Electric Power Systems Research 129 (2015) 217–226 Contents lists available at ScienceDirect Electric Power Systems Research j o ur na l ho mepage: www.elsevier.com/locate/epsr Statistical modeling of aggregated electricity consumption and distributed wind generation in distribution systems using AMR data Matti Koivisto ∗ , Merkebu Degefa, Mubbashir Ali, Jussi Ekström, John Millar, Matti Lehtonen Department of Electrical Engineering and Automation, Aalto University, Otakaari 5 A, PO Box 13000, Espoo FIN-00076, Finland a r t i c l e i n f o Article history: Received 24 March 2015 Received in revised form 17 July 2015 Accepted 5 August 2015 Keywords: AMR data Distributed generation Electricity consumption Load model Statistical modeling Wind power a b s t r a c t This paper presents a methodology for carrying out statistical analysis of electricity consumption and distributed wind generation in distribution systems in order to investigate their combined effect, e.g., to give a probabilistic estimate of the effective peak net load. Hourly consumption profiles and the expected deviations from the profiles are estimated for different consumer group sizes of different types using auto- matic meter reading (AMR) data. In addition, a statistical approach to wind power modeling is presented. The consumption and generation models are then combined to give a probabilistic estimate of their com- bined effect in the long term using Monte Carlo simulations. The presented methodology is applicable to new locations without measurements. To showcase the applicability of the proposed methodology, three wind power and two electricity consumption scenarios are presented and compared. © 2015 Elsevier B.V. All rights reserved. 1. Introduction 1.1. Motivation and technique Distributed generation refers to electricity generation that is decentralized and located close to the electricity consumption. Including generation at the distribution network level reflects the shift away from a top-down paradigm (centralized large-scale generation feeding consumption-only distribution systems) to the mixed consumption and generation nature of the contemporary distribution system. If the generation is wind or solar, it is stochastic in nature. Such stochastic distributed generation requires statisti- cal modeling to understand the behavior of the generators, e.g., the probability of all nearby wind turbines generating at full power at the same time. As the generation is located near consumption, which is also stochastic due to the uncertainties in the behavior of electricity consumers, a statistical approach for analyzing both con- sumption and generation together is required to understand their combined effect. ∗ Corresponding author. Tel.: +358 50 433 1581; fax: +358 9 47022991. E-mail addresses: matti.koivisto@aalto.fi (M. Koivisto), Merkebu.Degefa@aalto.fi (M. Degefa), Mubbashir.Ali@aalto.fi (M. Ali), Jussi.Ekstrom@aalto.fi (J. Ekström), John.Millar@aalto.fi (J. Millar), Matti.Lehtonen@aalto.fi (M. Lehtonen). As an example, the statistical analysis of the combined long term effect of wind generation and electricity consumption can be used to estimate how many hours during a year the net gen- eration (wind generation subtracted by the local consumption) is above a given threshold for a given set of consumers and stochastic generators, for example, the consumption and generation behind a primary high/medium voltage (HV/MV) substation. When using the maximum generation that can be transferred as a threshold, the information on how many hours are likely to exceed the limit in a year can be used for comparing different options for managing the high generation hours (e.g., energy storage, demand response, HV network reinforcement, wind generation curtailment). Simi- larly, when considering the net load (consumption subtracted by the local generation), the likelihood of the load exceeding a given limit can be estimated. The analysis of the combined effect can also be used to estimate the expected peak net load (or net generation). In addition to the expected peak, statistical modeling can be used to estimate, for example, the 99% prediction interval (PI) of the peak in order to achieve a higher confidence that the limit will not be exceeded. Both the behavior of the combined effect of generation and consumption throughout the year and during the peaks is ana- lyzed in this paper. This naturally puts the focus of the presented methodology on the long term. The Monte Carlo (MC) simulation method is a statistical method that can be used to estimate the probability distribution of an aggre- gate when the constituent parts are defined by statistical models. http://dx.doi.org/10.1016/j.epsr.2015.08.008 0378-7796/© 2015 Elsevier B.V. All rights reserved.