SBIR Phase I: Agile Model Reduction for Topology Optimization Software

The broader impact/commercial potential of this project is a substantial increase in the efficiency and reliability of the power grid, which would result in lower electricity rates to consumers and facilitate the integration of renewable energy into the grid. This project will enable the use of topology optimization in the operation of the power transmission grid and allow operators to adapt the grid configuration to address changes in system conditions in real time. These adaptive reconfigurations increase the ability of the system to transfer power across the network in the directions that matter for economic, reliability, or environmental reasons. The economic benefits of topology optimization represent a 50% reduction in the cost of grid congestion, which translates to $1-4 billion/year production cost savings in the US. In addition, topology optimization would consistently reduce, or entirely remove, the otherwise frequent overloads on transmission facilities, thereby increasing the reliability of the grid. Topology optimization would also facilitate grid operations with large amounts of variable renewable resources, such as wind and solar, by relieving their curtailment by about 40%. Given the increase in renewable energy in the generation mix, topology optimization is expected to become even more effective and important in the future. This Small Business Innovation Research (SBIR) Phase I project will develop fast and adaptive power system model reduction software fully integrated with topology optimization software. The technical challenge arises from the fact that finding a power flow solution for a large full nodal model of a power system requires the model to be reduced to an equivalent smaller model, to avoid numerical issues with the full model. For typical analyses, this reduction only needs to be performed once; however, for topology optimization, in the course of finding a good topology configuration, hundreds of different topologies need to be analyzed. Thus, the main technical hurdle preventing topology optimization from being used in online operations decision-support is the current state-of-the-art model reduction computation speed. As such, this project?s objective is to lower the computation time of the model reduction component by at least 10 times compared to existing capabilities in commercial software, when used as part of topology optimization routines. Model reduction calculations will be designed to take advantage of problem-specific attributes of the topology optimization routine, which will enable reduced computation time, by, for example, making use of partial recalculations and decomposition of the problem to support parallelization.