Wind Power Production Boosted by Supercomputers
Dashboard recently published an article in which we discussed the largely untapped potential of wind power, and how relatively low efficiency rates combined with high production costs are creating barriers to both development and adoption. It is therefore particularly exciting to hear that scientists at The University of Texas at Dallas (UT Dallas) have been working on a means of addressing this low convergence.
The UT Dallas team proposes to model the turbulence through large eddy simulations, in which movement can be integrated into supercomputers resulting in an accurate prediction of turbine behaviour. The developed code is able to mimic wind turbines, taking into account any interference they may experience, as well as regional variance, which was tackled using the Weather Research and Forecasting Model (WRF). When tested at a windfarm in North Texas. The predictions resulted in a 90% match against the physical behaviour of turbines.
These results, combined with calculations from numerical simulations would provide a predicated 6-7% increase in wind power efficiency. At this scale, these numbers represent a substantial development, as just 1% improvement would create $100 million in value if applied across all currently operational windfarms in the US alone.
One vital part of this supercomputer-lead approach was the development of the required algorithms, which are not only used to calculate, but also manage the operation of dynamic systems at wind farms. One of the control algorithms that was crucial to reach the 90% agreement rate was extremum seeking control, a model-free way of reaching best performance in dynamic systems even in conditions of limited knowledge of the system.
The lead researcher, Stefano Leonardi, associate professor of mechanical engineering at UT Dallas, says: “The important thing is that the control algorithm does not rely on a physics-based model. There are many uncertainties in a real wind farm, so you cannot model everything. The extremum seeking control can find the optimum no matter if there is erosion or icing on the blades. It’s very robust and works despite uncertainties in the system.”
Dashboard hopes that with more testing, systems such as these could make a real impact in terms of wind power’s conversion rate and subsequent adoption. Combining supercomputer intelligence with maximising wind output is a smart future-facing tactic, one we’re sure will be applied to more power production methods in the future.
Author: Nadja Kaukiainen