WHEN STANDARD DEFINITION TV WAS superseded by high definition, it was heralded as a major leap forward in viewers’ clarity – until 4K screens came along. The latest advances in data analysis are bringing a similar leap forward in the nuclear industry’s asset management programmes.

The indications from the first steps in data analytics are that it has the capacity to provide operational intelligence to transform the efficiency and availability of nuclear power stations, increasing revenues at a time when competitive pressures are greater than ever.

Rolls-Royce’s data analysis is facilitated by its digital services customers and plant control and monitoring systems, which harvest performance data from around 195 reactors around the world.

Rolls-Royce has been partnering with customers to showcase its data science, which has been a step-change in clarity and accuracy of analysis for operators. Now it plans to take a ‘4K TV-type’ leap. 

Collaboration on longer projects – Bruce Power in Ontario and PSEG at Salem and Hope Creek – is the way ahead for these services and plans for an expansion to European markets are under way.

During a collaboration, there is a period of data gathering and analysis to diagnose areas for improvement, and follow-up site services to make improvements quickly so the benefits can be gathered as soon as possible.

Higher definition from data

Using algorithms and human nuclear operations expertise, data science brings new clarity, insights and intelligence into nuclear power station operations. The results are better decision making, efficiency and availability.

Big data are largely meaningless in isolation, but analysis gives a much clearer picture. And the Rolls-Royce data science allows nuclear operators to see more clearly how their equipment is performing. They can also predict to a much greater degree how that performance will continue, so they can schedule site support services for the right time, so the management of assets and services is optimised.

The approach Rolls-Royce is about to bring to global nuclear power stations’ asset management builds on 30-years of experience applied to the aftercare of jet engines. Some 13,000 Rolls-Royce jet engines around the world carry monitoring systems which constantly send data to a control centre, which then analyses it and uses it to forecast the optimum time for maintenance.

Maintenance unavoidably incurs non-revenue-earning down time, so it is best to group together tasks so that downtime is minimised. If maintenance is carried out too early, it is unnecessary cost; and if it is too late it means unplanned outages.

Rolls-Royce believes that its technology with data analytics can forecast to a far higher degree of accuracy than ever before equipment degradation and remedial activities, so maintenance is timely and efficient and there are fewer unplanned outages. This improves cost avoidance and results in substantial savings. And because lots of data is collected from lots of identical components all the time, the quality of the forecasting keeps improving, as does the technology behind the analytics. This repository of information is called a data lake and the same principle is applied by Rolls-Royce to nuclear power stations.

Rolls-Royce has its own nuclear data lake, a secure repository of data on installed equipment and inventory from nuclear power stations around the world. It has been collected during the last decade from the routine data streams transmitted periodically from its plant obsolescence management system (POMS). It is this data Rolls-Royce uses to help predict the performance of components in a nuclear site.

Its aim now is to combine this existing data with new data from customers’ sites and make available digital systems that harvest new data in order to carry out real-time analysis that allows live reporting and forecasting for its customers.

Rolls-Royce is targeting five areas of known high value for customers:

  • Improved equipment reliability
  • Online asset monitoring
  • Preventative maintenance reduction
  • Inventory optimisation
  • Reduced operating cost

Targeting maintenance improvements

As one of the most forecastable and expensive parts of a power station’s programme, maintenance windows have always been near the top of operators’ lists for finding improvements and efficiencies.

One of the foremost thinkers in nuclear energy generation approached Rolls-Royce to trial maintenance optimisation analysis, outage planning and how more real-time analysis could be used to improve availability and efficiency.

Rolls-Royce used its data lake to forecast likely component degradation rates against windows for planned maintenance and identified the best opportunity for replacement or repair. This meant scheduling maintenance within planned outages. Adjusted preventative maintenance strategies saved 60,000 person-hours.

Another customer, after a trial of the so-called T-104 analysis, which looks at detailed planning prior to scheduled outages, found that 42% of all preventative maintenance activities could be less frequent. The customer accepting 90% of these outputs. Rolls-Royce now has an embedded a team with this customer to use its demand-driven forecasting to reduce overall preventive maintenance, improve equipment reliability and reduce material consumption.

Spares inventory, stocking logistics and sourcing

Unused components and spare parts sitting on shelves are a huge upfront cost, which can be optimised if operators know better when they are likely to be needed. The greater insights provided by data analysis increase certainty around, for example, stocking, obsolescence management and purchasing.

