To project overview

For Wind Turbines, Turbulence Is the Work of the Devil

Internationalisation Fellowships | 17/06/2016

Wind turbines convert kinetic energy of the wind into electricity. The more it blows - the more is produced. We all know, however, that the wind is rarely steady; a constant wind speed from a single wind direction rarely occurs. Instead it fluctuates in a chaotic and seemingly random way. These fluctuations are known as turbulence.

By Associate Professor, PhD Jacob Berg, Technical University of Denmark

Wind turbines convert kinetic energy of the wind into electricity. The more it blows - the more is produced. We all know, however, that the wind is rarely steady; a constant wind speed from a single wind direction rarely occurs. 

Instead it fluctuates in a chaotic and seemingly random way. These fluctuations are known as turbulence and still continue to puzzle scientists around the world. Many still consider turbulence as the biggest unsolved problem in the world of classical physics. Therefore, it might not surprise you that this article does not contain a solution to the problem nor does it try to solve it.

Besides being a scientific challenge, turbulence also constitutes one of the main pillars which modern wind turbines are designed after. The turbulence in the atmosphere close to the ground exerts forces on the wind turbines (called loads) which over time will limit its lifetime and therefore decrease the financial potential. With regard to wind turbines one could say that turbulence is the work of the devil.

In my work, I study atmospheric turbulence in order to address its qualities and quantify the levels encountered by wind turbines. I use a blend of atmospheric observations together with numerical Large Eddy Simulation (LES), which is computer models based on the famous Navier-Stokes equation and require extensive computer power. 

LES was the focus of my one and a half year stay at the National Center for Atmospheric Research supported by the Carlsberg Foundation and hence the main subject for this short overview article.

Wind energy on land (onshore) comprises one of the cheapest electricity forms today and in the next 20 years [1]. From a social-economic point of view, it therefore makes a lot of sense to study turbulence and in this context the turbulence of the atmosphere. The Carlsberg Foundation supports among others basic research within natural sciences i.e. the physics of turbulence. 

This subject also meet one of the UN Sustainable Development  Goals e.g. affordable and clean energy. This research project therefore manages to link basic research together with sustainability and “green tech” industrial development.

The Need for Turbulent Research in Wind Energy

The interaction between the Earth surface and the atmosphere above it generates a layer of turbulent motion, which depending on time and location can vary between a couple of kilometres and down to tens of metres. In this layer, the atmospheric boundary layer, we put turbines and expect them to operate for at least 20 years.

Various engineering models of turbulence based on a mathematical/statistical approach can tell us beforehand whether a given site can be used to host a turbine. By performing such a site assessment study most of the ‘bad locations’ are filtered away, and with perfect turbulence models we would in principle be able to predict the damage done over time from the turbulence to the turbine.

However, the problem is that no perfect turbulence model exists. Each model is built on a set of assumptions about mainly the external conditions. These could for example be ‘no heat exchange with the Earth’s surface’, ‘only flat terrain around the turbine’ or ‘no non-Gaussian turbulence with very gusty events embedded'. 

Figure 1: Sketch of LES methodology

One of the most advanced tool to run on a super-computer is a Large Eddy Simulation (LES) model, which is directly based on the (spatial filtered) Navier-Stokes equation (Figure 1). For all practical matters regarding siting of wind turbines LES is today too difficult to use. However, in a research context,  it is exactly the tool we need. 

Using the tool i.e. Being a big box of the atmosphere, we have the possibility to ask relevant questions and study them in an isolated manner. For example, we know that the atmospheric turbulence is not Gaussian as assumed in most turbulence models (Figure 2), but does the turbine care at all? Does this effect the calculation of loads for the turbine? Most likely not…[2]

Figure 2: To the right we look at a horizontal snapshot of the vertical velocity component of LES generated atmospheric turbulence - non-Gaussian by nature. Notice the swirly motion viz a Michelangelo painting. The same picture (same second order statistics, to be precise) to the right is Gaussian and clearly different in nature from its left counterpart.

LES and the Study of the Atmospheric Boundary Layer over Complex Terrain

For turbines on land, we often find that the landscape in which they are embedded generates very complex flow phenomena which not only increase the turbulence level beyond any turbine design criterion but also lower the power production and hence all parts of the financial aspects of a wind turbine project. 

Wind lidars, which are remote sensing instruments based on laser doppler techniques are being developed, and can monitor the atmospheric boundary layer above large areas of interest in a wind turbine context. 

The lidar system, however, still has its flaws [3], and LES therefore provides an alternative look into the flow occurring over complex terrain features such as hills, escarpments, forests etc. [4]. Especially flow over hills are well studied since the speed-up on top of the hill due to the compression of streamlines can be utilised by a wind turbine to produce more power (Figure 3). 

However, often the picture is not that simple and nothing is gained. On the contrary the opposite happens, kinetic energy is extracted from the mean wind field and converted to fields of fluctuating wind (turbulence) which ultimately will give rise to higher loads and less power production.

Figure 3: Double hill [5]. Snapshot of the streamwise velocity calculated from in LES [4]. Legends gives velocity in metres / second. Large recirculation zones observed behind the hill tops with huge levels of turbulence kinetic energy (not shown in this figure). Large speed-up is observed over the second hill top.

Time Spent Wisely at the National Center for Atmospheric Research

With the two Internationalisation Fellowships generously granted by the Carlsberg Foundation, I got the chance to work closely together with some of the best in the field of LES of atmospheric turbulence, namely Dr Edward (Ned) G. Patton and Dr Peter P. Sullivan at the National Center for Atmospheric Research (NCAR) in Boulder, Colorado. 

The happy memories of my years being a master and PhD. student came back to me through the very inspiring discussions and lively scientific environment present at NCAR. Bringing my new knowledge back to DTU Wind Energy has been both fruitful and inspiring. LES have until now been considered an overly expensive and not very useful tool in wind energy research. With support from the Carlsberg Foundation I believe this is about to change.


  1. EIA: Annual energy outlook 2015 with projections to 2040. Tech. Rep. DOE/EIA-0383(2015), US Energy Information Administration (2015).

  2. Berg et al.: Gaussian vs non-Gaussian turbulence: impact on wind turbine loads (2016), Wind Energy DOI: 10.1002/we.1963.

  3. Berg et al.: Addressing Spatial Variability of Surface-Layer Wind with Long-Range WindScanners (2015), J. Atmos. and Ocean Tech., 32, p. 518-527.

  4. Berg, Sullivan and Patton: Large Eddy Simulation of flow in complex terrain: Challenge with a pseudo spectral model (2016), Euromech Colloquium 576: Wind farms in complex terrain, Stockholm, Sweden.

  5. New European Wind Atlas, Ref. [2]-[4] are all produced / partly produced while at NCAR and hence supported by the Carlsberg Foundation.