Digital twin: taking energy and utilities condition-based maintenance to the next level
From my experience of digital transformation in the energy & utilities sector, the biggest potential – for cost savings and operational benefits – is around how plant assets are operated and maintained. Given the millions of euros required to keep plants going, the challenge has existed for decades: how to maximize the value and life of assets while maintaining quality and continuity. It’s one reason why CAPEX planning is so time-consuming; and allocations can depend on ‘who shouts the loudest’.
In the last few years, the availability of high volumes of data has enabled more evidence-based decisions. However, this data has mostly been used for bottom-up CAPEX investment cases. It’s only very recently that significant technological advances have converged to truly transform investment and maintenance planning in energy & utilities. Namely, new advances in digital twin technologies combined with the power of machine learning (ML) and artificial intelligence (AI) to optimize use of assets at every stage of their lifecycle.
Power of digital twins for energy & utilities
A digital twin is a live, evolving digital representation of a physical asset (or process, system or service) that is complete at any scale. This can be used to model almost anything – from a wind turbine, to an engine part or a generator – leveraging all the information created during its design, manufacture and operation, as well as real-time operational data from IoT sensors.
Using digital twins, energy & utility companies can simulate detailed ‘what if..?’ scenarios for strategic investment planning. They can change different parameters to see the impacts of different investment decisions and plans at an enterprise level. This capability can significantly simplify the planning process while increasing the value from investment portfolios through more precise risk-based CAPEX allocations. And it helps to optimize operations, using simulations to plan fleet capacity and downtime based on the outcome of different scenarios.
Combining simulations with ML and AI
When digital twin simulations are combined with ML and AI, it’s possible to detect abnormalities and deviations in the behaviour of assets to predict failures or risks in assets and plants. So, if you use ML and AI algorithms to find the correlations, and then work with those in the digital twin, you can see the precise trajectory of the predicted failure or risk in your simulations.
This optimum predictive capability enables businesses and their supply chains to evolve away from traditional schedule-based maintenance to condition-based maintenance – the ideal for energy & utility companies.
Major benefits include less downtime and lower costs through the preventive repair, maintenance and replacement of assets.
We have seen promising results from digital twins using ML in the energy & utilities industry - improved CAPEX allocation, prolonged component lifetimes, minimized downtime, maximized efficiency and reduced operating costs.
Making the shift to condition-based maintenance
So how realizable is this ideal? If you currently operate 100% schedule-based maintenance and want to move to condition-based, high-quality process data is required, so initiate a data strategy to ensure data availability and governance. Start to think about which components are most critical, then create categories in your maintenance strategy as a basis for moving forward.
While high-quality, available and accurate data in information technology (IT) and operational technology (OT) systems is critical for long-term condition-based asset investment and maintenance, you don’t need to wait to complete your data strategy project. A pilot project can be set up to develop a digital twin for a critical area or component so that you can start to simulate different ‘what if..?’ scenarios.
If your biggest pain point is CAPEX planning, start with asset investment planning. Many companies start with the high cost of asset unavailability; this is because it’s often the most straightforward to implement using real-time data. Within six months, you can start to use ML and AI to train models and get the first results. If there are unusual patterns and deviations that your engineers don’t recognize, these signal early warnings of a failure or risk.
Proven results for energy & utilities
We have seen promising results from digital twins using ML in the energy & utilities industry - improved CAPEX allocation, prolonged component lifetimes, minimized downtime, maximized efficiency and reduced operating costs. As part of a recent implementation, a digital model was enhanced with live data and expert knowledge to create a digital twin that provided a 360° view of the assets and predicted failures. This improved the Mean Time to Response (MTTR) by 20%, reduced the operational transportation expenses by 12%, and increased the asset availability by 0.2%.
While the specifics of the asset management lifecycle might be different for a power plant, wind turbine or water station, the challenges of high-performance, resilience and cost-efficiency are the same for all. Evidence shows that the combination of digital twins with ML and AI can now make a critical difference in this complex domain.