Measurement location for state estimation of distribution networks with generation A. Shafiu, N. Jenkins and G. Strbac Abstract: State estimation is important for the automatic management and control of complex distribution networks with significant distributed generation. State estimation has been used extensively on transmission systems where, generally, measurements of busbar voltages and line power flows exist. However, distribution systems normally have only a limited number of measurements. In such systems additional measurements are expensive and careful selection of location becomes important. The paper presents a heuristic approach to identify potential points for location of voltage measurements for state estimation as part of a proposed distribution management system controller. The developed technique identifies measurement locations to reduce the voltage standard deviation of the busbars which do not have a measurement. It addresses the problems of classical transmission meter placement methods, which are not directly applicable to distribution systems due to limited measurements, and unobservability of the network. 1 Introduction Electrical distribution networks have traditionally been used to transport electrical power from a high voltage infeed to consumers. These unidirectional power flows can be disturbed significantly by the introduction of distributed generation (DG). As more and more power is injected into the distribution network, the system voltage often rises, thus limiting the output of the distributed generators [1]. It has been shown that by using various strategies [2] to control the generator(s) and the transformer taps actively, the amount of generation that can be connected without reinforcing the network can be increased by 3–5 times. If these different control strategies are incorporated into a distribution management system controller (DMSC) to manage the distribution network actively, then increased penetration of DG may be achieved. With the help of a DMSC, area based voltage control and reactive power management procedures can be adopted to increase the penetration of DG. To implement these power management procedures and control actions, the DMSC needs knowl- edge of the state of the system in terms of network topology, line flows and busbar voltages. Acquiring the state of the distribution network is challenging. This is mainly due to lack of available real time measurements, especially at lower voltages. In the UK, supervisory control and data acquisition (SCADA) systems generally extend only down to 33kV networks. Such SCADA systems are seldom available on 11kV circuits. Adding real time measurements at remote points on the 11kV distribution networks is expensive, and careful choice of location of measurements is required. Some literature exists on the location of measurements on transmission systems but little emphasis has been given to the choice of measurement locations on distribution systems with DG. This paper describes a heuristic approach to identify the best locations for placing voltage measurements for distribution state estimation (DSE) with DG. The method identifies busbars on which to place a given number of voltage measurements by reducing the standard deviation of the voltages at those busbars in the network which do not have a measurement. 2 Measurement placement techniques Many measurement location techniques have been devel- oped for transmission networks. These techniques have evolved with the introduction of state estimation by Schweppe and Wildes for power systems [3]. Perhaps Schweppe was the first to recognise the importance of meter placement for transmission state estimation. He proposed placing meters to reduce the variance of the estimated quantities. Using the covariance matrix, meters are placed to minimise the standard deviations of the estimated states. The standard deviations of the estimated quantities are the diagonals of the inverse system gain matrix [4] . Monticelli and Wu [5] described an iterative process to place injection pseudomeasurements until the network becomes observable. The method is based on a linearised state estimation process (DC state estimation) that adds angle pseudomeasurements to zero pivots during triangular factorisation of the gain matrix. The addition of the angle pseudomeasurements allows factorisation and hence the DC state estimation to proceed. A DC state estimation is then performed with all the measurements set to zero except the angle pseudomeasurements. Injection pseudomeasure- ments are added only to candidate busbars. Candidate busbars are nodes without an injection measurement and whose connecting branches have a non-zero flow during each iteration of the DC state estimation. The authors are with the Department of Electrical Engineering & Electronics, UMIST, Ferranti Building/A4, PO BOX 88, M60 1QD, UK E-mail: a.shafiu@student.umist.ac.uk r IEE, 2005 IEE Proceedings online no. 20041226 doi:10.1049/ip-gtd:20041226 Paper first received 13th November 2003 and in revised form 20th October 2004. Originally published online: 24th January 2005 240 IEE Proc.-Gener. Transm. Distrib., Vol. 152, No. 2, March 2005