Doctoral Dissertations

Date of Award

5-2025

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Energy Science and Engineering

Major Professor

Yilu Liu

Committee Members

Fangxing Li, Ibukunoluwa Korede, Yi Zhao

Abstract

The electric power grid is undergoing a significant transition towards renewable energy resource. In addition to this, there is a growing energy demand due to increased manufacturing and data center growth. With this in mind, grid planners must develop innovative strategies to accommodate the increased load and renewable energy penetration while maintaining grid reliability in a cost-effective manner. This dissertation proposes several novel tools and algorithms to improve grid planning and address some of the challenges currently faced in power system planning.

One of the key challenges faced by grid planners is the convergence issues encountered in the creation of accurate power flow models that represent the grid under various loading and generation scenarios. To address this convergence issues, three algorithms are developed to achieve power flow convergence in previously non-converging cases. The first algorithm achieves convergence by applying a deep learning neural network initializer to predict better initial conditions for the Newton-Raphson AC power solution method. Next a hot-starting algorithm with switched-shunt control is developed to solve previously non-converging power flow models and third algorithm applies a generator redispatch methodology to converge previously non-converging power flow models. These algorithms were successfully applied to the Electric Reliability Council of Texas (ERCOT) 6102 bus grid system under various operating conditions.

Beyond power flow convergence, this dissertation introduces a hosting capacity tool using an adaptive gradient descent algorithm. It determines the maximum generation capacity a bus can accommodate without violating thermal and voltage constraints, both in base case and N-1 contingency scenarios. Compared to the conventional sequential search method, this algorithm achieved higher accuracy with fewer iterations and was validated on the 243-bus Western Electricity Coordinating Council (WECC) system.

Finally, a steady-state interconnection tool was designed to automate the generation interconnection studies and was applied to a real US power distribution system.

Available for download on Thursday, May 15, 2031

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS