Consumer preferences for automated electric vehicle charging

EV-Flex is a first of kind randomised control trial, delivered in partnership with the The Abdul Latif Jameel Poverty Action Lab (J-PAL)

Context

The opportunities that the electrification of mobility presents are well-documented; switching from fossil-fuel powered motor vehicles to electric vehicles (EVs) reduces carbon emissions and the adverse effects of combustion engines on air quality. Research covering multiple geographies shows that those exposed to the worst air pollution are more likely to be from low-income groups, and racial and ethnic minorities. The switch to EVs presents a major opportunity to tackle the distributional impacts of the global carbon emissions that come from transport, whilst recognising the ongoing utility of access to transportation for socioeconomic mobility.

Yet without intelligent design, the future energy system could be challenged by issues of mass electrification. Fortunately, EVs can be highly flexible in their operation, especially if they are charged via an automated response to market or grid signals.

The global evidence base for how to incentivise consumers to switch to smart charging systems that are responsive to cheap and clean electricity is currently limited, given the low level of penetration of EVs. Yet sales of electric cars are set to continue increasing globally - nearly one in five cars sold in 2023 was electric - making our ability to understand their potential intelligent use critical.

Field trial

We ran the first randomised control trial that investigates the level of incentive required for a consumer to shift to a managed EV charging tariff. We researched:

  1. Consumer preferences for automation tariffs. The costs of switching EV owners to a managed charging tariff, via an empirical assessment of their willingness to accept a regular incentive in exchange for agreeing to managed charging.
  2. The grid and consumer benefits of automation tariffs. This will be explored via an empirical assessment of how the managed charging tariff changes customers’ electricity consumption profile towards half-hours that are lower carbon-intensity and less grid-constrained, and an assessment of the reduction in what consumers pay per month and/or per kWh of electricity they consume.

Results

The trial provides novel empirical evidence that AI-managed EV charging can reshape electricity demand at scale, with considerable benefits for consumers and the grid.

① Targeted consumer engagement can increase take-up of EV charging tariffs. Among a ‘harder-to-reach’ sample, simply an email raised tariff adoption by 3.4 percentage points, while an extra offer of £50/month for three months nearly doubled that effect. Customers that switched to the tariff largely remained on it for the 12 month trial period.

② Managed charging provides significant load-shifting. The tariff led to a 42% reduction in peak household electricity use, with the entirety of EV demand shifted to off-peak hours. This load-shifting did not change total household electricity consumption.

③ Automation is accepted by users and highly responsive to system conditions. We saw high user adherence to the automated schedule, with manual overrides only accounting for 2.3% of total electricity consumption and more than half of households never using this feature. The AI-driven managed charging tariff was more responsive to wholesale electricity prices than a static time-of-use tariff that does not use managed charging.

④ Significant benefits for society and the electricity system. Our results suggest that AI-managed charging generates a consumer surplus, including a reduction in electricity bills by £343 per year - or £650 when compared to a standard flat tariff. For the electricity grid, we found strong empirical evidence that managed charging can provide EV demand flexibility at scale to lower system costs.

Check out the full summary and working paper.

The trial was being funded by The Abdul Latif Jameel Poverty Action Lab (J-PAL)’s, King Climate Action Initiative (K-CAI).