Wind Energy

Understanding wind resource: wind measurements, long-term estimation, and vertical extrapolation

By Alberto Quintana Infante

Each stage of a wind resource study is crucial for estimating a wind farm’s energy production with the least possible uncertainty.

During the development phase of wind farms, this evaluation is essential to understand the energy potential of the site and analyze the characteristics of the wind regime in terms of speed, direction, frequency, variability, turbulence, or extreme events. These factors not only help predict production and optimize the project but also determine its economic feasibility.

Below, I’ll review the main phases of these studies, focusing on their utility and importance, as well as their impact on the final uncertainties of the project. In this first part, I’ll cover the first three, leaving the rest for a second article:

  1. ? Wind measurements
  2. ? Long-term estimation
  3. ?? Vertical extrapolation
  4. ?? Horizontal extrapolation
  5. ? Power curve and wind turbine model
  6. ?? Wake models
  7. ? Technical and operational losses
  8. ? Variability and uncertainties

All of these are fundamental to determining the AEP (Annual Energy Production) values commonly used in project financial models. The most commonly used are the P50, P75, and P90 percentiles.

P50, P75, and P90 percentiles indicate confidence levels regarding a wind farm’s energy production. In other words, they express the probability that the calculated energy value will be exceeded.

  • P50: The expected average production value (50% probability of being exceeded).
  • P75: A moderately conservative production value (75% probability of being exceeded).
  • P90: A more conservative production value for risk analysis (90% probability of being exceeded).

? 1. Wind measurements

Although the use of “virtual wind series” or mesoscale data is increasingly widespread for preliminary studies and the quality of these products is improving, it is well known in the sector that real on-site measurements are still absolutely necessary.

The questions here are always the same: When should we start measuring? How long should the campaigns last? At what heights should these measurements be taken? What sensors should we install, and how many?

Whenever possible, the key is to design a measurement campaign that aligns as closely as possible with our initial project concept (extent, number of wind turbines, turbine model). This will determine the number of measurement points, their location, the necessary equipment, and the measurement heights.

It’s also important to note that, to properly characterize the wind, measurements must be taken for at least one year (although extending this to 2–3 years is always recommended) to capture the full annual cycle and the wind’s seasonality. Starting the measurements earlier or later depends on the developer’s criteria, but having this data as soon as possible can help avoid future surprises.

The most common option in wind farm development remains the guyed lattice tower equipped with cup anemometers, wind vanes, and sensors for temperature, relative humidity, and atmospheric pressure.

A good selection of equipment, proper calibration, and installation (complying with MEASNET requirements and IEC 61400-12-1 standards) are crucial to ensuring the lowest uncertainty in the campaign.

Once the campaign is completed, it’s essential to process and clean the records to obtain high-quality statistics: frequency distribution, Weibull distribution, wind roses, seasonal characteristics, turbulence intensity, among others.

Moreover, daily monitoring of the measurements becomes a key process to ensure high availability of the data and to act quickly when issues or errors are detected.

All the above will shape the course of the resource study, as the measurement campaign forms the foundation not only for energy estimation, which we’ll discuss later, but also for validation by the turbine manufacturer for site suitability.

A site suitability study is an evaluation conducted by wind turbine manufacturers to ensure their turbines are suitable and safe to operate at a specific location, analyzing aspects such as wind speed and turbulence, adverse weather conditions, terrain characteristics, and structural durability.

? Long-term estimation

Continuing with the questions: Was our measurement campaign representative of the long term? It could happen that, due to random meteorological variability, the measurements were taken during an especially good or bad year.

To answer this, the temporal variability of the wind regime at our site must be analyzed.

This is where reanalysis datasets (ERA5, MERRA2, CFSR, etc.) come into play. These global databases, with records spanning up to 40 years, serve as a reference and allow for identifying trends, temporal cycles, or potential anomalies in the campaign by performing a temporal extrapolation (MCP—Measure-Correlate-Predict processes).

This way, we can determine the expected long-term wind speed and any necessary corrections to the original measurements.

Temporal extrapolation also allows for generating production series of up to 30 years, which are essential when evaluating financial models or conducting studies such as complementarities with other generation sources (e.g., hybridization with solar PV).

?? Vertical extrapolation

If we’ve designed a good measurement campaign, we’ll have data at different heights to characterize the vertical wind profile at our site. We might even have measurements at the same hub height as the wind turbine model to be installed.

But what happens beyond the maximum height at which we’ve measured? What if the hub height is greater?

Source | ZXLIDAR

Different models exist to perform vertical extrapolation of the measurements. However, it’s always recommended to first evaluate the different behaviors of the vertical profile depending on wind direction, time of day, and temporal distribution, to detect trends and ensure these results align with terrain characteristics or the presence of potential obstacles.

