Assessing low-voltage network headroomunder EV and heat pump uptake using large-scale empirical data

Extended abstract 4-213-26

The ongoing uptake of low-carbon technologies (LCTs) such as electric vehicles and heat pumps is reshaping peak electricity demand on low-voltage (LV) distribution networks. Where this additional load exceeds existing capacity, substation reinforcement is required, which represents significant cost. The challenge is that the LV network is vast and sparsely monitored, making it difficult to assess where headroom genuinely exists and where constraints are most likely to bind. This motivates the central question of this work: how much headroom remains in the LV network under plausible trajectories of LCT uptake?

We combine two large empirical datasets: network topology and substation ratings from Electricity North West Limited (a UK network operator) GIS records, and building-level attributes from the UCL 3DStock project. A spatial join associates every mapped substation with the real set of buildings it serves and their use class. Of 28,797 substations mapped, 12,578 were identified as predominantly residential (>50% domestic buildings) and retained for analysis.

A simple Bernoulli demand model is applied at the household level: each residence is independently assigned an LCT with some probability, with non-LCT households modelled at a 3 kW peak and LCT households at 10 kW, scaled down by a coincidence factor to capture diversity. The resulting distribution of substation peak demand is evaluated across a sweep of LCT uptake and coincidence values and compared to real rated transformer capacity.

Results show that under modest coincidence factors, very few substations exceed 100% loading even at high LCT penetration. Under higher coincidence, which is plausible if, for example, houses with LCTs respond to shared dynamic price signals, a substantial share of residential substations breach this threshold (see figure). Exceeding rated capacity does not guarantee thermal failure, since this requires sustained overload, but it marks networks operating close to their limits which may be at elevated risk. Future work will extend the analysis to line loading, voltage drop, and the geographic heterogeneity of LV headroom constraints.

Read the full abstract.

Download presentation.