Feature

The role of AI in remote power management

As the talk around artificial intelligence increases, Nnamdi Anyadike gives an overview of the systems embracing AI in the power sector.

AI power management promises efficiency gains. Credit:Art Stock Creative via Shutterstock

In recent months, the discussions around artificial intelligence (AI) have started to match the level of change that the technology could have on the power sector. GlobalData, the parent company of Future Power Technology, expects the overall AI market to be worth $383.3bn by 2030, growing at a compound annual growth rate of 21.4% from 2022.

Using AI, grid operators can better forecast demand by cross-referencing large datasets covering weather, consumer behaviour, previous trends, and local variables. This should save power and money, enabling more efficient grids and helping grid operators to make better decisions about when to store or release energy, ensuring a more stable and reliable power supply. “Smart grids” that incorporate advanced sensors, communication technologies and automation systems to facilitate the efficient and dependable distribution of electricity are a key application of AI in the energy and power market.

The systems on offer

Power trading technology company Gridmatic uses AI in its self-titled system to automatically sell power at the best possible rates. It offers fixed-price offtake agreements, removing risk for generators, while using its AI to optimise power sales on the company’s end. 

In January, the company unveiled a consumer system for power consumers. Called Gridmatic Retail, the system is designed to use AI to hedge power costs. This effectively helps companies reach carbon reduction goals by using sustainable energy contracts that lower costs, offer predictability and stability. The company says it can match consumption with renewable generation on an hourly basis at any time.

Combined with robotics, AI is proving to be a winning combination for the power generating industry. In March, Sarcos Technology and Robotics Corp, a US robotics company, announced that it had validated a technology to streamline solar power plant construction using autonomous machines.

In March, a US robotics company announced that it had streamlined solar power plant construction using automated machines

The company has now received $1.9 million from the US Department of Energy to develop the robotics system that will allow crews to install photovoltaic solar modules in the field faster, or increase a project’s size using the same crew. Further field tests will be conducted this year to validate the technology’s preliminary cost-saving analysis. Sarcos expects to launch the solution on the commercial market in 2024.

Meanwhile, renewable energy technology company Heliogen has demonstrated an AI-powered autonomous robot. Called Icarus, the robot is designed to lower the installation and maintenance costs of building a full-scale concentrated solar plant. Final enhancements to the robot are expected during the course of this year before its deployment across the entire range of Heliogen’s concentrated solar facilities. These employ arrays of mirrored heliostats that require upkeep and cleaning to ensure peak optical performance and a high level of energy generation efficiency.

The Icarus system uses GPS, ultrasonic rangefinders and Lidar sensors for a completely autonomous operation. “By taking advantage of huge boosts in processing power we aim to make solar energy more affordable,” said Bill Gross, founder and CEO of Heliogen.

AI essential for cutting edge predictive maintenance

But one of the most promising applications of AI in power management is in predictive maintenance. Any wise power company looks to avoid the costly mistake made by Siemens Energy, which recently took a $500m hit after a main turbine failed. Proponents of AI argue that the alternative is often over-maintenance. Routine maintenance keep operations running smoothly, but reducing this to the minimum necessary is a risky process.

Using sensor data, AI is capable of predicting when a component is likely to fail. This allows operators to schedule maintenance before a failure occurs, reducing downtime and minimizing the risk of unexpected outages.

The ‘knowledge graph’ used here currently contains about half a billion data points

Jan Weustink, head of Omnivise performance at Siemens Energy, said that at some point AI will also have the potential to make full-scale commercial autonomous power plants (APPs) a reality. Developers at Siemens Energy are now working on a power-plant database that is capable of being evaluated by a machine. This contains not just the properties of the power station components, but also all the connections between them. The company’s database can now bring together data from 50 power station projects based on 12 data sources using the data integration system. “The ‘knowledge graph’ used here currently contains about half a billion data points,” Siemens engineer Saskia Soller explains.

The dawn of automated power plants

A white paper from Adex, the developer of its patented Self-Tuning Artificial Intelligence platform, says the need for APPs is growing. It argues: “Most of the existing thermal power plants in the world are 10 years old, or older. While previous improvements were possible through expensive hardware retrofits and conventional control systems upgrades, these solutions do not fix all new grid demands and do not fit into the speed demands to update a plant.

“Novel developments, allow power plants to upgrade their performance, flexibility, and automation using advanced analytics software. As a result, APPs will be more efficient, more dispatchable, less pollutant and more economical.”

So far, APPs remain at the margins of power generating technology. The world’s first APP is Mitsubishi Power’s T-Point 2 plant, a next-generation power plant in Takasago, Japan that has been in commercial operation since the summer of 2020. The gas-fired plant supplies 566MW of power to the regional grid in Japan.

The facility brings together a range of technologies, but it is expected to provide the building blocks for a new breed of smarter, more sustainable and more integrated power plants. The T-Point 2 was designed as a digital facility from its inception. Virtual reality simulations of construction were conducted in 3D and used as the physical plant was constructed. Mitsubishi Power used its Tomoni suite of systems in planning, which used AI-enabled analytics to automate operational processes.

The offering of US software company AutoGrid enables the prediction, optimisation and real-time control of energy assets remotely. Rahul Kur, chief operating officer at AutoGrid, told CleanTechnica, “In just over a decade, the application of AI to the energy sector is producing dramatic results and achieving outcomes that would be impossible for humans to replicate. As the grid grows increasingly complex, the need for AI only deepens.”

He added, “With AI-powered virtual power plants, operators can forecast and optimize energy use. Wind forecasting is currently a challenging, but also a much-localised, problem. However, as more advanced modelling and AI systems emerge, it will become even easier to predict wind generation.”

GlobalData, the leading provider of industry intelligence, provided the underlying data, research, and analysis used to produce this article.

GlobalData’s Thematic Intelligence uses proprietary data, research, and analysis to provide a forward-looking perspective on the key themes that will shape the future of the world’s largest industries and the organisations within them.