The rise of distributed generation and electric vehicles is changing the way consumers use the networks – and that means DNOs have to rethink the way they model usage on their LV networks.

New Thames Valley Vision (NTVV), a Low Carbon Network Fund (LCNF) project led by Scottish and Southern Electricity Networks, has looked at innovative methodologies for modelling the distribution network. The ability of distribution network operators (DNOs) to model loading patterns will allow improvements in the way in which network problems are identified and solved. This will help them to decide where network reinforcement is needed or whether more cost-efficient, smart solutions can be used.

Previous modelling of LV networks has been limited. Efficient and safe network operations demand that DNOs must ensure the power which flows through LV network circuits remains within loading limits and reaches the end customer within an acceptable voltage tolerance.

Since the inception of widespread LV networks, DNOs have refined their designs and operations within a regulatory environment which incentivises good reliability at least cost. As was shown in last week’s article in this series, this has resulted in a standardised approach to LV network design and operation which favours the simple and practical over the complex.

Unlike upstream systems, it has been rare for DNOs to maintain computer models of their low-voltage (LV) networks for power flow analysis. This is because LV network load growth has tended to be driven by increased connections rather than by higher household usage. This generalisation allowed DNOs to focus planning activity on new connections and load growth.

The radial architecture of the LV network makes manual calculation of total feeder load a straightforward matter of aggregating an average demand per customer. These records allow DNOs to develop a basic model which describes the composition of the feeder, but does not always show the finer points of the system.

The maximum demand measured by a domestic meter would typically be about 6-10kW. However, the average maximum demand per customer connected along an LV feeder is often shown be lower than 2kW. This effect is explained by the fact that neighbouring customers along an LV feeder use power at different times and in different ways.

For this reason, when planning capacity requirements of LV feeders, DNOs have traditionally applied formal assumptions for household consumption. These are chosen to reflect the average power drawn per property after the effect of behaviour diversity has been allowed for.

Low carbon technologies bring new challenges
We are now witnessing the adoption of domestic low-carbon technologies, such as solar panels, and an increase in the ownership of electric vehicles, which are being charged at home. These changes mean lower diversity in customer usage patterns, which worsens the loading profile on LV feeders. These new technologies typically have a large demand in comparison to existing household demand and are therefore likely to, when aggregated, take the load above the feeder rating.

In cases where there are large amounts of domestic PV, the nature of the feeder can be completely transformed to change the direction of flow such that the feeder exports upstream and raises the voltage along the feeder, potentially to unacceptable levels. The way in which customers use the network is changing, which means that the traditional approach to forecasting load flows and voltage drop along LV feeders can no longer be relied upon.

So, how NTVV is helping?

To calculate the amount of power flowing along a circuit, a consumption profile has to be assigned to each end point in the connectivity model. Because of the widespread availability of telecommunications, NTVV added monitoring to points of entry and exit of the LV network. A notable NTVV innovation was a smart methodology, which avoided the need to add monitoring to every customer. The approach buddied data from customers which have half-hourly monitoring to similar customers where no monitoring was available. This allowed NTVV to assign meaningful load profiles to all customers within the model.

The ability of DNOs to confidently model the loading patterns will also allow improvements in the way in which LV network ­problems are identified and solved. For example, once the load profile of an overloaded feeder is understood, a DNO can review whether traditional reinforcement is required or whether a smart solution such as network connected energy storage could be deployed.

A number of lessons have been learned through NTVV about how to build a more effective network modelling environment. Specifically, the use of the DNO’s GIS connectivity model is an appropriate methodology for the scale of network being modelled. This is supported with plant and asset data, customer address data, and information from customers related to their load or generation. In the future, this modelling tool can provide a platform for the DNO to host available smart meter data for direct use in the assessment of the network.

There is still some work to do to move from trials to the running of business-as-usual studies to make real-life network investment decisions; more data needs to be gathered and tested. A DNO that wishes to benefit from the methodology described here needs to invest in improved data gathering.

The building blocks of the models themselves are commercially available and largely already exist within DNOs. Increased experience with the integration of these components will increasingly show that modelling can bring positive financial benefits that justify the investment and, most importantly, benefit customers.

by Scottish & Southern Electricity Networks and DNV GL

This article is the second of five looking at how Scottish and Southern Electricity Networks and the NTVV project have been exploring better ways of managing distribution networks.