Overcoming accuracy anxiety: Smarter data to build EV response programmes

Electricity companies are taking varied approaches to charging and demand response programmes when it comes to supporting today’s electric vehicle (EV) revolution.

Some are focused on building charging infrastructure. Some are developing managed charging programmes. Others are focused on optimising existing grid structures by way of behavioural load shifting.

While choosing the most appropriate approach is heavily dictated by service area, customer needs and regulatory environment, they all rely on a foundational need to know not only who drives an EV but also when and for how long vehicles are being charged.

Unfortunately, this information is something many utilities either struggle to access, or have limited insight into. And, that’s a huge reason why EV programmes have been hard for utilities to put into action, despite much talk to do so.

Where does EV data come from?

Historically in the US, utilities have relied on their motor vehicle department (DMV) to provide EV ownership data based on vehicle registrations within the territory.

DMV data relies on the owner re-registering their car each time they move. Because this is something many fail to do, there can be more or less EVs in the utility’s territory than listed in the DMV database.

Even if the customer is only moving houses within the same zip code their EV load may now be on a new feeder, resulting in unknown EV load on grid assets. Inaccuracies even at this level can cause major disruptions to a utility’s EV programme.

And, that’s only half the equation. There’s still a black hole around charging that the DMV does not have. Is the customer charging during peak times? Do they charge once a week or every night? Do they have a Level 1 or Level 2 charger? 

Customer-provided methods to collect charging intel, like telematics, exist but even those only provide a limited view of EV activity. Since it requires drivers to proactively opt-in to monitoring programmes, often accompanied with the installation of intrusive and costly monitoring hardware (like plug-level sensors or current transformer clamps), not every driver participates. So again, utilities may have some idea of the charging in its territory, but it’s not the complete picture.

Have you read?
New study aims to test benefits of bidirectional EV charging
French heavy EV charging to reach 3.5TWh annual demand by 2035

As utilities begin designing EV programmes, having that complete picture is what will ensure these programmes are built to support all EV loads across the entire grid, with the least amount of false positives.

Overcoming accuracy anxiety

With EVs changing the grid, and more so in the decades to come, getting EV intelligence programmes off the ground now is imperative. But without a solid foundation of accurate data it is no wonder utilities are hesitant to invest in these.

How can utilities effectively manage EV programmes if they can’t even see the EVs on its grid?

Fortunately, utilities have a better solution than relying on DMV data for their programmes. A solution most already have in their hands.

Smart meter data coupled with AI-powered analytics is providing an easier, faster and more reliable option for collecting EV ownership and charging intelligence. Best of all smart meter data is something many utilities already have.

The advanced metering infrastructure of 100 million smart meters is currently deployed across the US, covering 75 percent of all households and giving utilities insight into how much energy is being used in any given home. Layering on AI-powered analytics, utilities can break down exactly how that energy is being used. 

This includes heating, air conditioning, refrigeration, lighting – and you guessed it, EV charging. EVs have very distinctive charging signatures when extracted from smart meter data. In addition to consuming large contiguous blocks of energy, EV’s generally exhibit a clear pattern of sloping decay toward the end of charging. AI detects this, letting utilities know when and for how long EV charging occurred. Collecting this information over the course of the past week, month and even year, utilities are able to identify ongoing patterns, isolate areas of the grid most impacted and more accurately forecast future demand.

Now, without having to ask the DMV, utilities can gather addresses for all of the customers in their territory who are charging EVs in their homes. This same data set also exposes relative intelligence essential to the management of EV loads, including: differentiation of Level 1, 2, and 3 chargers; average hourly charging patterns; geographic patterns of EV charging and growth; and amplitude of chargers.

EV intelligence of this granularity gives utilities peace of mind knowing they can implement EV programmes and manage grid planning based on real-time supply and demand. The fear of wasting time and money on the unknown is eliminated.

For example, utilities can contextualize how an increase in EVs would impact its existing feeder mapping. Can the grid withstand additional EVs or should the utility consider reconfiguration? Should different customer sets be moved to a different feeder to optimise grid performance in that specific neighbourhood?

This intelligence also enables utilities to more accurately engage with EV drivers. By incorporating personalised information about their individual charging patterns, utilities can better motivate EV owners to shift charging to off-peak times. Knowing how much money they spend on charging during peak times versus the amount they could save by charging at an alternative time is more effective at motivating change than generic, mass marketing messages.  

AI-powered smart meter data

Utilities are acutely aware that understanding how EVs charge on the grid is the only way to properly manage peaking energy loads, particularly as EV adoption continues to rise. Smart meters are a gold mine of information, and by applying AI-powered analytics utilities are able to unearth valuable consumption insights from the meter up rather than the transformer down.

By ditching the DMV in favour of AI-driven smart meter data, utilities are better equipped to build data-driven EV programmes with greater accuracy and success. 

About the author:
Maria Kretzing, Bidgely’s General Manager of Electric Vehicles and Analytics, leads innovation, product development and go-to-market strategies for EVs, grid analytics and decarbonisation solutions. She started her career in the generation space before migrating to work on utility customer engagement and creating solutions for a flexible energy future.