Title: Autonomous Quadratized Optimal Power Flow via Convex Solution - Sequential Linear Programming (CS-SLP)
Dr. Sakis Meliopoulos, ECE, Chair, Advisor
Dr. Santiago Grijalva, ECE
Dr. Daniel Molzahn, ECE
Dr. Valerie Thomas, ISyE
Dr. Nikolaos Sahinidis, ISyE
Abstract: This dissertation develops high-fidelity physically-based object-oriented models (of grid-connected devices) and advanced computational tools and optimization models for the optimal and reliable operation of power systems in the presence of FACT devices. Specifically, a generalized convexification method for the high fidelity quadratized optimal power flow and an OPF solution algorithm named Convex Solution-Sequential Linear Programming (CS-SLP). CS-SLP starts by solving the convex quadratic optimal power flow with a commercial convex solver. Then starting from the convex solution, the initial problem is solved using SLP. Achievement of the objective will occur through the following contributions: (1) High fidelity physically-based device modeling that is cast into a universal object-oriented syntax, allowing for the integration of detailed models of Inverter Based Resources (IBR); (2) The use of the universal object-oriented syntax allowing for the seamless integration of protection, optimization, and control, as well as the integration of the network model in different OPF applications; (3) The use of the universal syntax to form the OPF allows for the convexification of the problem around the current operating point by minimal relaxation (minimal term additions). (4) SLP based on the co-state method is used to correct the convexification error. It efficiently linearizes the problem around the present operating conditions, resulting in a problem in terms of control variables. It features dynamic limits on control movement (ensuring that the linearized model remains in the valid region) and adaptive introduction/removal of model constraints.