Design Mining VAWT Wind Farms. (arXiv:1506.08781v2 [cs.NE] UPDATED)

The small body of previous work considering the design of wind farms has used arrays of turbines originally intended to operate alone, optimising the layout of homogeneous turbines essentially as an afterthought in the design process. In this paper, we consider designing wind farms composed of collaborating wind turbines. Computational intelligence is combined with rapid prototyping whereby candidate designs are physically instantiated and evaluated under fan-generated wind conditions. It is shown possible to use surrogate-assisted coevolutionary algorithms to aerodynamically optimise the potentially heterogeneous morphology of an array of 6 small scale and closely positioned vertical-axis wind turbines using the total angular kinetic energy of the array as the objective. This approach performs optimisation in the presence of complex inter-turbine wake effects and multi-directional wind flow from nearby obstacles, which is extremely difficult to achieve accurately under high fidelity computational fluid dynamics simulation. The general approach is equally applicable to the design of other forms of sustainable energy where the characteristics of the environment and/or materials involved are too difficult to accurately simulate.

from cs.AI updates on arXiv.org http://ift.tt/1Nx7TnZ
via IFTTT