The age of the electric vehicle (EV) is dawning, with two million of them sold in 2018. Whether EV purchases are motivated by environmental concerns, lower operating costs, increases in range coverage, or just plain envy of the neighbour’s EV’s constant torque acceleration, sales of all EV models are on the rise.

Sustainable electric vehicles herald the beginning of the end for fossil fuel-dependent combustion engines. But for grid managers known as distribution system operators (DSOs), this transition brings problems. Too many EVs in the same neighbourhood can lead to power disruption or even blackouts.

How can we encourage EV uptake without bringing down the grid?

We need to make the transition to EVs sensibly, which means understanding and preparing for the unintended consequences of electrification.

Imagine a typical American neighbourhood. Let’s say it has 1000 cars, only 50 of which are EVs. While EV charging represents a significant power drain on the grid, this neighbourhood may be perfectly capable of charging 50 EVs in the early evening, back home after a day spent transporting their owners to work or running errands.

Now let’s add another 50 EVs to the neighbourhood. After arriving home, these new 50 EV owners join the first 50 in plugging in. It’s only a matter of time before the EV draw on power becomes too great for the infrastructure to bear. This grid congestion - the inability of the local grid to meet the immediate power demands on it - can result in grid failure in the form of disruption or even outage.

Image: Oracle

Power outage is anathema for a DSO. A DSO takes over distribution of electricity from the high-voltage grid operator and brings it, via local infrastructure, to your home. It’s their job to make sure, first and foremost, that the supply of electricity is reliable. There are a few straightforward ways DSOs can prepare for potential EV-produced grid congestion:

1) align local and state politicians, as well as power market managers, toward large-scale expansion of grid infrastructure

2) secure the massive public finances required to build the infrastructure necessary to meet sporadic spikes in electricity demand

3) disrupt the neighbourhood and surrounding areas with months of construction work until the grid has the ability to deliver enough power at a moment’s notice to satisfy the maximum needs of the entire EV-owning neighbourhood at once

Unfortunately, these solutions are fraught with problems, so most DSOs take a different route. In some geographies, it’s quite normal to offer a cheaper rate for electricity drawn off the grid late at night when most people are asleep, and dishwashers and TVs have been turned off. Savings-oriented EV owners might opt to stay up until 11pm to turn on their charger.

Image: Oracle

Another, slightly more sophisticated option is currently in use by the St. Croix Electricity Cooperative in semi-rural western Wisconsin. New EV owners are offered a subsidy for a level two charger, the kind that can charge an EV faster, if they agree to an external control by the DSO that allows the charger to run only between the hours of 11pm and 7am. As an EV owner, you just plug in the EV whenever you get home and the charge doesn’t start until 11pm. Whether the car is fully charged or not, the charger turns off at 7am.

Eventually, and regardless of level two charger subsidies, once there are too many EVs in a neighbourhood, a DSO would most likely be pressed to allow only level two chargers with external control mechanisms. That’s a level of top-down control that’s understandable but not always popular.

What if we could prepare for the EV draw on power differently, balancing out the demands on the grid by utilizing the individual preferences of the EV owners themselves? We could offer EV owners the option to choose convenience of charging start time and duration over cost savings, or reliability of charging amount over convenience, and so on. Every day, every EV owner would make the charging choice that works best for them, while still managing to keep EV charging demands on the grid within acceptable levels.

Introducing ‘preference-based decision support’

While St. Croix has a static, one-time, incentive-based approach to encouraging nighttime charging, our research shows the value to both EV owner and DSO of offering a more subtle ‘demand-response’ approach. It is possible to satisfy the individual preferences of EV owners and prevent congestion by creating an agile market for electricity using a combination of dynamic pricing and remote control of charging ability, all facilitated by an automated, machine-learning ‘decision-assistant’.

Imagine, as a policy, if all new level two chargers were installed with the potential for an external control, but that EV owners were put in charge of the decision to exert that control.

Now, instead of the DSO automatically shutting down or opening up charging ability, every household could make the decision, via an app, to allow or disallow charging at their home based on willingness to pay.

In a user-preference lab study, participants simulated being an EV driver for four to 12 weeks, during which a phone-based app connected with ‘their EV’ to let them know how much charge was left on the car. The app showed, on a sliding scale, how much it would cost to charge the car now compared to later. At any given point in the day, the EV owner could decide whether or not to charge the car. The more people wanting to charge their EVs at the same time, the higher the price charged. When participants saw how much they would save by waiting until prices dropped, they opted to activate their chargers when the price was cheaper, thereby alleviating grid congestion.

Now imagine that in the future, a machine learning assistant, controlled by you and with the capacity to interact with real-time electricity markets, learns your preferences and offers you the flexibility to make choices based on your immediate preferences for cost savings, convenience or range security. Our further research is starting to show the potentially powerful ramifications of using automated decision-making information systems in the form of machine learning assistants to create the kind of flexibility and demand response that will help grid operators cope with increasing demand and supply volatility. The usefulness of automated information systems in managing the transition to sustainable energy and electrification applies not only to the rise of EVs, but extends to the integration of households that put energy they produce back into the grid.

The European Organization of DSOs (Distribution System Operators) already seem to know that preference-learning assistants and dynamic pricing are part of their future, and are preparing for volatility by favouring ‘smart-charging’ infrastructure options.

The world must continue to prepare for the transition to sustainable energy quickly and gracefully. Consumers should be buying EVs now, for any and all of the good reasons they find. We need to prepare the grid to accommodate those EVs quickly. We prefer user-oriented, market-based methods over top-down control, and we suspect you will too.