With a growing amount energy on the grid coming from renewable energy sources, techniques that capture the intermittent nature of renewable energy must be used to inform the planning process. State-of-the-art electric grid capacity expansion models are memory and compute limited, making it harder to fully represent the impact of variable renewable generation and mitigation strategies via storage to enable an understanding of 100% penetration of renewables.
Scalable Power System Economic Expansion and Dispatch (SPEED) is a capacity expansion model that takes advantage of the availability of a stochastic optimization framework and a mathematical isomorphic coincidence to unlock the power of HPC to address the above challenges using parallel computing. By considering a suite of operational scenarios within the optimization process, SPEED produces planning decisions informed by spatial and temporal variations in renewable energy resources. To ensure scalability, the model is constructed to leverage the horizontal decomposition technique progressive hedging, which enables the model to be solved via parallel computing.
Energy planners use capacity expansion models to inform power system infrastructure planning decisions to meet future electrical power demand on the grid economically and reliably. To meet these goals, such optimization models must consider the operational implications of the infrastructure built.