A New Paradigm in Atmospheric Simulation

Thanks to the enormous engineering successes of recent decades, wind energy is now expected to be the central pillar of the energy transition. To continue the success story, however, several grand challenges still need to be met. On the one hand, the challenges are all related to increasing scale: of wind turbines themselves, of clusters of wind farms and of the share of wind energy in the electricity system. On the other hand, the challenges stem from our limited understanding of the atmosphere: of blade aerodynamics, wake effects, interactions between the wind farm and the atmosphere, and running an entire energy system based on the weather.

 

 

Large-eddy simulation (LES) is a technique for dynamic simulations of fluids with a two-faced reputation: it is very accurate but is computationally too demanding. As a consequence, it was used mostly in academia and by a handful of industry experts. But advances in high-performance computing using graphics processing units have now made atmospheric modelling with LES so powerful and realistic that it is in a position to add value in virtually all parts of the wind energy supply chain: from blade design to wind farm control and from installation to trading. The key to this wide applicability is this: because LES stays so close to the laws of physics, it provides a unified approach that can capture a wide range of scales, from a resolution of several metres that captures flows around, and loads on, the turbine to domains of several hundreds of kilometres capturing the interactions between clusters of wind farms and the atmosphere. When the industry fully embraces these new simulation techniques, some of the grand challenges wind energy faces today can be tackled efficiently. In this article we highlight some of the issues at different scales and illustrate how atmospheric LES can be used to address them.

 

The Micro-Scale
Wind turbines face huge forces over their lifetime. These loads are the result of a complex interplay between moving parts, controls and material properties, but in the end it is always the turbulent wind field driving the forces. Aeroelastic models compute loadings on a turbine, and they are fed with a turbulent wind field. So far, the community still uses synthetic turbulence as input for such load calculations, but these techniques do not capture the wonderfully complex turbulent structures present in the real atmosphere. Even a wind tunnel does not do so, because atmospheric turbulence looks very different than what a flow in a wind tunnel looks like: thermal effects, clouds and land–atmosphere interactions, to name a few, are not present in the laboratory.

 

In the offshore environment, loadings are even more complex due to forces acting both underneath and above the water level. Understanding how waves and currents interact with the atmosphere and how all this influences the turbine requires a next step in modelling. High-resolution, turbulence-resolving models, coupled to a wave and ocean model, have the ability to provide physically consistent wind and wave fields. For floating wind turbines, the relevance of such an approach is even more evident.

 

Wind turbine design could thus benefit in many ways from the realistic turbulence fields produced by modern LES codes. Figure 1 shows wake turbulence in an LES of two turbines represented by an actuator line model. The LES framework allows study of advanced controllers, computation of aeroelastic loads and simulation of structural dynamics in a virtual but very realistic atmospheric wind tunnel.

 

The Wind Farm and its Environment
On a larger scale, the wind farm is both influenced heavily but also interacting intimately with its environment. Understanding the environment of the wind farm is crucial for planning and operations.

 

In the planning phase, wind resource and annual energy production assessments determine not only the technical design but also have a huge influence on the financials of the farm. To put it simply: when the wind resource and wake effects are not estimated well enough, the project will not be profitable. Engineering wake models have been around for many years, but the industry has only recently realised that ignoring a subtle but persistent effect like global blockage (the blocking effect of the wind farm as a whole on the flow) comes at a price. In addition, suitable land and sea basins are getting more crowded by the day, so wind farms are starting to influence each other. More physics is thus needed to get wind, wake and production estimates at the desired accuracy.

 

During its operational life, a wind farm functions in a pretty harsh environment. Relevant aspects of the environment include not only the wind but also precipitation, icing, waves, currents and turbulence. Once built, a wind farm has to withstand it all or face the consequences – costly repairs (that, by the way, need to be planned in the right weather windows).

 

Given the often remote locations of (offshore) wind farms, asset management based on damage prevention and weather forecasts makes a lot of sense. Furthermore, despite the fact that turbines are equipped with a range of sensors and measurement systems, understanding the wind in operational wind farms can still be a daunting task. Combining atmospheric models and measurements can therefore bring a lot of insights into the performance of a farm.

 

The Energy System
There are many open questions with respect to the integration of renewable energy in the power system. Because they are driven by the weather, wind and solar energy have a degree of variability and unpredictability that make them hard to absorb into the system. Planning and operating a power system based totally on the weather can thus be quite a puzzle, to use an understatement. Highly accurate atmospheric simulations and predictions can alleviate some of the pain.

 

In the planning phase, decision makers base their investments in power systems on expected production patterns. It is crucial to understand spatio-temporal correlations of weather systems and, hence, renewable energy production to plan for the right amount of generation capacity, transmission capacity and storage. Take, for example, the expected role of green hydrogen production with wind energy: the design of the electrolysers and possible storage capacity is closely related to the production patterns of the wind farm.

 

When operating a power system, supply and demand have to be met at all times. In electricity markets, a central role is therefore assigned to the day-ahead markets, where wind energy producers have to act based on forecasts of their production. Atmospheric models in combination with machine learning algorithms are the workhorses of renewable energy forecasts and there is plenty of room for improvements. With high-precision models, forecast errors can be reduced, saving both wind farm owners and transmission system operators millions of euros per year.

 

Figure 3 shows a snapshot of a wind field over the Dutch part of the North Sea with future wind farms. This figure shows that the atmospheric feedback of a massive wind energy rollout cannot be ignored. Wake effects between wind farms, even from one country to another, will alter the wind resource. One can wonder whether the term wake effects is even appropriate, since the flow is affected so profoundly that one can almost describe this as a completely different weather situation.

 

Conclusions
Wind energy faces a number of challenges as its share in the energy system continues to grow. A useful perspective to understand the challenges is to divide them into different levels of scale: from turbine scale to wind farm scale and finally to energy system scale. Today’s most advanced atmospheric simulation techniques in combination with modern cloud computing environments have given the wind energy sector a new set of tools to tackle some of the most daunting engineering tasks. Atmospheric science and wind energy are a dream couple, of which we will hear many good things in years to come.

 

Biography of the Author
Remco Verzijlbergh is Co-founder of the Dutch weather forecasting company Whiffle. He holds an MSc in applied physics and did PhD research on renewable energy systems. He is currently Chief Technology Officer at Whiffle and Associate Professor at Delft University of Technology. His professional career focuses on research and innovation to support the energy transition.