Differential Geometric Foundations for Power Flow Computations

TL;DR

This paper introduces a differential geometric framework for analyzing power flow solution spaces, providing new computational methods for tracing and estimating distances to critical solution boundaries in power systems.

Contribution

It develops differential geometric tools and a high-precision continuation method for power flow analysis, focusing on the solution space boundary and its curvature properties.

Findings

Methods for computing tangent vectors and normals of the solution space boundary

A new high-precision continuation method near the solution boundary

Effective techniques for estimating distances to the solution boundary

Abstract

This paper aims to systematically and comprehensively initiate a foundation for using concepts from computational differential geometry as instruments for power flow computing and research. At this point we focus our discussion on the static case, with power flow equations given by quadratic functions defined on voltage space with values in power space; both spaces have real Euclidean coordinates. The central issue is a differential geometric analysis of the power flow solution space boundary (SSB, also in a simplifying way, called saddle node bifurcation set, SNB) both in voltage and in power space. We present different methods for computing tangent vectors, tangent planes and normals of the SSB and the normals' derivatives. Using the latter we compute normal and principal curvatures. All this is needed for tracing the orthogonal projection of points on curves in voltage or power space onto the SSB on points closest to the given points on the curve, thus obtaining estimates for their distance to the SSB. As another example how these concepts can be useful, we present a new high precision continuation method for power flow solutions close to and on the SSB called local inversion of the power flow map from voltage to power space, assuming the dimension of power flow's Jacobean zero space, called KERNEL, is one. For inversion, we present two different geometry-based splitting techniques with one of them using the aforementioned orthogonal tracing method.