Modelling turbulence

We cannot see air - we can just sense its movement. But stuck as we are at the bottom of an ocean of air, our perception of how air moves is limited by our experience. We know that sometimes the wind blows and that sometimes it is gusty and at other times it is more constant.

But we cannot see how air often rolls and eddies in turbulence - that random variation in instantaneous wind speeds, occurring at different scales and different speeds.

We can't see turbulence - but before wind turbines can be effectively sited, the turbulence characteristics of the location must be understood.

Windlab is one of the world’s leaders in turning science into technology for the wind industry.Windlab’s Chief Technology Officer, Dr Keith Ayotte, spends much of his time working with scientists from the worlds leading research institutes, creating and distilling the complex output from sophisticated turbulence simulations and using that information to create advanced tools for the prediction of turbulence over wind farms sites.Making the invisible visible.

Here's the simple version of what it is all about.

Turbulence and Wind Turbines

Wind turbines are sited to take advantage of high average wind speeds. But knowing only wind speeds is insufficient - without also quantifying the degree of turbulence experienced at a location, the story is incomplete. In fact, when turbulence for a location is analyzed, a site that initially may have looked economically viable for wind turbines can become a site to immediately remove from the list.

A high degree of turbulence has several negative impacts on wind turbines.

First, the turbine has to be designed to cope with the peak loads it will experience. If turbulence is high, the maximum loads on the blades, gearbox, generator and tower will also be higher. Typically, turbines designed to cope with high turbulence loads cost more and yield less energy.

Second, turbulent wind gusts occur at random frequencies and speeds. These rapidly changing loads have the potential to excite resonances (large vibrations) in the turbine and its supporting tower. Stiffer and stronger structures are needed to combat this.

Third, cyclically varying loads can cause fatigue problems in turbine structures. For example, if the blades are constantly flexed by turbulent wind gusts, they will need to be designed to safely withstand a greater number of these movements before failure.

Finally, rotor blades moving through turbulent air meet constantly varying conditions. In turbulent conditions the local wind environment is randomly changing in a way for which each blade cannot compensate. For example, a change of apparent wind direction or speed may cause localized stalling of the blade, so reducing the amount of energy being yielded by the generator.

Turbulence and Topography

So if a high degree of turbulence has negative impact on wind turbines, why not simply site them where there is no turbulence? That's rather harder than it first sounds!

Except at dawn on a still day, all air that a turbine is likely to meet is turbulent. There are numerous causes of this, but we can see most of what is going on by looking at two things.

 

First, as air moves over hills it speeds up. That is good for wind turbines, and is the reason that wind turbines are usually sited on ridge and hill-tops. The air increases in speed because it is subjected to what is called a positive pressure gradient. That is, the push behind is greater than the resistance ahead, and so the air moves faster.

However, on the other side of the hill, the opposite occurs - the air experiences a negative pressure gradient. As a result, some of the air slows and in fact can actually reverse direction.

If air high above the hill is moving in one direction, and air down low in the lee of the hill is moving in the other, swirls and eddies of air are formed ... turbulence.

If the wind turbines are sited downstream of hills, this topography-caused turbulence has the potential to impact on the turbines.

There are also other sources of turbulence.  To understand this we need to consider how high we can go in the air before we escape the influence of the ground.

Imagine a wind blowing across a flat plain. If we measure wind speed just a few tenths of a millimeter above the ground we'll find that the air is not actually moving at all - it's held stationary by its frictional contact with the surface.

But move the measuring instrument two meters above the ground and you'll find that here the air is moving.  Go higher again and the wind speed will be even faster.

For the air at ground level to be stopped and yet, as we go higher, to be moving faster and faster, layers of air must be sliding over one another. (This is called shear.) Air that is in shear is more easily 'tripped' to become turbulent.

(The air that is being affected by the presence of the earth's surface - for example by showing an increase in speed with height - is called the boundary layer.)

And in addition to shear within the boundary layer, there's another potential cause of instability. Air in contact with the warm ground surface is heated. As this air warms, it becomes lighter - and rises. This uplift is called convection.  Convection (that we can see in tall cumulus clouds) can also lead to turbulence.