How can we reduce the uncertainty associated with this extrapolation? There are two main approaches:

  1. Reach measurement heights of at least 2/3 of the hub height being considered.
  2. If there are doubts, install a remote sensing device (SODAR/LIDAR) in parallel for a sufficient period, as these devices can measure up to 300 meters in height.

In the second part, we’ll continue analyzing the remaining aspects:

  • ?? Horizontal extrapolation
  • ? Power curve and wind turbine model
  • ?? Wake models
  • ? Technical and operational losses
  • ? Variability and uncertainties

After analyzing wind measurements, long-term estimation, and vertical extrapolation in the first part, we now move on to:

  • ?? Horizontal extrapolation
  • ? Power curve and wind turbine model
  • ?? Wake models
  • ? Technical and operational losses
  • ? Variability and uncertainties

?? Horizontal extrapolation

We’ve already measured at least one point of the site (where the measurement tower or towers have been installed), but how do we determine the wind speed at every turbine location in the wind farm?

This is another critical step, and choosing the right modeling software to generate the wind map for the area is essential. It’s important to input terrain information (orography, roughness, obstacles, etc.) with as much detail as possible. These maps will allow for the optimization of the turbine layout and provide insights into areas with energy potential.

Analyzing terrain complexity is also crucial since it may require specific models (e.g., CFD) capable of characterizing highly complex sites. However, modeling doesn’t always hit the mark. To improve confidence in the results, additional measurement towers or campaigns with SODAR/LIDAR devices running parallel to the main tower may be needed, especially if the project area is extensive.

? Power curve and wind turbine model

At this stage, the key question is: how do we transform wind measurements into energy production? This is done using the power curve provided by the turbine manufacturer.

Two key points: adapting the curve to the site’s air density (which requires measuring variables like temperature and atmospheric pressure) and selecting the appropriate wind turbine model. This involves assessing factors such as the turbine’s Class and Subclass, which determine the suitability of a turbine for a specific site.

Permitting constraints can sometimes complicate this process, as limitations on tip height, rotor size, or turbine spacing can reduce flexibility when switching models or manufacturers. Additionally, reprocessing permits for a turbine change can cause unacceptable delays for the project.

?? Wake models

The effects of operating turbines on the wind flow are complex. How does the airflow redistribute after passing through the rotor?

It’s well-established that closely spaced turbines affect each other. Analyzing these phenomena is no easy task.

Various numerical models aim to calculate the speed reduction caused by this effect, known as the wake effect. Some names to keep in mind include Jensen, PARK, or Large Eddy Simulation (LES) models, which are also used to simulate turbulence generated by wakes. Each has its own limitations, advantages, and drawbacks.

? Technical and operational losses

Production estimates must consider potential losses to deliver realistic results. What types of losses can we encounter beyond plant availability or electrical transmission?

Losses can be grouped into three categories:

  • Turbine performance:
    • Deviation from the theoretical power curve.
    • Hysteresis at high wind speeds.
    • Inclined flow or rotor misalignment with actual wind direction.
  • Environmental factors:
    • Blade degradation from dirt or ice formation.
    • Operational temperature ranges.
    • Derating.
  • Operational restrictions:
    • Temporary noise constraints.
    • Shadow flicker.
    • Birds or bats.
    • Wind sector management.
    • Power limitations or curtailments.

? Variability and uncertainties

All the variables and calculations discussed in both articles involve uncertainties that must be correctly assessed and quantified. Since the ultimate goal is to understand energy production and make investment decisions, all the analyses above are essential to minimize project risk.

How are uncertainties in calculations evaluated?

Even if the P50 estimate is perfectly defined, wind farm energy production varies yearly due to factors like the natural variability of the wind regime, system availability changes, or fluctuations in technical losses as described earlier.

Net energy uncertainty provides a metric for determining the risk of a project’s production decreasing or increasing over a specific period (typically 1 year, 10 years, or the project’s lifetime). To determine this, uncertainties related to wind speed data and models, energy loss factors, and interannual production variability must be considered.

These factors, combined with the site’s specific sensitivity to wind speed, are used to generate a probability distribution for the project’s annual net energy production. Probabilistic stochastic processes are the most complex methodology but offer the best characterization of randomness in each variable.

All these evaluations and measurement campaigns should aim to improve estimation quality and minimize overall project uncertainty, which is critical for justifying financing.

Park sensitivity determines how production responds to changes in wind speed. Similarly, wind speed uncertainties translate into energy uncertainties using this parameter, which depends mainly on wind speed distribution and the turbine’s power curve. For example, with a sensitivity of 2, a 2.0% reduction in wind speed would result in a 4.0% reduction in net energy production.

? Conclusions

To conclude, it’s worth emphasizing that each stage is important and impacts the results and the final uncertainty assessment, with the quality of the measurement campaign being the key starting point for the studies.

Each section could undoubtedly warrant more than one article, so if this has been of interest, we can dive deeper into any of them.

This content was originally published in Windletter, a weekly newsletter with the latest news from the wind energy sector. You can subscribe for free at this link.