Would it make sense to design wind projects based on the price of energy?

Traditionally, the goal of wind farm design is to maximise annual energy production (AEP) within the safety margins of the turbine. But today, with electricity prices differing depending on the time of day, it may make more sense to maximise expected revenues even at the cost of reducing output.

 

Let’s start with the basics

The AEP (Annual Energy Production) of a given turbine is given by its power output as a function of wind speed (power curve) combined with the wind distribution over the year.

 

The real power curve of a turbine is constructed from the cloud of power measurements over a period as a function of the measured wind speed at the hub height.

 

 

 

As can be seen in the graph above, the data cloud forms what is called the Real or Measured Power Curve and should be as close as possible to the theoretical power curve defined by the manufacturer in the wind turbine specifications.

 

Regarding the wind distribution, first the average wind speed is measured (normally at a ten-minute frequency) and this provides a data set where the great variability of the wind can be appreciated.

 

 

To make the wind data more manageable, they are grouped by speeds indicating the number of hours that each speed occurs throughout the year, so that the total sum is the 8760 hours of the year. This is called the wind distribution. In fact, any measured wind distribution can be represented by a Weibull distribution. This statistical approximation is very convenient and useful to generate wind distributions from an average speed and a factor “k”.

 

 

As can be seen in the above graph, the way to maximise the AEP is to increase the power at wind speeds with many hours. At sites with low or medium wind speeds (which are the majority), most of the wind hours will be at low values, so the most common way is to increase the rotor and thus move the curve to the left, capturing many more wind hours.

 

 

But what happens when we introduce the electricity price variable?

The strategy of maximising annual production makes sense when each kWh generated has the same value, as was the case with feed-in tariffs or now with PPAs. But what happens if a project sells its production on the wholesale electricity market?

 

As can be seen in the data for any given day in the Spanish daily market, the price differences between peak hours (around 21h) and off-peak hours (around 14h) are more than €100/MWh. As renewables (and in particular solar) continue to increase their penetration, and pending the massive deployment of storage, we will see these differences grow. This raises the question of trying to introduce this variable in order to maximise revenue rather than maximise production.

 

 

In fact, wind turbine manufacturers are already aware of this and try to design their products with this factor in mind. A good example is Nordex and one of their latest promotional videos, where they explain in a very didactic way how their N175 large rotor turbine produces more when electricity is more expensive, thus generating more value.

 

 

How do we model this new variable in a simple way?

Just as the Weibull distribution is a tool that makes wind distributions tractable, in order to introduce the market price distribution into the modelling of wind farms, we should create a simple statistical model.

 

And this is exactly what a team from DTU (Technical University of Denmark) led by Andreas Bechmann has done and presented at WindEurope 2024 in Bilbao with a poster and a presentation.

 

 

What the study does is to create a statistical model for the price curve of a given region or country. This new model is introduced into the traditional AEP calculation together with the Power Curve and the wind distribution to obtain the AEV or Annual Energy Value.

 

 

What are the practical implications of the AEV?

 

• The most obvious is that it can drive the design of turbines for “very low winds”. This topic was explained very well in our preferred wind newsletter, Windletter by Sergio Fdez Munguía, back in issue 3, and is worth revisiting.

 

• It would also be possible to modify the control of the installed turbines to increase power at low speeds but with high prices, and yet reduce power or even stop at times with very low prices. Something similar is already being done with “voluntary curtailment”, but the idea would be to do it in a more sophisticated way.

 

• In line with the first point, perhaps concepts such as Vin and Vout could be rethought, perhaps by making them dynamic as a function of the price curve.

 

Chinese manufacturers are already marketing turbines for very low winds with very large rotors. In fact, Goldwind announced at the last China Wind Power 2024 fair a model with a 204m rotor and 4 MW power, which results in an incredible 122 W/m2, the lowest power density (by far) in the market.

 

Undoubtedly, introducing the market price variable in the design phase of the turbine can increase the value generated for the operator. However, given the variability of market prices 30 years from now, flexible strategies that can be adapted over time would be advisable.