23 rd International Conference on Electricity Distribution Lyon, 15-18 June 2015 Paper 0915 CIRED 2015 1/5 EXTRACTION OF 9,163 REAL LV NETWORK MODELS FROM DNO GIS DATABASE TO ASSESS OVERVOLTAGE FROM PV AND CONSEQUENT MITIGATION MEASURES Andrew CROSSLAND Neal WADE Darren JONES Newcastle University – UK Newcastle University – UK Electricity North West Limited - UK a.f.crossland@gmail.com neal.wade@ncl.ac.uk darren.jones@enwl.co.uk ABSTRACT The scale of the overvoltage problems that will be seen across a Distribution Network Operator’s low voltage network is quantified for increasing levels of solar photovoltaic integration. This is completed on 9,163 real low voltage network models which have been extracted using a bespoke method from a Geographical Information System. Two configurations of distributed energy storage are investigated; home storage randomly taken up by customers; and storage directed by the Distribution Network Operator to areas where it will give the greatest effect on overvoltage problems. It is found that a planned approach is more effective in mitigating overvoltage than if consumers adopt storage in an ad-hoc manner, although the later can bring advantages. INTRODUCTION Rooftop solar photovoltaics (PV) are the most widely connected form of distributed generation in European low voltage (LV) power networks in terms of the number of systems installed (e.g. [1]). As well as positive impacts, PV can have negative effects on distribution networks. These are widely studied on small numbers of LV networks (e.g. [2]–[4]). Such issues can be problematic and costly for Distribution Network Operators (DNOs). DNOs do not always maintain models of their LV networks. Therefore, to investigate a particular LV network, a DNO might manually build a network model by looking at their Geographical Information System (GIS). However, this slow process prohibits production of the large number of network models needed to allow assessment of the cost implications from PV across all of a DNOs LV distribution networks. Alternatively, a large number of LV models can automatically be extracted from a DNO’s GIS database, such as in [5]. A custom made procedure for doing so is presented here. By using models extracted using this method from an entire DNO’s operating area, real technical and financial implications of PV on a large number of LV networks are presented. To mitigate any problems due to PV, new methods are needed to reduce costs. Therefore an assessment is also performed into how to install energy storage in LV networks to mitigate problems associated with PV. Such results have relevance for both DNOs and policy makers. THEORY As shown in Figure 1, an LV network has a secondary (MV/LV) transformer with a number of LV fuses/circuit breakers. From these fuses, a number of LV feeder cables run above or below ground. The feeder cable can split into sub-feeders and is connected to homes and businesses through service cables as shown. Figure 1: Overview of a typical LV network GIS maps the configuration and location of cables/other components in a DNOs distribution network. Typical parts of GIS are shown in Figure 2 and Figure 3 in which the various components previously described can be seen. It is noted that service cables usually enter a home or business perpendicular to the orientation of the roof. This is useful for estimating where PV might be installed in residential LV networks. Since the GIS system is held electronically, it is possible to computationally extract LV network models from it. Figure 2: Components around a MV/LV transformer in GIS Figure 3: Typical section of LV network in GIS MV/LV Transformer MV busbar LV busbar LV fuses LV feeders Service cable LV feeder cable Domestic load A B MV cables MV/LV transformer LV fuses LV busbar LV feeders LV feeder House Service cable Cable type Feeder cable connector Service cable connector