Taking the Guesswork Out of Power Plant Dispatch

Abstract Effective power plant dispatch relies on precise performance predictions, particularly for gas turbines operating in changing environmental and operational conditions. This work builds on prior efforts constructing tools for performance visualization in both simple and combined cycle systems. Previous work developed a refined digital twin modeling framework to enhance performance predictions under changing weather conditions. With a set of training data, the resulting model requires only three ambient conditions—pressure, temperature, and humidity—to predict plant heat rate and power output. The goal was to strike a balance between modeling accuracy and practical usability, which tends to be missing in traditional approaches. This project introduces several advancements in the digital twin model. Extrapolation techniques were enhanced to improve reliability in scenarios outside the original training data. Augmentation modeling was developed to account for auxiliary systems and their effects on plant performance. Additionally, dispatch optimization integrates the performance prediction models with cost and operational constraints, enabling 7-day-ahead dispatch scheduling. Improvements were tested via a case study on Chevron Pipeline and Power gas turbines—the result is a tool, Optora, supporting better management of power plant performance. Optora offers operators a clear understanding of how plants perform across diverse scenarios while accommodating uncertainties inherent to real-world operations.