Searcharxiv⌕ Search

arXiv subjects

Sanjay Johnson

Publications and source records attributed to Sanjay Johnson.

2 recordsLinked to original sources

Quantifying the Cost of Restoring AC Feasibility for Approximate Optimal Power Flow Solutions

Power system operations rely on approximate solutions of the AC optimal power flow problem, most notably the DC optimal power flow used in security-constrained unit commitment and economic dispatch, while convex relaxations, machine learning surrogates, and other approximations are studied extensively in the academic literature. Because these approximations neglect or relax parts of the AC power flow equations, their solutions must be adjusted before dispatch; for example, the DC optimal power flow, which neglects transmission losses, reactive power, and voltage magnitude variations, is known to produce solutions that are always infeasible. The cost of this adjustment is not adequately understood. This paper proposes a method that quantifies, for any approximate solution, the minimum cost of the adjustments that restore AC feasibility, which is a lower bound on the cost of the heuristic adjustments used in practice. We apply it to the DC optimal power flow, the second-order cone and quadratic convex relaxations, and the AC optimal power flow solved to a loose relative tolerance of $10^{-4}$ instead of $10^{-8}$. Relative to the AC optimal power flow solution obtained from a local nonlinear optimization solver, and even with the minimum adjustment, the DC optimal power flow requires adjusting, on average, generators that account for 15.6% of total generation capacity and incurs a 1.10% cost increase; the convex relaxations require more adjustment (17.2% and 16.9%) but smaller cost increases (0.68% and 0.78%); and the loose-tolerance AC optimal power flow requires 6.1% and incurs 0.42%. These results indicate significant value in operating markets directly on the AC optimal power flow, which avoids the distortion of the dispatch and prices caused by heuristic feasibility restoration.

eess.SY↗

ExaModelsPower.jl: A GPU-Compatible Modeling Library for Nonlinear Power System Optimization

As GPU-accelerated mathematical programming techniques mature, there is growing interest in utilizing them to address the computational challenges of power system optimization. This paper introduces ExaModelsPower.jl, an open-source modeling library for creating GPU-compatible nonlinear AC optimal power flow models. Built on ExaModels.jl, ExaModelsPower.jl provides a high-level interface that automatically generates all necessary callback functions for GPU solvers. The library is designed for large-scale problem instances, which may include multiple time periods and security constraints. Using ExaModelsPower.jl, we benchmark GPU and CPU solvers on open-source test cases. Our results show that GPU solvers can deliver up to two orders of magnitude speedups compared to alternative tools on CPU for problems with more than 20,000 variables and a solution precision of up to $10^{-4}$, while performance for smaller instances or tighter tolerances may vary.

eess.SY↗