A Massively Parallel Computational Framework for Grid Integration of Wind Energy over Complex Terrain

Variable nature of winds along with congestion in existing transmission lines are two major challenges in grid integration of wind power. Driven by a vision to create a sustainable energy future, wind farms are increasingly being built in regions of complex terrain, as potential areas over flat terrain are diminishing. Because complex terrain enhances wind turbulence and its variability, a short-term (0-6 hours) forecasting capability that can resolve wind speed and direction accurately is needed to balance the load on the electricity grid and overcoming transmission congestion via the dynamic line rating concept. Equally important, a forecasting capability can be used for trading profitably in the energy imbalance market. However, there are several modeling challenges to overcome. Additionally, any wind solver has to operate in forecasting mode to be a relevant tool for the aforementioned applications.

To address these challenges, a multi-scale forecasting engine is proposed for emerging supercomputers. In this engine, a regional weather forecast model will inform a micro-scale wind forecasting model that is accelerated with multiple graphics processing units (GPU). The microscale wind solver adopts a Cartesian mesh immersed boundary method to faithfully represent arbitrarily complex terrain while mapping well to the computer architecture of modern GPUs. The hardware oriented numerical approach enables more than an order of magnitude speedup in computational turnaround time. To enhance the accuracy of wind simulations at the microscale, major error sources are identified and tackled using benchmark cases to obtain a collective improvement in predictions. Large-scale wind simulations for a region in Southern Idaho are used to demonstrate the potential of a dynamic rating concept for increasing transmission line capacity.

Speaker Bio: Inanc Senocak is an associate professor of mechanical engineering and the director of High Performance Simulation Laboratory for Thermo-fluids at the Boise State University, where he leads computational science and engineering research projects related to energy and environment. Senocak obtained his B.Sc. degree in mechanical engineering from the Middle East Technical University in Ankara, Turkey and his Ph.D. degree in aerospace engineering from the University of Florida in 2002. Prior to starting his faculty career in 2007, he worked as postdoctoral researcher first at the Center for Turbulence Research, jointly operated by NASA Ames Research Center and Stanford University, and later at the Los Alamos National Laboratory, working on different aspects of atmospheric transport and dispersion. At Boise State he has taught courses in thermodynamics, fluid mechanics and parallel scientific computing and advised graduate-level research of mechanical engineering and computer science students. Senocak received an NSF CAREER award in 2011.