Distributed and Decentralized Control of Residential Energy Systems Incorporating Battery Storage

Abstract

Abstract-The recent rapid uptake of residential solar photovoltaic (PV) installations provides many challenges for electricity distribution networks designed for one-way power flow from the distribution company to the residential customer. For gridconnected installations, intermittent generation as well as large amounts of generation during low load periods can lead to a degradation of power quality and even outages due to overvoltage conditions. In this paper we present four control methodologies to mitigate these difficulties using small-scale distributed battery storage. These four approaches represent three different control architectures: centralized, decentralized, and distributed control. These approaches are validated and compared using data on load and generation profiles from customers in an Australian electricity distribution network.

Valuation of stored energy in dynamic optimal power flow of distribution systems with energy storage

“…Another possible extension of such models to take into account stochasticity within each day would be to retain the 24-hour planning horizon as a rolling horizon, updating the schedule for the operation of the ESS e.g. each hour [14,27,37].…”

A Distributed Optimization Algorithm for the Predictive Control of Smart Grids

“…In [16], we studied the problem of optimally scheduling battery storage at the residential level so as to minimize variability in energy supply and demand at a local level; for example at the level of a residential neighborhood where residences have installed both energy generation and storage technologies such as solar photovoltaic panels and batteries, respectively. Unfortunately, this optimal scheduling problem does not lead to a global cost function that is decomposable as a sum of cost functions at each residence.…”

“…Unfortunately, this optimal scheduling problem does not lead to a global cost function that is decomposable as a sum of cost functions at each residence. Hence, in [16], we proposed and compared centralized, distributed, and decentralized algorithms, where a degradation in performance was observed from the centralized solution. Of course, the benefit of the distributed and decentralized algorithms resulted from their scalability.…”

“…In this paper, we present a hierarchical, iterative, distributed optimization algorithm that converges to the solution of the centralized optimization problem for the specific cost function and system dynamics used in [16]. In contrast to [10], the algorithm presented herein relies on the presence of a central entity with which all agents communicate.…”

“…The paper is organized as follows: in Section II we recall the system dynamics and optimization problem considered in [16]. In Section III we recall centralized and decentralized model predictive control schemes from [16].…”