With the increasing uptake of new technologies (such as PV, wind, EVs and LED lighting) there is a need to study their effect on the low voltage (LV) network, and this is one objective of the Green Grid project. Although detailed modelling of an LV network will provide a more accurate assessment of their impact for a given scenario and network, to develop guidelines this type of impact assessment can be performed on representative networks. These studies will identify the penetration level, for example of PV, that a typical system can withstand without adverse effects. This will be a function of the characteristics of the PV inverter and its controls and the benefit of different control schemes can be evaluated. It allows quick evaluation and can identify which scenarios need further evaluation with more particular details. The generation of representative LV networks that are truly representative is not a trivial task and is the focus of this paper. The k-means clustering technique is applied to cluster LV networks supplied by 10558 MV/LV transformers on the Orion distribution network. The extent and largely high quality of the Orion data has enabled the silhouette method to be used to evaluate how good the cluster fit is. The clusters identified from the data are city centre, urban, industrial and rural. The technique gives the median and two extremes for each cluster. Several examples of using these representative networks for evaluating the impact of PV on the LV network are also provided.