How AI, IoT and data analytics are transforming nuclear safety

By Christopher Crosby, Global Nuclear Industry Principal, AVEVA

While the pandemic may have accelerated digital transformation across virtually every industry, a number of pre-existing factors have seen technological investment and innovation in the nuclear power sector rising over the past few years.

One of these is the increasing focus on sustainability. With global governments under pressure as worldwide temperatures rise year after year, nuclear power is seen as a means of meeting the energy requirements of a growing population without the need to emit the same level of emission by-products as hydrocarbons.

According to figures published by Statista, the cumulative capacity of nuclear power plants in operation worldwide reached approximately 382.5 gigawatts in 2021, a slight dip from its high of 396.6 in 20218. As of May 2022, there were 439 reactors in operation worldwide.

While safety has remained a concern for much of the general public – many of whom recall Chernobyl and, more recently, Fukushima – a raft of technologies are helping ensure that today’s nuclear power plants are safer than they have been at any point over the six decades the world has been using them.

AI: Enabler of safety and performance
One of the key technologies that have helped make nuclear energy plants safer is artificial intelligence (AI). Before even building a plant, AI models can be used to simulate the behavior of nuclear reactors, and use the data collected to improve a reactor’s safety, as well as its design, performance and fuel loading.

In a running nuclear energy plant, AI can – with a combination of machine learning and physics-based models collecting data from sensors – predict possible accidents, and alert personnel to the faults in equipment that can cause these. With machine learning, scientists can analyze substantial volumes of computational data to more efficiently model reactor behavior.

The Tianwan Power Plant in Jiangsu will be the largest nuclear plant in 2027, providing electricity to millions of people on China’s east coast. Here, engineers use the AVEVA PI System data management software to reduce the risks of radiation and extend equipment life by analyzing equipment performance parameters in real time, and by comparing their plant’s performance to other plants across China. Plant operations data is streamed into easily understandable flowcharts and dashboards, so that engineers are alerted to divergences from expected results and can quickly and proactively analyze and diagnose issues – and share these lessons with their peers and partners.

The power of data and IoT
Another important tool that has seen advancement in recent years on the nuclear energy front is data analytics.

A 2018 paper by the New Mexico-based Sandia National Laboratories offers a useful breakdown of data analytics and the industrial internet of things (IOT) for nuclear power. Sensors create large amounts of structured data, which is analyzed by machine learning tools, such as neural networks, that use edge analytics to process incoming data, including from accidents and abnormal events. The incoming data is processed and streamed to plant operators.

If the data is in text form, it may be unstructured and less useful from an analytics perspective, since it relies on forms that may not be easy to read, such as an engineer’s notes. However, natural language processing (NLP) tools can help bodies such as the IAEA parse large bodies of text. Many rely on text analytic question-answer models, including IBM Watson commercialized technology.
AVEVA’s Integrated Management System (IMS), first developed and implemented for use with Finnish nuclear company TVO’s European Pressurized Water Reactor (EPR), factors in more than 800 pipe and instrumentation diagrams (P&IDS). The 3D PDMS model of the EPR is highly detailed, and includes all objects – including smallbore pipes, instrument lines and details of every support and anchor plate. Despite its large size, AVEVA’s 3D model can still fit on an average laptop.

Information flowing in from the model is fully automated, and not only offers alerts related to the condition of each component in the system but is also used to update AVEVA’s materials management software, VPRM, ensuring that both maintenance and procurement needs are taken care of.

The solution, deployed by the French power and renewable energy major AREVA at TVO’s Generation III+ reactor in Olkiluoto, Finland, offers reliably up-to-date design information at each location. Integrating data and work processes in ways that have not been possible before has brought greater control and achieved higher quality and consistency, leading to on-time delivery of the project. AREVA is now putting the toolset to work at new projects in France, Germany, Japan, and the US.

Technology is vital for both safety and efficiency
There is a lot of important work being done in the nuclear energy space, and technology offers a reliable, improving platform for this.

● As the world seeks out sustainable energy solutions to combat climate change, nuclear energy offers a reliable, efficient method of power generation
● While past incidents have fed public mistrust in nuclear energy, digital transformation and technology have made today’s facilities safer than ever
● AI and machine learning can build models that predict future accidents with increasing accuracy
● Data analytics, fed by IoT sensors, can both gather and analyze information coming in from a number of sources in a plant

It’s clear that advanced technologies offer a means of not only reducing the risks of accidents but also allow plants to function with greater safety and efficacy. Nearly 70 years after the world’s first nuclear power station began operating, in Obninsk in the Soviet Union in 1954, this means of sustainable energy generation has never been more valuable.