Work Smarter, Not Harder
Agent-Based Modelling enables us to unlock the benefits of EV smart charging
Road transport is being electrified. This increasing load on national electricity grids around the world is going to be a problem, amplified by our increasing reliance on variable renewable energy supply. We introduced this problem in a previous blog, highlighting the need to reduce the number of trips people need to take by car, and to shift the charging patterns of those who own and use electric vehicles (EVs).
In the City Modelling Lab, we leverage Agent-Based Models to target reductions in car trips. Using results from our BatSim tool, we can now also simulate the extent to which EV charging patterns can be shifted, in both time and space. By analysing the simulated plug-in behaviour of individual EV owners, we can quantify the value of smart charging. Read on to find out more!
TL;DR: There is a lot of flexibility in when to charge EVs, given how long they are parked when plugged in. By using Agent-Based Modelling outputs to understand where and at whom to target policy and technology interventions, we can help unlock the benefits of smart charging.
NOTE: All results we visualise in this article are illustrative. They are generated from artificial data that aligns with our analysis using confidential client data.
What is smart EV charging?
Given complete freedom, we are most likely to plug in our EVs to charge at the maximum rate that the plug allows. If you plug in to a socket at home, this rate might only be 2–3 kW. If you have a dedicated charger, it could reach 7 or even 22 kW. If all 33.6 million cars in the UK were EVs that plugged in at the same time, an additional 67 - 740 GW would be added to the country’s peak electricity demand. The UK electricity network can currently meet a peak of 76.7 GW, if renewable energy is producing at its maximum. Clearly, something must give.
Thankfully, our BatSim modelling indicates that the activities in which we are most likely to charge our cars are often the longest in our day; very few of our charging events are likely to happen part way through a journey (enroute charging), when we’re time constrained. Therefore, our cars are parked and plugged in a lot longer than they need to charge. Our daily plans also differ — amongst private EV owners and between private and commercial vehicle fleets — meaning there are cars parked and plugged in at all times of the day.
Smart charging involves shaping the charging schedule of a plugged-in EV to meet external objectives. EV smart charging is expected to be a key component of our future network. Indeed, private chargers installed in the UK since June 2022 have mandated inbuilt smart charging capabilities. Homeowners might leverage smart charging to maximise use of their own rooftop solar panels, charging only in the middle of the day. They might also opt to trickle charge their EV overnight to maintain the car’s battery health, as we do with our mobile phones.
Smart charging is most valuable when it’s used to shape the charging schedules of millions of EVs in a country to meet a common aim. It can be used to minimise peak electricity demand, taking pressure off the transmission network. It can also be used to better align demand with available renewable electricity supply.
At the Lab, we have simulated the daily plans and charging preferences of millions of future EV owners to quantify how powerful smart charging can be. What’s more, we have done so within the constraint of EVs having to already be plugged in. That is, we can demonstrate the value of smart charging without inconveniencing EV owners. We even enforce the conservative rule that cars must be fully charged 80% of the way through the activity in which they’re plugged in, so that we account for the potential of people wanting to get away sooner than we have simulated. Even with these self-imposed constraints in place, smart charging has an enormous potential to contribute to a stable electricity system. Our focus on the individual also enables us to quantify its value when applied across sub-regions and sub-populations.
Reducing electricity demand peak load
Electricity transmission and distribution networks have limits on the amount of power they can move around. To increase those limits, substantial investment is required. Balancing existing constraints in the network requires us to rely heavily on power plants fuelled by natural gas, to the tune of more than £1bn a year. Uncontrolled EV charging will push network constraints further; smart charging could turn the tide by contributing to the balancing market.
By focussing smart charging appropriately, we can drastically reduce peak demand. We achieve this by filling the overnight troughs in existing electricity demand with EV charging. This strategy can be localised where network constraints are most acute (we’ll discuss this more later), leaving a part of the EV fleet to focus on minimising the electricity supply-demand gap.
