OPTIMIZATION FOR RENEWABLE ENERGY

Currently, the optimization of local placement of wind turbines on potential wind farms (also called micrositing) has a wide degree of variability in terms of accuracy of modeling, inclusion of design variables and parameters, and overarching objective. Though many researchers have used computational optimization tools to determine local wind turbine placement, the current work seeks to incorporate considerably more relevant information into the three-dimensional layout optimization by using state-of-the-art optimization algorithms. Though the exact objectives vary, the fundamental goal of installing a wind farm is usually to maximize the energy capture of the site, while minimizing other objectives – such as cost of development and cost of operation and maintenance. Maximizing the annual energy production of a farm can involve many complex variables, though the primary factors include the wind resource at the site, the turbine and topographic geometry, and the direct effect these have on the aerodynamic interactions of the turbines on the site.

The potential of a wind turbine to produce energy is primarily contingent on the wind resource of the proposed site, as the effective wind speed directly preceding the rotor determines the rotational capability of the blades. The wind flow through the rotors of a turbine transfers some of its momentum to the turning blades, causing a wind speed deficit downstream. Placement within this region of momentum deficit (known as the wake) therefore limits the amount of power a downstream turbine can develop.  When turbines are placed in close proximity, this effect can be dramatic. One must consider the geometry of the turbines, the preliminary wind data from a proposed site (both wind speed and direction), and the region’s topography in order to account for wake deficits mathematically when micrositing. Our work seeks to do this by developing novel optimization algorithms and extensions that directly support the inclusion of local variables, while incorporating up-to-date modeling of wake interaction dynamics and farm costs as part of the design objectives.

To date, our published results include that of a wind farm layout optimization using an Extended Pattern Search Algorithm for three different wind cases. The Extended Pattern Search is a deterministic search algorithm that, given an initial layout, selects potential new turbine locations based on a set pattern of directions and varying step sizes, such that global movement is favored at the start of the search, ending in precise local placement. Multiple extensions to this algorithm allow for random characteristics to be infused into the search, which aids in the avoidance of selecting local optima.  The benefit of extensions to the search was made evident as the EPS was able to develop superior layouts than a strictly deterministic pattern search. In addition, the EPS was able to develop more efficient layouts than those of previous literature that employed other optimization algorithms.

Primary Researcher: BRYONY DUPONT