What with topographical turbulence and boundary layer turbulence, it becomes effectively impossible to site wind turbines in areas of zero turbulence!

But if we can predict the amount of energy in turbulence that occurs at a potential wind turbine location, a better decision can be made.

Predicting Turbulence

We've said that turbulence comprises the random changes in instantaneous wind speeds. It takes place in all three dimensions and can occur over time spans as short as tenths of seconds, right through to tens of minutes. The latter is fine - the wind turbine just 'sees' this turbulence as a relatively slow change in wind speed and so its control systems can make the most of this potential energy yield. But as we've seen, turbulence occurring on a shorter time scale can have major negative impacts on wind turbines.

If we are to model and predict turbulence, we need to know how to quantify it. Three different measurements can be made: the energy contained within the turbulence (that is, the mass and speed of air movements); the size of a region in the lee of a topographic feature producing the turbulence; and how far downstream the turbulence exists before it naturally abates.

That's how to quantify it - but what about modeling it?

Three different levels of complexity of turbulence modeling can be applied. The simplest model suggests that as air passes over a hill, separation occurs in the lee of the hill. This separation bubble changes the effective shape of the hill - and this idea can be fed back into how the flow is accelerated over the hill.  Although relatively inexpensive in terms of computer power and required skill, this method often poorly predicts turbulence levels.

At the other extreme, massive amounts of computational power can be applied to simulate individual eddies in the flow. Windlab currently works with scientists at the National Center for Atmospheric Research in Boulder, Colorado undertaking such calculations.  Unfortunately the latter approach has two problems with use for siting wind turbines: the modelling requires the aforementioned enormous computer resources and is best only applied to individual, idealized hills rather than complex real-world terrain.  In short, for now it is still at the cutting edge of atmospheric science.

Windlab has developed a model that sits midway between these two types of modeling.  The model calculates the pressure changes in the flow as it accelerates and decelerates over topographic features.  It then uses an understanding of how turbulence is created and transported downwind, gained from a number of years of research of turbulent flow over hills, within algorithms to predict levels of turbulence.

Hills and Vegetation

Landscapes on which wind turbines can be placed vary in innumerable factors but the two key ones for predicting turbulence are the steepness of the hill and its type of vegetation cover.

The steeper the hill, the greater the amount of turbulence generated behind it. In terms of vegetation, the degree of turbulence that is created is influenced by the canopy height, its variability in height and the density of the vegetation.

 

Examples

 

So what do the outputs of this turbulence modelling look like?

Using the most complex models currently available (the ones that are too large and expensive to effectively apply to wind turbine siting), this diagram shows turbulence occurring in the flow over a smooth hill. The wind is blowing from left to right and the 'hotter' the color, the higher the instantaneous velocity of the flow in the turbulence. Note the reverse flow (darkest blue) that occurs in the lee of the hill.

 

This diagram shows the simulation of flow over a 20 meter high vegetation canopy. The rapid change in wind speed with height results in a strong shearing action.

 

This diagram shows the simulated flows around a hill with a vegetation canopy.  The complexity of flows near ground level, including reverse flows, gives a strong indication of the potential difficulties in siting wind turbines - especially on adjacent hills!

 

In these complex modelling results, this pair of diagrams shows the simulated flows around a hill, with and without a vegetation canopy.

Finally these two images show the results from eddy resolving (LES) calculations of flow over a shallow and a steep hill.These are results are from Windlab’s most recent collaboration with scientists at the National Center for Atmospheric Research (NCAR) in Boulder , Colorado.These calculations were done using 4096 processors of a massively parallel computing cluster and represent a Herculean effort at the very leading edge of science.

Modeling1Modeling1

Modeling2Modeling2

And the Windlab model? Many of the results (and the detailed calculations behind them) remain confidential, but this map of an actual potential wind farm location shows the output of the turbulence model, for a single direction.

Modeling3Modeling3

Conclusion

Predicting turbulence at potential wind turbine locations is a critical element in finding economically viable locations.  Using cutting-edge technology and sophisticated modeling, Windlab is able to remove much of the uncertainty that previously surrounded the impact of turbulence on locating the best sites for wind turbines.

Turbulence modeling is one reason Windlab leads the industry in locating viable wind turbine sites.