An integrated software to improve grid reliability and efficiency by assessing current grid status and providing predictive capability for system cascading failure analysis using advanced high-performance computing, statistical, and visualization techniques.
POC: Yousu Chen
State estimation (SE) and contingency analysis (CA) are two critical energy management system (EMS) functions in a control center. Currently, state estimation computation time in a control center is based on sequential computation, which is not fast enough to accurately capture the system state, especially in the case of a severe disturbance. Except for a certain number of pre-selected contingency cases with a solution time of about 1 minute, today’s typical practice of contingency analysis is required to be run at an interval of a few minutes within a balancing authority or a region.
Unfortunately, these preselected cases cannot guarantee coverage of all critical contingencies for different operation points with different generation and load patterns, particularly with the uncertainty brought by the penetration of distributed resources and smart loads. Meanwhile, the impact of the wind/solar power on the power grid operation is increasing, bringing greater amounts of uncertainty. The industry is facing the challenge of integrating new mechanisms like 15-minute scheduling and probabilistic flow forecast systems for handling intermittency, but a look-ahead predictive capability is an urgent need for system operation, planning and market. Thus, there is a significant need for running state estimation and contingency analysis with this type of stochastic behavior.
However, as the nature of the grid becomes more dynamic, today’s tool are even less likely to achieve real-time and predictive analysis of system events.
Analyzing events more quickly, understanding behavior and operations of distributed and intermittent resources, and studying the impact of multiple contingency events requires the help of High Performance Computing (HPC) techniques and considering forecasting information into grid operation. Due to the large size of simulation data, advanced data management and data mining techniques have to be developed to aid operators to identify the critical information extracted from simulation outputs that can come from multiple data resources.
Typically, commercial tools still present results in a tabular data form, but that is not easy to interpret within a short timeframe when the system is stressed. Advanced visualization techniques are needed to convert a large amount of data, including uncertainty, to actionable information which allows operators easily digest and make timely informed decisions.