Electric vehicle charging demand: data and AI to the rescue | Eleven

Electric vehicle charging demand: data and AI to the rescue10 November 2023


Data science


Electric vehicle charging, a world of possibilities

The number of electric vehicles registered in France is rising impressively every year: +34% over the last twelve months (1). The number of vehicles on the road has risen by almost 225,000 to reach a total of 915,000 (September 2023). This explosion can be explained by a combination of many favorable structural factors: growing environmental awareness among the general public, regulatory constraints (e.g. the famous ZFE – Zones à Faible Emissions – low-emission zones in city centers), lower vehicle sales prices, a wider variety of vehicles, improved vehicle features (e.g. range), tax and regulatory incentives such as the purchase bonus, conversion of corporate fleets, rising fossil fuel costs, etc.

In parallel with this growth in the number of vehicles on the road, Electric Vehicle Charging Points (EVCP) are being rolled out at an accelerated pace, with the number of charging points set to increase by 59% between September 2022 and 2023 (2). This further growth is also being supported by the public authorities (PPE, Advenir program, etc.), but also by investments from private players, who are raising more and more funds.

The apparent parallelism of these two growth trends masks several decorrelation factors. Between the increase in the number of electric vehicles and the evolution of charging demand, there is a complex transfer equation that Data and AI (and therefore eleven) can help solve.

Anticipating demand for EVCP: more than just a volume x value equation

The tried-and-tested strategy consulting approach of breaking down a complex issue into simpler problems provides some initial insights. Firstly, in terms of volume, the quantity of energy to be distributed via charging stations depends on the distance covered – and to some extent on driving conditions: behavior, weather conditions, etc. This gives rise to an initial segmentation between, on the one hand, the “short-haul driver”, whose journeys are organized around home-to-work and leisure trips (evenings, weekends, and vacations) and, on the other hand, the “long-haul driver”, whose vehicle accumulates journeys during the day – rounds, deliveries, breakdown service, etc. The first is a private individual or a beneficiary of a car charging station. The former is a private individual or the user of a company vehicle, the latter a “road worker”, or at least someone who spends a significant part of his or her professional time behind the wheel.

The needs of these two types of consumers (and let’s leave aside truck drivers, who operate in a specific parallel world) will be met by 3 types of charging:

  1. Recharging at home or at the workplace (or even in a hotel), often without time constraints and therefore slow, and often at private charging stations.
  2. Recharging at destination (supermarkets, public parking lots), in places where the driver goes to take advantage of the opportunity to recharge.
  3. Recharging on the road (or while roaming), when the driver stops to recharge and avoid running out of fuel, the world of fast or ultra-fast charging…
A car parked in front of a building

Fast-charging station “on the road” in the Swiss Alps

We therefore have two types of consumers, whose demand can be broken down into 3 types of recharging: the short-haul driver will mainly use private recharging at home, and for his few long-distance journeys, public recharging on the road, or even at his vacation destination. To cover his 8,000 km per year, he’ll recharge his battery 40 to 60 times, by around 40-50 kWh each time. The long-haul driver will behave in a totally different way, as his need for recharging will be much more frequent and critical.

This first-level segmentation illustrates why it’s not just the number of vehicles that counts, but the nature of the driver behind the wheel. For example, the arrival of more affordable electric vehicles increases the proportion of short-haul drivers in the fleet, which distorts demand. At eleven, we have developed a much finer segmentation of uses to better understand and anticipate the market.

But the complexity doesn’t stop there. Demand for the same segment is constantly evolving. Over the long term, the evolution of electric vehicle offers, particularly in terms of range, limits the need for on-the-road recharging for short-distance drivers. Conversely, the deployment of public EVCP at affordable prices is reducing the need for long-range charging and supporting demand.

Other factors linked to user behavior, such as their maturity in the face of these new uses, influence the demand for recharging. For example, mature consumers are less likely to rush to a charging station at their destination when they have more than enough range to get back home.

In the short term, demand is subject to seasonal effects (in vacation travel and weather-related consumption), weekly cycle effects (home-work during the week, leisure and shopping at weekends) and hourly peaks. This variability in demand necessarily influences supply, both in terms of quantity and price.

Finally, as if this weren’t complicated enough, all these dimensions of demand depend on local parameters. They are not uniform across a territory. So, is it possible to see clearly in this universe of complexity?

Data science helps us see things more clearly

Like many new paradigms, the adoption of the electric vehicle and its corollary, the deployment of a suitable ECVP system, is still subject to a great deal of uncertainty, but this doesn’t mean that players in the sector are reduced to making blind bets.

On the one hand, data is already accumulating, some of which is available in Open Data: electric vehicle registrations, installed base of charging stations by type of supply, surrounding building types (residential, commercial, office), local purchasing power, etc. On the other hand, these problems of predictability, which often exceed the abilities of the human brain, are particularly well addressed by Artificial Intelligence models (time series, Bayesian networks, Machine Learning…). Those we have implemented on public data are already enlightening. When we enrich them with private data from the operation of existing ECVP (e.g.: actual history of charging sessions in terms of number and duration), they will be all the better.

Anticipating demand for electric vehicle charging is a major challenge for all market players. A detailed understanding of the associated economic stakes requires a comprehensive vision of the electric mobility environment. eleven’s in-depth knowledge of the world of mobility and energy, combined with its cutting-edge expertise in data-science, will help you make the most of this constantly evolving environment.

Sources :
1. https://www.avere-france.org/publication/barometre-septembre-2023-les-immatriculations-des-vehicules-electriques-et-hybrides-rechargeables-representent-255-des-parts-de-marches/
2. https://data.enedis.fr/explore/dataset/nombre-total-de-points-de-charge/table/?sort=-trimestre

Sur le même sujet

follow us

All rights reserved Eleven Strategy ©2024