Jan 24, 2020 |
Read time: 5 mins
With continuing pressure on wind project margins throughout the entire value chain, it becomes increasingly important to apply a holistic view of the design and micro-siting of wind farm projects. A diligent focus on optimising annual energy production or net capacity factor will not be sufficient to ensure a project providing the highest investor value (risk-adjusted return). At a minimum, there is a risk that potential investor value is being ignored in the wind farm design stage, which can lead to a misguided conclusion on what would constitute an optimal wind farm design.
Previously, we have discussed the possible value of energy advantage for high rotor to generator ratio turbines, such as the Vestas EnVentus 138, 3.0 MW. Another optimisation possibility, related to the WTG type, is its potential impact on production uncertainty. Specifically, the turbine type and micro-siting can have a significant impact on the inter-annual variability, as we will demonstrate below.
A reduction in inter-annual variability will contribute to reduce uncertainty in the production estimate and allow the wind project to carry a higher debt burden, thereby increasing Return on Equity (RoE). Further, as projects increasingly rely on non-subsidised PPAs set under market conditions, the less variable the production, the higher the value of the PPA will become.
Another and normally undervalued point is the impact a reduced inter-annual production variability has on improving the risk-adjusted return for institutional investors approaching renewable projects, as an alternative to the diminishing return of fixed income assets, such as bonds. To have wind projects approximate fixed-income investments, it thus becomes an important wind farm design parameter to minimise underlying variability.
To demonstrate the potential of selecting a wind turbine type for reduced uncertainty, we simulated the same wind farm layout with two significantly different rotor-to-generator ratio turbines. Our initial hypothesis was, everything else being equal, that the higher the rotor-generator ratio of the turbine, the more stable the annual AEP will be. As illustrated in the figure below, we see such an effect when we model production across an 18-year period. Despite having the same turbine locations, and hence free wind speed, we can significantly reduce the inter-annual production variability from the wind farm. For Scenario A, the high rotor to generator ratio [25 m2 of swept area per kW rated capacity], the standard deviation of the inter-annual AEP was 3.2%, while for the lower rotor to generator ratio, Scenario B [14 m2 of swept area per kW rated capacity], the same standard deviation was 5.6%. In other words, in Scenario B, the interannual production variability was 75% higher than in Scenario A.
The turbine type has a bearing on more than the interannual variability alone. It also influences key elements in the financial model: annual energy production as well as turbine supply cost, installation cost, BoP cost and operational expenditures. In our example, the combined effect of these differences results in a higher ROI for scenario B. In our standard model; Scenario B provides a return which is 90 bps higher than in Scenario A.
Scenario A and Scenario B comparison
|Scenario A||Scenario B|
|Rotor-to-generator ratio||25 [m2/kW]||14 [m2/kW]|
|Inter-annual production variability||3.20%||5.60%|
|Other uncertainties related to energy yield estimates ||7.20%||7.20%|
|Combined energy yield uncertainty||7.87%||9,10%|
However, the question remains if this higher return compensates for the higher volatility associated with Scenario B. The Sharpe ratio a measure from the field of investment and portfolio theory has gained wide recognition as the basis for comparing investments options with different risk profiles. It is calculated by subtracting the risk-free rate from the expected return of the investment and dividing that excess expected return by the risk associated with the investment, expressed as the standard deviation.
In the above calculations, we have excluded the risk-free rate, since, in current market conditions, it is considered to be very close to zero, at least in a European investment climate. The Sharpe Ratio assumes that the return of an investment is normally distributed around an expected mean return, which aligns well with standard assumptions made for energy yield related uncertainties.
By comparing the Sharpe ratio for the two scenarios, it is evident that the lower volatility of Scenario A offers a more attractive risk-adjusted return compared to Scenario B. With the increased interest from institutional investors assessing renewable projects as an alternative to fixed income assets, a strategic approach to optimisation for a risk-adjusted return should become an increasingly important agenda for project developers.
Above, we demonstrated a usually overlooked element of turbine type selection on wind farm design optimisation. However, this is not the only lever for project developers intending to take a deliberate approach to improve the risk-adjusted return of their projects. Other levers include assessing the uncertainty imposed by on-site climate measurements as well as both horizontal and vertical extrapolation wind flow models.
Assessing the uncertainty imposed from horizontal extrapolation of the wind flow modelling from the met mast to the turbine location makes it possible to apply the same risk adjustment at the turbine level. In other words, when micro-siting turbines, it might be worth choosing turbine locations with slightly lower AEP, if that turbine location imposes a smaller uncertainty. Consider the following example with the evaluation of two alternative turbine locations, I and II. WTG location I is within 800 meters of the met mast installed as part of the wind measurement campaign, while the latter turbine location is located 9.5 kilometers from the measurement mast.
The wind flow model conducted based on the wind measurement campaign suggests a slightly higher free wind speed at WTG location II compared to I. Resulting in a higher AEP and net capacity factor for location II. In most automated layout optimisation tools in standard software applied across the industry, such as WindPro or Open Wind, location II would be favored over location I. However, location II would carry a higher uncertainty due to the longer distance from the wind measurement point (met mast) to the turbine location. As wind flow models and surface maps are not perfect, this horizontal extrapolation introduces an uncertainty in the production estimates. And the longer the distance, the more uncertainty there is.
In conclusion, WTG location II may produce the highest AEP and thereby ROI, but WTG location I would offer a better risk-adjusted ROI for the project. Micro-siting of turbines usually is more complex than choosing between two alternative turbine locations, but the example seeks to demonstrate the need to integrate the consideration of project uncertainty into the wind farm design stage. Here it should also be remembered that the developer may have more profitable alternatives for improving the risk-adjusted return by reducing the uncertainty through additional investments in the wind measurement campaign.