In this blog we will discuss expert insights on evolution of dgital twin from 3E co-founder and CTO Werner Coppye, and editor in Chief of Foresight magazine David Weston. We will explore their thoughts about how digital twins have developed over recent years, and how new technologies like artificial intelligence are changing the way digital twins are used.
Why are digital twins important in renewable energy?
A ‘digital twin’, which is a concept that originated in the early 2000s, is a digital representation of a physical system. This could be a car, airplane, manufacturing plant, or telecom system for example. Digital twins serve various purposes throughout the lifecycle of the physical system they represent, from design and construction through to operation and decommissioning. These purposes might include actions such assimulation, integration, testing, monitoring, or maintenance.
It is this value that digital twins provide across the lifecycle of a system that makes them so important for the renewables industry. Renewable energy assets typically undergo years of development, design, engineering, construction and commissioning. These assets are then expected to function for 20 years or more before decommissioning.
Throughout this time, digital twins offer a cohesive, data-based digital model, that can generate valuable insights which benefit the entire lifecycle of the asset. In this way digital twins allow for the translation of large volumes of data into insights, and those insights help to maximise the value and longevity of renewable energy assets.
The evolution of digital twin technology has transformed the landscape of operations and maintenance
For operations and maintenance within renewable sites, the evolution of digital twin technology has been transformational.
In relation to solar assets for example, initially, digital twins were characterised by fragmented datasets. Data relating to geographical information, PVSYST design files, CAD data, business plans, as-built files, monitoring data, and maintenance records was available, but was not linked together. This lack of integration among datasets led to significant workflow inefficiencies, as well as information loss.
Then, with the emergence of drone technology, drone imaging platforms began to build the links between different datasets. Initially between the site thermography and the electrical system design. Developments like these have since made solar asset management much more efficient and resolute.
More recently, drone imaging has also gained traction in construction management. Drone inspections can be used to compare the deviations between the geographical data of the design against the actual construction. This creates much more accurate and insightful as-built files, that are crucial for efficiency in the operational phase of an asset.
PVCase and similar design tools have improved digital twins by combining CAD data with electrical designs and yield simulations. In another blog, we discussed how digital twins, when used from start to finish in renewable projects, give an advantage in optimizing performance.
Different methods of digital twin development
Black box v Whitebox
In the realm of digital twin development, there are two primary methods: the black box and the white box approaches.
The black box method relies on statistical analysis of operational data, using machine learning to create a model of asset behavior. This method correlates actual behavior with various operational conditions, offering robust and reliable results due to advancements in artificial intelligence.
On the other hand, the white box method is based on physics. It employs simulation models during the design and engineering phase to predict asset behavior under real operating conditions. These models typically incorporate physical formulas defining relevant processes.
Combining both methods can create a more precise digital twin, known as grey-box modeling. Such grey box modeling leads to hybrid digital twin that is today the most optimal approach to performance optimization. Want to see why? check our deep-dive blog on hybrid digital twin here.
Utilizing a digital twin that incorporates physics, data, and AI yields numerous benefits, especially in renewable energy
At 3E, we advocate for both white-box (physics-based) and grey-box (hybrid) modeling approaches for several key reasons:
Firstly, a pure black box method necessitates a training period, delaying the assessment of asset performance. In solar, this could mean waiting at least two seasons or six months for insights.
Secondly, the statistical approach views a training period as a reference, assuming the plant operates as expected, often not reflecting real-world scenarios and potentially overlooking performance losses.
Furthermore, the black box method lacks the versatility of a true digital twin, focusing solely on operational phases. In contrast, white box and grey box methods encompass all lifecycle phases.
The physics-based digital twin method significantly enhances decision-making
This approach furnishes detailed time-series data on asset behavior at different component levels, enabling robust identification of failure root causes in operational systems when combined with high-quality monitoring data and AI techniques. Compared to machine learning, the physics-based method improves early fault detection, identifies more loss categories, and accurately quantifies losses. Consequently, this:
- Increases transparency on asset performance among operational stakeholders.
- Facilitates more efficient decision-making and reporting processes.
- Allows for proactive corrective actions, leading to significant long-term performance improvements.
Discover how a physics-based digital twin, developed from the project's inception, can enhance accurate performance analysis in our related blog post.
What is generative AI and how is it relevant to renewable energy?
What is generative AI?
Generative AI is a form of artificial intelligence. It uses advanced machine learning models to create original content in the form of text, images, or code. It is expected to revolutionise various sectors in the near future. Much like the industrial revolution and robotisation transformed blue-collar work over the past century.
What role might generative AI play in renewable energy operations?
Generative AI is set to heavily impact several aspects of renewable energy operations, spanning from research and development to support services, financial services, and software development.
For EPC contractors, it will streamline the configuration of digital twins by leveraging all available data sources. For O&M service providers it will automate the dispatching of interventions. For technical asset managers, it will improve stakeholder reporting capabilities. While PPA parties will also benefit from more accurate reporting.
Overall, the impact of generative AI should not be underestimated. It is crucial that we prepare our organisations and our industry for this revolution.
One great way to prepare is to learn as much about the topic as possible. If you’d like to learn more, why not give the full podcast a listen.
