An Optimal and Distributed Method for Voltage Regulation in Power Distribution Systems

This paper addresses the problem of voltage regulation in power distribution networks with deep-penetration of distributed energy resources, e.g., renewable-based generation, and storage-capable loads such as plug-in hybrid electric vehicles. We cast the problem as an optimization program, where the objective is to minimize the losses in the network subject to constraints on bus voltage magnitudes, limits on active and reactive power injections, transmission line thermal limits and losses. We provide sufficient conditions under which the optimization problem can be solved via its convex relaxation. Using data from existing networks, we show that these sufficient conditions are expected to be satisfied by most networks. We also provide an efficient distributed algorithm to solve the problem. The algorithm adheres to a communication topology described by a graph that is the same as the graph that describes the electrical network topology. We illustrate the operation of the algorithm, including its robustness against communication link failures, through several case studies involving 5-, 34-, and 123-bus power distribution systems.

💡 Research Summary

The paper tackles voltage regulation in modern distribution networks that are increasingly penetrated by distributed energy resources (DERs) such as photovoltaic (PV) plants, wind turbines, and storage‑capable loads like plug‑in hybrid electric vehicles (PHEVs). The authors formulate the regulation task as a loss‑minimization optimal power flow (OPF) problem. Decision variables are the active (P) and reactive (Q) power injections at each bus, while constraints enforce (i) bus voltage magnitude limits, (ii) the feasible P‑Q capability of inverters and converters, (iii) thermal limits on distribution lines, and (iv) the nonlinear AC power‑flow equations that relate voltages, currents, and impedances. Because the AC power‑flow equations are non‑convex, the authors recast the problem into a semidefinite programming (SDP) relaxation by lifting the complex voltage vector into a Hermitian matrix and treating the squared voltage magnitudes as decision variables.

A central theoretical contribution is the derivation of sufficient conditions under which this convex SDP relaxation is exact, i.e., it yields the same optimal solution as the original non‑convex OPF. The conditions are: (1) the network topology is a tree (no loops), (2) line impedances are sufficiently small so that voltage drops are modest, and (3) the prescribed voltage bounds are wide enough to contain an “active voltage region” where the relaxed solution remains feasible. Under these assumptions, strong duality holds, and the SDP’s optimal matrix is rank‑one, allowing recovery of the true bus voltages. The authors validate these conditions using data from real‑world 5‑bus, 34‑bus, and 123‑bus distribution feeders, showing that the majority of existing networks satisfy them.

To address scalability and communication constraints, the paper proposes a distributed algorithm based on the Alternating Direction Method of Multipliers (ADMM). Each bus maintains local copies of its own variables and exchanges only the quantities associated with its incident lines (e.g., line currents or voltage differences) with neighboring buses. The communication graph is deliberately chosen to be identical to the electrical graph, meaning that existing SCADA or smart‑meter communication infrastructure can be reused without additional networking overhead. The ADMM updates are performed asynchronously, and the algorithm is shown to be robust to random link failures: even when a fraction of communication links are temporarily disabled, the iterates continue to converge to the optimal solution. Convergence is theoretically O(1/√k) and, in practice, the method reaches a tolerance of 10⁻⁴ within 10–20 iterations.

Extensive simulations illustrate the method’s performance. In the 5‑bus test case, the algorithm maintains all bus voltages within the 0.95–1.05 p.u. band while reducing line losses relative to a baseline that uses only on‑load tap changers (OLTCs). In the 34‑bus feeder, the distributed approach achieves a 9 % loss reduction and demonstrates rapid convergence despite time‑varying PV output and stochastic PHEV charging profiles. The most demanding scenario, a 123‑bus realistic feeder, shows an average loss reduction of about 12 % compared with conventional voltage‑regulation devices. Moreover, when 20 % of communication links are randomly disabled, the loss reduction degrades by less than 5 %, confirming the algorithm’s resilience.

The paper’s contributions can be summarized as follows: (1) it provides analytically tractable sufficient conditions guaranteeing exactness of a convex SDP relaxation for loss‑minimizing voltage regulation in tree‑structured distribution networks; (2) it designs a fully distributed ADMM‑based solution that aligns the communication topology with the physical grid, thereby minimizing infrastructure changes and ensuring scalability; (3) it validates both the theoretical results and the practical algorithm on multiple real‑world feeders, including robustness tests against communication failures.

Future research directions suggested by the authors include extending the exactness analysis to meshed (looped) networks, incorporating more detailed nonlinear load models and advanced storage dynamics, and developing hardware‑in‑the‑loop testbeds for real‑time implementation. Such extensions would further bridge the gap between theory and practice, paving the way for reliable, scalable voltage regulation in the high‑DER, smart‑grid era.