Using big data, Rolls-Royce teams helped a customer analyse stocking and availability of spare parts, on site and in the marketplace. The result was a dramatically reduced inventory that allowed storage space to be freed for other projects.

For another customer, Rolls-Royce was requested to analyse and forecast inventory levels, identify possible savings and propose a strategy for managing future inventory sourcing. We investigated the quality of information being communicated and purchasing and stocking practices, then we analysed the customer’s data and compared it with its existing data to identify excess material based on projected activity levels by using operational and industry data. The team used a data algorithm using historical and forecasted use of certain materials and components, factoring in stranded material, criticality and material reconstruction bills to make recommendations. The result halved the cost of inventory, by buying more smartly, finding ways to liquidate existing unnecessary stock and changing re-order rules.

The same customer used the improvements to target sourcing: what does the global components market have to offer; what’s required and when; for how long is that component likely to be serviceable before further maintenance or replacement is required. Purchasing strategies were then agreed so that the cost of critical spares could be reduced.

Reduced operating costs

A utility customer asked Rolls-Royce to improve maintenance schedules, keeping them in a strict resource and regulatory envelope. Particular attention was paid to adjusting preventative maintenance strategies, using best practice experience from other areas, targeting programme waste and identifying non-value adding tasks. It created $11 million in savings and improved equipment reliability across the fleet.

Scratching the surface

While ‘data into intelligence’ is tried and tested in aviation, it is immature in nuclear power stations, and has huge potential.

Rolls-Royce has created a sub-brand called R2 Data Labs. Its firm belief is that data is everywhere and understanding it can improve everything. R2 Data Labs is a hub through which all new data science ideas are incubated, developed and deployed into its aerospace, defence and power systems business units as customer propositions, to improve efficiency, availability and compliance.

This clearly ticks the boxes of nuclear operators but it requires changes to culture and process.

At first, it may not appear too great a leap in practical terms because using digital monitoring systems to harvest real-time operating data is a purely logistical exercise. But what do you do with all that data? It drives change in how business systems and processes are set up to respond to the intelligence that the analysis provides. This in turn requires a workforce that sees data science as enhancing, not threatening, their activities.

In the near future, machine learning, artificial intelligence and automation will be deployed for standardised tasks, and wearable digital technology will start to blur sharp lines between humans and the systems around them. Humans will always be in charge, but their role will be controllers, handling nuanced activities not deliverable by technology alone.

Rolls-Royce plans a seamless end-to-end process, where it installs systems that securely send data to regional hubs for analysis. Nuclear site operations experts in those centres provide 24-hour support to customers across their fleet and when maintenance or remedial activities are identified, they are automatically resourced and programmed for delivery by a site service team. The result is ‘just in time’ delivery, greater efficiency and availability.  


Data science is driving change in our industry

Data science is proven, but it challenges us as an industry to think differently to unlock its value.

This change is happening – a hugely positive development when competitive pressures require us to open up our minds and innovate for our survival.

As a supplier, I see progress in the data science space in the move to collaboration and away from simple service procurement. Collaboration is the only way to maximise the value of data science because each organisation has different opportunities that must be explored in partnership.

This can feel like exposing a station’s vulnerabilities. But adopting a data science approach is not about identifying what you think you need, running your finger down a list of services and choosing one. It’s a joint process that requires customer and supplier to trust each other to work for the same goal.

At both a human and organisational level, collaboration is a strength, because that is how we make the biggest advances. But it is a journey to get to that point and that takes time and transparency.

As a population largely made up of engineers and scientists, we love technological advance, enhancing our knowledge and skills. Data science upskills teams by giving them more information so they can be more rigorous about decision making and planning.

For our ageing workforce, machine learning and artificial intelligence remove the pressure to perform repetitive standard tasks, thus helping meet our labour challenge. Technology is ready to counter a physical decline, and increase operational resilience for standard tasks in future.

Our revenue challenge is stark. Unlocking more value is achievable now if businesses are willing to embrace the support that is available – from a collaborative approach, to using data science to reveal areas of improvement invisible to the eye.

As an industry, we love as much definition as we can get and data science creates levels of clarity we never knew we could achieve. Making the most of that clarity requires a strategic change in how operators run their nuclear stations.

We have never had such a capability to capture, harvest, analyse and understand what is going on in our stations and organisations. My conservative estimate is that there are $4bn of savings available in the North American sector alone over the next three years. Those savings will remain invisible and unreachable until data science is deployed. 

Paul Tobin (pictured above) is Executive Vice President, Nuclear Engineering Services at Rolls-Royce