Transmission Planning Improvements with Probabilistic Convex PCH Models (GR-14-01)

Principal Investigator: Dr. Roy McCann

This project develops a method for analyzing electric transmission and generation systems that incorporates aspects of long-term planning models with operational (real-time market and security/thermal constrained) system constraints. Recognizing that optimal power flow problems are generally non-convex, the research considers non-parametric probabilistic (Bayesian) techniques for optimizing planning scenarios with consideration of variable wind generation. In the case of market-based (LIP/LMP) operational cost constraints, a convex optimization method will be investigated. The topic of probabilistic convex optimization methods have been previously developed in the context of machine learning algorithms (e.g., speech recognition, automated VLSI design, etc.) but not for electric power system planning and operations. However, the challenges of optimizing complex systems with uncertain operating conditions under cost constraints have many similarities.

This project benefits member companies whose operations include the real-time electricity markets (LIM/LMP, day-ahead, hour-ahead, 5-minute intervals) by providing planning models that more closely align with actual operating conditions. This will also include the effects of large-scale wind farm generation in optimizing future asset utilization. That is, the intent is to provide improved power flows over transmission lines within their respective thermal/stability/security constraints in order to best support electricity market transactions.