Aligning demand with supply
In every instant of every day, there needs to be enough electricity supply to meet demand. In 2022, it cost £1.2bn to ensure the right amount of electricity was available to meet demand in the UK. If there’s too much, “non-dispatchable” producers (usually of renewable power) need to be paid to switch off. If there’s too little, expensive “dispatchable” suppliers (usually fossil fuel power stations) need to be paid to switch on.
EVs can be used to manage this balance. By focussing smart charging on balancing renewable energy supply in each day of the year, electricity system operators can maximise the output of renewables and reduce reliance on expensive, dispatchable supply.
National governments mandate how often in a year the system can fail to maintain this balance. One measure of this is the Loss of Load Expectation — LOLE — the average number of hours in a year that we expect there to be insufficient supply. In the UK, a LOLE threshold of three hours is used in system adequacy planning (99.765% uptime). In Ireland it is eight hours. We can direct EV smart charging as a first line of defence to reduce LOLE, which mitigates the need to rely on expensive options like dispatchable supply.
Place-based solutions
EV charging varies geographically, depending on the number of EV owners, the chargepoints they have available, and how they use their cars. We build our charging demand schedules from individual EV profiles, so we can pinpoint those locations where smart charging measures could be most successful. We can also vary smart charging strategies geographically, to meet local requirements. For instance, if the electricity network is particularly constrained in one area, we can prioritise smart charging that minimises peak demand in preference of meeting available renewable energy supply.
People make the profile
It’s the aggregation of millions of EV owners’ charging schedules that allows us to mitigate electricity network and balancing constraints. However, we cannot expect all EV owners to completely buy in to smart charging strategies. For instance, if smart charging is implemented with financial incentives and penalties, then the price sensitivity of individuals or businesses will impact its effectiveness.
We can use EV owner attributes to identify the key target sub-populations whose buy-in will maximise the effectiveness of smart charging. There are many ways we can slice our modelled EV owners, including:
- Vehicle use type (private car, light-duty commercial vehicle, buses, etc.).
- Owner socio-demographics (age, income brackets, etc.).
- Charge point type / location (work, shopping, education, home on- and off-street, commercial depot).
These sub-populations affect overall EV demand to varying extents and at different times of day. For example, commercial vehicles are more likely to be plugged in during the daytime, so their charging can be shifted to align with solar electricity supply. Private vehicles contribute more to overall demand and are more likely to amplify the existing peak demand, so shifting their demand overnight can really help manage network constraints. Although we have found that relatively few EV owners will charge while out and about — at work, shopping, etc. — those that do can have their demand pushed away from the morning peak to the early afternoon, when a slight dip in non-EV demand tends to occur.
We already build these attributes into our ABM populations and use them to conduct equity assessments of transport interventions, including disparities in exposure to air pollution and differences in travel behaviour between genders. Indeed, not only can we maximise smart-charging effectiveness with these attributes, we can also demonstrate how to do so equitably.
Summary
Here at the City Modelling Lab, we build our understanding of transport systems from the bottom up. This allows us to pull levers in our simulations that others cannot: who is using an EV, where they are plugging it in, what charging infrastructure they have available, and to what extent they might engage with smart charging strategies.
With our approach, we can provide the depth and breadth of insight required to unlock the benefits of EV smart charging. And there are clear benefits. We may not be the only ones banging this drum, but we believe that our insights into the daily interactions of individuals with infrastructure, policy interventions, and each other are essential to the conversation.
More work still needs to be done. We plan to explore how to achieve even better results by influencing when in their day EV owners plug in, by implementing price incentives alongside changes in the availability of charging infrastructure. We also need to improve our understanding of how price-sensitivity differs across sub-populations, and to what extent we can incentivise smart charging whilst keeping it cheaper than other options to balance the system (dispatchable supply, batteries, etc.). We’re excited to explore this further and welcome opportunities to collaborate!
