By Dr. Edouard Leonard
Wind farm layout optimization is a key point to maximize power production. Today, the main challenge for wind farm optimization is to find the most energy production while also considering the global operating conditions with financial aspects, wake effect losses, and potential impacts on turbine loads; optimizing a wind farm is a multi-objective optimization problem. Minimizing wind farm cost leads to turbine cluster solutions, while maximizing production with wake effects leads to increasing the distance between each turbine. Special attention should be paid to considering realistic operating conditions to prevent a premature wind turbine failure. While wind farms located on complex terrain offer high potential for energy production, extreme conditions can induce varying mechanical stresses that lead to gradual degradation of components. For instance, high sheer values observed in forested areas or high inflow angle values observed in mountain areas can have strong effects on the wind turbine lifetime and should be avoided (IEC norm 61400-1 edition 4).
Multi-objective wind farm optimization cannot be solved with standard approaches like gradient methods. Instead, they require specific heuristic methods. Evolutionary algorithms like genetic algorithms are widely use in this field. They offer a greater chance of finding global solutions while considering several constraints and different criteria. Just like the process of natural selection, this approach exploits information from previous generations to create new layouts with a higher objective function.
Providing ultra-precise resource maps is an essential preliminary step for wind farm optimization. CFD methods are widely accepted as accurate method of simulating the wind flow in complex terrain. By solving the Navier-Stokes equations with a RANS (Reynolds averaged Navier-Stokes) approach, the wind flow can be modeled considering the geographical orography and roughness data for the site. Some methods are even able to propose custom forest canopy models. The influence of forest parameters (like density, height, or shape of trees) on wind shear and turbulence intensity may be known according to the location of the wind turbines with respect to the forest.
In addition to computing resource maps based on the Weibull parameters, these advanced models provide resource maps for turbulence intensity, wind shear, wind veer, flow inclination, extreme wind speed, etc. Considering these parameters is crucial for the lifetime of the wind turbine. Figure 1 illustrates the influence of the wind shear value map on a multi-objective wind farm optimization located in a forested area. A 30-wind turbine optimized layout without any constraints shows a high energy potential in Figure 1a. However, as the site is located is a forested area, the influence of the wind shear value should be considered carefully. Figure 1b shows the new optimized turbine layout with the most energy production in good agreement with the recommendation of the IEC norm 61400-1 edition 4, specifically respecting a wind shear value between 0.05 and 0.25.
Figure 1a) A 30-wind turbine optimized layout without any constraints.
Figure 1b) A 30-wind turbine optimized layout respecting a wind shear value between 0.05 and 0.25.
Financial aspects, wake effects, and constraints
Based on resource maps, computational systems have the objective of finding a turbine layout with the highest production contemplating wake effects, and reducing cost while having multiple constraints. To formulate the optimization problem with realistic conditions, the different constraints that may be due to either the site location (forbidden areas in the site) or resource maps values (shear, turbulence, inflow angle, extreme wind, etc.) must be specified. Although adding constraints is essential for wind farm optimization, it can critically affect the algorithm’s results. Hence, the importance of meta heuristic approaches.
Because reducing the cost and decreasing wake effects have opposite effects on the optimized layout, a tradeoff must be found between these variables. On one hand, financial optimization reduces the cost of electricity by reducing cable and road networks, leading to wind turbine clusters with closely spaced wind turbines. On the other hand, wind turbine clusters create significant wake effects. In multi objective wind farm optimization, the best layout can be found by optimizing a complex objective function like the levelized cost of energy (LCOE) that includes many criteria, such as annual energy production (AEP) considering wake effects, transportation costs, installation costs, number of turbines and turbine type.
Figure 2 illustrates the influence of wake effects on the LCOE optimization. In one example (as shown on figure 2a), optimizing the LCOE without considering wake effects leads to a cluster of wind turbines. In contrast, the new layout optimization (displayed on figure 2b) simulates wake effects and presents a satisfying tradeoff between wind farm cost and wake effects.
Figure 2 a) An LCOE optimization layout without considering wake effects.
Figure 2 b) An LCOE optimization layout considering wake effects.
In order to provide a suitable wind turbine layout that accounts for location and highest return on investment, advanced computational systems must consider realistic conditions as accurately as possible. Wind farm optimization is best approached and solved as a multi-objective problem that utilizes complex algorithms to find a global solution within a short time.
Dr. Edouard Leonard is a Meteodyn WFO expert. WFO is a wind farm optimization software designed by Meteodyn, a numerical engineering and climatology company dedicated to the sectors of renewable energy, wind simulation, meteorology, and wind safety.
Meteodyn | meteodyn.com