL includes losses caused by energy theft, billing errors, or faulty meters (in contrast to energy losses for technical reasons, such as heat dissipated from power lines).

Energy theft is a growing problem in many countries, including in the United States. Energy theft takes place when consumers break into power boxes and short circuit connections to bypass the meter.

On-site inspections are costly. Suspected theft investigations need to target more accurately for the company to get net cost savings. With so much data about usage patterns collected from smart meters, companies can now use AI to find anomalies in usage patterns, which in turn could help improve focus for investigative departments within energy companies.

Atos, a Paris-based global services company, uses its advanced analytics technology to support European utility companies with their revenue protection processes by custom-developing machine learning algorithms for NTL pattern detection. In one case, they managed to increase the energy company’s ability to identify energy theft incidences six-fold and reduce the cost of their entire revenue protection process by 75% (the project demonstrated ROI and paid for itself in less than 12 months).

2 – Revenue Maximization with Power Purchase Agreements (PPAs) using AI

Energy companies use several tactics to sell power at optimal prices. One way is that they maximize their revenue while negotiating and selling Power Purchase Agreements (PPA).

Independent Power Producers (IPP), selling power generated from renewable sources, face unique challenges. Energy from wind turbines and solar panels typically gets transmitted through existing electric power grids. As production timings are based on weather and may not always match demand schedules, energy production either has to be stalled (e.g., wind curtailment), or energy gets stored until demand increases. Energy storage with batteries, compressed air, or hydrogen fuel cells can be expensive and reduces efficiency levels.

To help energy clients address this, StrategyWise has built models using machine learning that have seen an 8–10x accuracy increase over traditional manufacturer power curves in predicting power generation levels from wind turbines. To achieve this, the team uses a combination of weather data, location data, and real-time factors such as acceleration and blade angle. In combining this with models that forecast market trading price and volume, users can optimize a whole host of operational decisions.

For example, utility users can now use real data in place of pure intuition to work out the optimal power to sell. Their real-time traders use this model as a benchmark to continually improve their asset optimization and capitalize on opportunities. The company can now enhance bidding strategies and take more calculated risks that will ultimately lead to higher revenue.

3 – Avoiding Disaster from Equipment Failure & Optimizing Infrastructure Maintenance with AI Get a strategy for using data and AI in your energy utility

In 2019, PG&E filed for bankruptcy and agreed to pay $13.5 billion in damages due to likely device failure in their power lines that caused the 2018 California wildfires.

In the US alone, there are over 200,000 thousand miles of power lines. The typical process for maintenance involves a team of people actually walking the routes of these lines and manually inspecting for possible interference or damage caused by debris, such as fallen branches or bird nests.

Technology has become sophisticated enough that we can start to advance from these manual processes to ensure power line reliability. Avitas Systems, a GE funded company, is using AI and predictive analytics to process data from flying drones that can now do the manual work of power line inspection. With the rapid collection of visual imagery, this vast data can be utilized with AI to prioritize where to dispatch maintenance crews to perform preventative maintenance in order to limit further damage and potential disasters. Companies can save substantial money and lower risk to society associated with possible large-scale outages.

In other instances, StrategyWise has worked with gas utility clients to deploy predictive models that anticipate leaks before they occur. Using a variety of factors like pipe type, pipe age, and seasonality, the team has built neural network-driven models that showed a 40% boost in accuracy over traditional methods.

4 – Enhancing Customer Experience, Satisfaction, & Loyalty with AI

Consumers are spoiled for choice today, and not just for boxed cereal. Given the increasing trend for deregulation in the power sector and more open markets, many new entrants are competing in the power supply industry.

Consumer behavior has also evolved substantially in the past decade. In a world of give-me-everything-now, Amazon Prime delivery, and instant downloads, consumers want a fast response, the self-service capability to manage their lifestyle, and they want the best price to match it all. They will switch providers in an instant to get this.

The smart home has finally arrived. With devices such as Amazon Echo and Google Home linked to Philips Hue bulbs and Nest thermostats, consumers can use natural language to switch off the lights in the basement or increase the temperature in the bedroom. AI plays a huge role in this new technology. The interpretation of language and processing into real actions is made possible by neural networks and machine learning that gets more accurate over time.

Power companies can embrace these technologies by syncing their smart meters to the mix and offer customers real-time solutions to save money on their bill (e.g., by telling them to do their laundry after 4 p.m.). Power companies can develop features like this to build loyalty with customers (while also improving capability to effectively manage load distribution over the course of a day).

Cognizant, a large IT services company, worked with one US utility to develop automated systems using AI to deal with frequent customer service inquiries. Aside from providing faster service to the end-users, the innovation using AI resulted in cost savings by reducing the time customer agents spend on the phone by 7%.

5 – Innovation Planning & Optimizing for Growth in the Renewable Sector using AI

There is a growing shift toward consumer preference for sustainable energy sources. Families are producing electricity from solar panels on the roof of their home and selling the excess through the power grid. Families like this are now net producers instead of consumers and as a growing aggregate, a source of competition for incumbent power companies in a market where they historically enjoyed monopoly status.

For incumbent power companies, this situation creates many unique challenges. Even with plans in place to make some shift toward renewables, it is tricky for companies to find the right balance. Traditional fossil fuel or nuclear energy sources offer reliability and predictability to meet consumer demand. Renewable energy sources such as wind and solar have complications with storage and can be volatile due to unpredictable weather patterns.

No matter where companies are on the spectrum in regard to renewables, one thing is for sure: incumbents, especially in the oil and gas sector, can no longer afford to be complacent about managing costs.

Last year, ExxonMobil announced a partnership with Microsoft. They plan to make their Permian Basin operation “the largest-ever oil and gas acreage to use cloud technology…” Their operation “…is expected to generate billions in net cash flow over the next decade through improvements in analyses and enhancements to operational efficiencies.”

These examples are just the beginning of AI applications in the energy sector. StrategyWise specializes in helping energy and utility clients maximize the power of their data. If you want to learn more about how we can help you build an AI or advanced analytics strategy, contact us today.