[Paper Review] Distributed Load-Side Control: Coping with Variation of Renewable Generations

Zhaojian Wang, Shengwei Mei+3 more|arXiv (Cornell University)|Apr 13, 2018

Microgrid Control and Optimization25 references3 citations

TL;DR

This paper proposes a distributed load-side control framework that jointly uses consensus-based optimization and adaptive internal model control to manage time-varying power imbalances caused by renewable generation. It decomposes imbalance into steady, low-frequency, and high-frequency components, enabling economic allocation, local mitigation of unknown variations, and robustness against disturbances—demonstrated via simulations on the New England system and real wind data with sub-0.003Hz frequency deviation.

ABSTRACT

This paper addresses the distributed frequency control problem in a multi-area power system taking into account of unknown time-varying power imbalance. Particularly, fast controllable loads are utilized to restore system frequency under changing power imbalance in an optimal manner. The imbalanced power causing frequency deviation is decomposed into three parts: a known constant part, an unknown low-frequency variation and a high-frequency residual. The known steady part is usually the prediction of power imbalance. The variation may result from the fluctuation of renewable resources, electric vehicle charging, etc., which is usually unknown to operators. The high-frequency residual is usually unknown and treated as an external disturbance. Correspondingly, in this paper, we resolve the following three problems in different timescales: 1) allocate the steady part of power imbalance economically; 2) mitigate the effect of unknown low-frequency power variation locally; 3) attenuate unknown high-frequency disturbances. To this end, a distributed controller combining consensus method with adaptive internal model control is proposed. We first prove that the closed-loop system is asymptotically stable and converges to the optimal solution of an optimization problem if the external disturbance is not included. We then prove that the power variation can be mitigated accurately. Furthermore, we show that the closed-loop system is robust against both parameter uncertainty and external disturbances. The New England system is used to verify the efficacy of our design.

Motivation & Objective

• To address the challenge of frequency control in multi-area power systems with unknown, time-varying power imbalances due to renewable generation.
• To design a distributed control architecture that enables fast, localized response using controllable loads without centralized coordination.
• To decompose power imbalance into three timescale components: known steady, unknown low-frequency variation, and high-frequency disturbance.
• To achieve economic allocation of steady imbalance, accurate tracking of unknown variations, and robustness against disturbances and parameter uncertainty.
• To validate the controller’s performance using the New England system and real offshore wind farm data.

Proposed method

• The power imbalance is decomposed into three parts: a known constant (predicted imbalance), an unknown low-frequency variation (e.g., from wind/solar fluctuations), and a high-frequency residual (treated as disturbance).
• A slow-timescale consensus-based controller allocates the steady imbalance across areas to minimize economic cost, solving a distributed optimization problem.
• A medium-timescale internal model controller is designed to track and compensate for unknown low-frequency variations using an exosystem model of the variation dynamics.
• A fast-timescale control layer ensures robustness via L2-gain analysis, attenuating high-frequency disturbances and parameter uncertainties.
• The controller is implemented in a hierarchical, distributed fashion, with each area using local frequency measurements and communication with neighbors.
• Theoretical stability is proven: asymptotic convergence to the optimal solution under no disturbance, and robustness under uncertainty and disturbances.

Experimental results

Research questions

• RQ1 How can distributed load-side control effectively manage unknown, time-varying power imbalances caused by renewable generation?
• RQ2 Can a unified control framework handle multiple timescales of imbalance—steady, low-frequency variation, and high-frequency disturbance—simultaneously?
• RQ3 How can economic dispatch of steady imbalance be achieved in a distributed manner without centralized coordination?
• RQ4 To what extent can an internal model controller accurately track and compensate for unknown low-frequency power variations?
• RQ5 How robust is the controller against parameter uncertainty and high-frequency disturbances in real-world conditions?

Key findings

• The closed-loop system achieves asymptotic stability and converges to the optimal solution of the economic dispatch problem when external disturbances are absent.
• The internal model controller accurately mitigates unknown low-frequency power variations, ensuring minimal frequency deviation.
• The controller demonstrates robustness against both parameter uncertainty and high-frequency disturbances, with maximal frequency deviation under 0.003Hz in simulations.
• In simulations using real offshore wind farm data, the proposed controller maintained frequency stability with sub-0.003Hz deviation, while PI and prior distributed controllers failed to eliminate variation.
• The controller outperformed conventional PI and the Mallada et al. (2017) distributed controller in frequency regulation, showing superior suppression of fluctuations.
• Controllable loads accurately tracked real wind power variations across all areas, confirming effective local control action.