Artificial intelligence is much more than just one technology – it comprises expert systems, machine learning (ML) programmes, prescriptive and prognostic models, reinforcement learning, large language models (LLMs) and generative AI.
Jim Chappell, global head of AI and advanced analytics at AVEVA, explains how this broad spectrum of AI technologies work together with the cloud, data management and analytics solutions to provide utility companies with powerful insights and capabilities to optimise their operational sustainability.
How is AI being used to make the energy sector more sustainable?
AI-infused solutions can turbocharge industries’ progress towards efficiency and sustainability. Although AI has been helping to lower carbon emissions for many years, we have only scratched the surface of its potential in the overall area of sustainability. For example, predictive analytics can quickly identify underperforming assets or incorrect control settings that result in additional fuel being burned and generating excess greenhouse gas emissions.
Furthermore, ML systems can help predict energy consumption. For example, data-driven AI can be combined with physics-based simulation to better emulate operations as part of a grey-box modelling system.
Carbon capture simulation is another area where AI can support sustainability. AI can also help make renewable and alternative fuels, such as wind and green hydrogen, more economically viable and competitive with traditional sources, thus accelerating the global green energy transition. All of these factors will support industrial companies on their journey to net zero.
These types of AI solutions do not involve training massive LLMs, so they don’t consume large volumes of power. In the case of grey-box modelling, where AI models replace physics-based models, AI runs substantially faster, thus requiring significantly less energy than its alternative.
As AI evolves and becomes more objective-driven, it will play an ever larger role in climate change control and overall sustainability. This will also include more AI-driven closed-loop control for optimal performance.
What types of industrial insights can AI-powered solutions offer?
Using AI models trained on various types of historical data allows utility companies to more accurately predict demand, manage varying energy sources, forecast asset failure and fine-tune their operations. For example, AI can analyse sensor data from many types of energy storage systems and zero-carbon power generation, including wind, solar, nuclear, hydroelectric and geothermal. AI can detect problems long before a human or other type of software can, significantly increasing operational efficiency and reliability.
In addition, AI can model alternative energy production, including green hydrogen. Large advances can be achieved by using AI models intermixed with physics-based simulation.
AI black-box models run hundreds of times faster than a corresponding physics model and are much easier to set up, enabling them to identify issues in near-real time. Furthermore, by combining prescriptive, prognostics and predictive analytics, organisations can optimise resource allocation and minimise waste, resulting in significant savings.
How can organisations use AI to identify energy-saving opportunities?
On the demand side, AI-driven energy management systems are empowering consumers and businesses to take charge of their own energy usage. By analysing consumption patterns and providing personalised recommendations, these applications promote energy efficiency and reduce costs.
For example, US-based utility provider Dominion Energy is using AVEVA’s CONNECT Data Services, powered by Microsoft Azure, to collect and share real-time data on energy sources and power flows. This enables customers to track their usage, supporting their progress toward net-zero goals.
In what ways can AI help to track and reduce global carbon emissions?
AI-powered models can predict the impact of operational changes, enabling companies to make real-time adjustments for efficiency and cost savings. For instance, AVEVA is working with TotalEnergies to monitor over 110 greenhouse gas reduction projects. Data from various global operational sites is fed directly into dashboards at the company’s French headquarters using AVEVA’s PI System. After verifying and contextualising the data, the team analyses performance and measures results against specific key performance indicators. This enables them to monitor 85 per cent of emissions accurately and use AI models to forecast the impact of operational changes. In one case, optimising power delivery at a single site cut carbon emissions by 15 per cent annually. Scaling this approach across TotalEnergies’ portfolio could result in tens of thousands of tonnes of carbon savings each year.
Another example involves predictive and prognostic AI being implemented at a major US power producer as part of an overall reliability-centred maintenance programme. When a generating unit was brought back online after repairs, an operator inadvertently set the extraction steam temperature too low. Thankfully, predictive analytics caught this issue very early, so the company was able to quickly fix the problem, allowing it to prevent 2,000 tonnes per week of excess carbon dioxide going into the atmosphere and saving 100,000 gallons of water.
What role does open-source technology play in amplifying AI-infused solutions?
Open-source collaboration can accelerate the potential of AI and ML-enhanced energy networks. Open technologies establish a shared language for data exchange, ensuring compatibility across different systems. This enables data from sources like sensors, smart meters, weather stations and maintenance logs to be seamlessly integrated and analysed together.
Historically, the proprietary nature of energy systems limited innovation. Today, open technologies allow data and ideas to be shared across departments, companies, industries and countries, driving deeper collaboration and faster progress in the energy sector.
How will quantum computing impact the future of AI and energy efficiency?
AVEVA recently signed a memorandum of understanding with Oxford Quantum Computing to explore how quantum power can reduce the energy load of AI. A Forbes study suggests that simple quantum apps can cut energy use by up to 50 per cent. Currently, it takes the equivalent energy of running 47 US homes for a year to fire up a supercomputer, while it takes the energy needed to boil just seven kettles of water to run a quantum computer.
However, quantum has a number of hurdles to cross in order for it to become mainstream. It is prone to error, susceptible to noise, requires complex hardware and lacks scalability and reliability. A lot of work is being done to overcome these challenges. In the future, quantum is likely to provide a very energy-efficient alternative to intensive computing, such as AI.
Can you give an insight into how AI will transform AVEVA’s operations?
Successful AI implementation will depend on building partnerships that drive large-scale innovation. A prime example is our collaboration with Microsoft, where we leverage AVEVA’s industrial expertise in the cloud to benefit customers worldwide. This approach is crucial to delivering the advantages of AI at scale.
Additionally, we are actively working with governments and our industrial clients to continually refine AI models, enhancing efficiency and supporting the energy transition.
Learn more about the advantages of implementing AVEVA's industrial AI-based solutions on the AVEVA website
Discover more insights like this in the Winter 2024 issue of Technology Record. Don’t miss out – subscribe for free today and get future issues delivered straight to your inbox.