Feature
The business of producing and delivering energy is vast, and energy companies are turning to artificial intelligence (AI) to predict energy demand, cut the costs of production, streamline analysis of data and find new customers for their services. The following case studies offer some examples of how AI is being used by different segments of the energy industry:
1. Engie Energy Access
Many places in sub-Saharan Africa are off the electric grid, which means there’s not only a need for access to power, but also a huge potential market for companies offering solar panels for homes. One such company, Engie Energy Access, targeted Kenya as a market for its products, but it needed a way to identify potential customers, according to a case study.
The organization turned to Atlas AI, which relies on Google Cloud and uses geospatial data and machine learning (ML) to provide predictive analytics.
The challenge was to find areas with a high-density population but inconsistent grid access, figure out who could afford Engie’s products and identify individuals that might be interested. Atlas AI created models with various factors, such as consumer spending levels and villages with electricity. It predicted which customers to target and with which product, and it identified new markets for expansion.
In a pilot program comparing two regions where Engie followed traditional marketing tactics to one using Atlas’s recommendations, the AI-targeting region outperformed, with an increase in monthly sales of 48%.
“With over 700 million people around the world still without energy access, new tools are needed to forecast where consumer demand will be and connect complex supply chains to the identified communities,” says Abe Tarapani, CEO, Atlas AI. The latest advances in geospatial AI, he said, can help “unlock this capability for sustainable infrastructure providers."
Results
- Hyperlocal predictions allowed the creation of a targeted marketing plan
- Model identified the best timing and location to expand into new territories
- Expected increase in business in Africa of over $100 million a year
2. Duke Energy
Duke Energy set a goal to achieve net-zero methane emissions by 2030. It is difficult to make frequent checks of the natural gas pipelines that may leak, so the main existing method for finding methane leaks is to make calculated estimates, according to a case study.
To improve on this, Duke turned to Microsoft, which used its Azure and Dynamics 365 to develop a new method and then had Accenture and Avenade build the solution.
Duke Energy explored new monitoring systems, but it needed a way to make effective use of the data those systems produced. The new platform started with measures of emissions from natural gas utilities, then added data from satellite monitoring and ground-level sensors. AI assessed the data and provided reports in near real-time, so crews can be sent out quickly to repair the leaks, preventing an important source of greenhouse gas from entering the atmosphere.
“Our work with Duke Energy and Microsoft demonstrates how technology, innovation and artificial intelligence can help address sustainability challenges,” says Mark Schuler, managing director, Accenture.
Results
- Graphic dashboards quantified and prioritized measurements
- Geolocation data allowed crews to pinpoint leaks and fix them more quickly
- Scalable with other equipment and types of emissions
See more: 10 Top AI Certifications for Energy Pros
3. Marathon Oil
The oil and gas production company Marathon Oil has nearly 4,000 wells and wanted to create intelligent alerts that warn the company about changing conditions at those wells and other facilities, according to a case study.
The company had time-series production data from its wells, but it took months to produce alerts that it could act upon. Applying AI and analytics to that data helped dramatically shorten the process.
Marathon was already working with Amazon Web Services (AWS), which introduced the company to Seeq, an analytics company for the process manufacturing industry that specializes in applying AI and machine learning (ML) to time-series data. Seeq provided various applications to power the analytics process and join that with Marathon’s own data. When someone in the field identified an issue, they could use the software to investigate the related time-series data, set up an alert to warn about potential future issues and assign someone to check it out, all in a matter of hours.
“This solution is increasingly important every day,” says Mark Betts, IT manager of digital solutions, Marathon Oil. “Keeping wells online and limiting deferred production are the biggest opportunities to assist the assets going forward.”
Results
- System allowed Marathon Oil to connect production data over all of its wells
- Automatically generated Information
- Generated 1,500 tasks and 1,500 notifications a month
4. AES
AES was shifting its energy business from fossil fuels to renewables. As such, it needed methods to predict energy output from changeable renewable sources, predict failures and optimize distribution of the electric load, according to a case study.
With the help of H20.ai, the company put in place predictive maintenance programs for wind turbines and smart meters and developed a bidding strategy for its hydroelectric plants.
To predict failures in different wind turbine components, the AES used data from the manufacturers that relied on physics-based models and combined that with the H2O.ai Cloud to create new models that predicted failures with 90% accuracy. This allowed AES to avoid unnecessary repair trips or sending the wrong equipment, cutting the cost for each repair job from $100,000 to $30,000. They also used AI to distinguish actual problems with smart meters from attempts to tamper with them, eliminating 3,000 non-essential trips. They also measured demand, predicted rainfall and market conditions to get the best value from energy bidding.
“We didn’t have a single data scientist on payroll when we started our AI transformation,” says Sean Otto, director and head of AI, AES. “We started with a few high-value use cases, built a team and found a platform to accelerate our results.”
Results
- Achieved $1 million in annual savings by eliminating unnecessary maintenance trips
- Saw 10% reduction in customer power outages
- Addressed 85 business challenges over two years
5. SLB
The oil field services company SLB constructs oil and gas wells for energy companies, often at a fixed cost. They needed to analyze historical data to predict how long it will take to drill a well, so it does not face cost overruns and to figure out the correct size of the project to fit a contractor’s bid, according to a case study.
Part of the challenge in making their assessments was that much of the relevant data existed in unstructured form, such as in daily drilling reports. Additionally, the time to respond to a bidder was typically short, and human error can creep in.
SLB was able to use an AI-driven system to streamline its prediction process. SLB turned to Dataiku, which used a natural language processing (NLP) application to mine client data. Another application predicted the sequence of operations needed to drill a particular well and also the time the project will take.
“The applications provide SLB engineers with a data-driven way to predict the time and risks associated with well construction,” the case says. “This is of critical importance, as 30% of SLB revenue is associated with integrated well construction projects.”
Results
- Reduced the time for classifying data for a well from about eight hours to 20 minutes
- Assessed over $10 billion worth of bids
- Implemented application as part of SLB’s standard operating procedure (SOP)
See more: Energy Hungry AI: Is It Sustainable?
