Part three of a multi-part series discussing the shortcomings of the recent NEVI final rule and what that may mean for EV charging in coming years.
In my first NEVI article I discussed how the NEVI final rule, the most important set of government standards for determining how EV infrastructure is deployed along the nation’s corridors, disregarded the issue of cord amperage and in this way missed the opportunity to set a real baseline for what constitutes ‘fast’ charging. In my second article, I explained that the NEVI-specified minimum site size is too small to serve even current demand and that incentivizing “make-readies” would have helped ensure sites are equipped for easy expansion to meet inevitable future demand. Here I put aside the question of future efficiencies that could have been captured and stick to analyzing the constraints the NEVI places on the efficient use of even the power that is supposed to be available to EV drivers today.
The problem starts with the NEVI’s undue emphasis on charging speed ratings (which, as I have discussed, are more theoretical than actual because the NEVI does not account for the voltage of the typical EV and the amperage of the cord). The first principle to understand when discussing speed ratings is that they specify the maximum charging speed, not the average charging speed for a vehicle over a charge session. That matters because, for technical reasons I get into below, a car battery can charge at or near the maximum charging speed for only a relatively brief portion of the period required to get a full charge. In fact, most EVs today spend a relatively small amount of time charging at a speed above 100kW, let alone above 150kW. The efficient response to this reality would be to allow chargers to dynamically share power between multiple cars, so that at any given point in the charge session the charger dedicates to a car only as much power the battery can take and reallocates the rest.
But one of the NEVI’s principal requirements is that each of the minimum four ports at a station must be capable of delivering 150kW simultaneously for the whole duration of a charge session.
The provision is well-intentioned and presumably designed to ensure a fast and consistent charging experience wherever an EV goes. But what it signifies in practice is that for a lot of the charging time, much of the hardware and grid capacity that is dedicated to each charger will go unused. Thus sites that adhere to the NEVI’s minimum requirements standards as written will experience wasted power capacity and unnecessarily high hardware, construction, and electricity costs.
First things first: car batteries charge on a curve. In fact, depending on the car, they charge on many different curves, as reflected in the below chart (from P3 Group):
Many factors affect charging speed, but the most important to consider is the fundamental chemistry of lithium ion batteries. As shown in the chart, whatever a lithium battery’s charging curve, generally the lower the battery’s state of charge (SOC), the more power it can accept. In plain English, the deader the battery, the faster it charges. As the battery becomes more full, the rate of charge slows down. A simple way to picture the charging process is as follows. The lithium ions need to move from one electrode in the battery to the other. But those ions need to find cavities in the electrodes to fit into. When the battery is at a low (SOC) most of these cavities are empty so it is easy for many ions to find a home at the same time. The fuller the battery becomes, the more “searching” each ion has to do to find an available spot, with those very last spots at nearly 100% SOC taking quite a long time to find. As you might notice with your laptop or phone, going from 10% to 20% charge typically happens very quickly, whereas 90% to 100% takes far longer. The same principles apply to the chemically similar, far larger lithium batteries in electric vehicles.
As can be seen in the charge curve graph above, maximum charge rate can typically only be maintained until around 40-50% SOC, after which the maximum power the battery can take tapers off gradually. It is for this reason that charge times are often quoted from 10%-80% SOC rather than going to 100%, because that last 20% can take as long as the previous 70%, and in practice drivers are unlikely to wait around that long.
There are other factors that impact charging speed, many of them temperature-related. The curves above represent a battery and charger in ideal conditions. Lithium batteries are sensitive to temperatures that are too high or too low, and so being at either extreme can impact charge rate. Like humans, batteries tend to work best in the 70-80 °F range. The battery management system (BMS) that controls charging in an EV knows this and is constantly balancing between keeping the battery at the right temperature and using up energy to do so. Most EVs are capable of actively heating and cooling their batteries to bring them to the ideal temperature, but this uses a significant amount of energy, so batteries are often allowed to stray outside of the ideal range to save energy. Some vehicles, like Teslas, feature what is called “smart preconditioning,” where the car recognizes when you have input a charging station as a destination and brings the battery up to temp just in time to charge, trading a bit of energy for a significant time savings once you start charging. If a car arrives at a charging station with a very cold battery, it might spend 10-20 minutes charging at only 10kW while the battery is brought up to temperature. Similarly, the charge rate suffers at the upper end of the temperature spectrum, and on hot days you can often hear the vehicle’s fans running at full blast to keep the battery cool and maintain charge rate. In fact, Rivian just patented chargers with fans to blow cool air vaguely in the direction of the vehicle to enable faster charging.
All that is to say that there are myriad reasons why a vehicle at a busy four-port charging station will not be able to accept the full power that the charger is capable of delivering for the full duration of the charge session. So the energy it isn’t using could theoretically be used by any of the other three cars plugged into the station. But the NEVI doesn’t permit that kind of sharing except where the available power exceeds 150kW.
Tesla enabled power sharing between pairs of chargers at the inception of its Supercharger program because it recognized that it was unlikely that two cars would each be requesting power from a charger at the then-100kW max at the same time, or for very long. These chargers were designed to dynamically share 150kW total between two charge posts (rather than, say, capping the charge posts at 75kW each). As the batteries got larger and chemistry improved to allow them to charge faster, Tesla moved to V2 Superchargers (150kW split into two 120kW posts), then to V3. V3 actually represented a step change in power sharing: not just power sharing between posts fed by a cabinet, but between multiple cabinets at a particular site.
Each 350kW charge cabinet was directly connected to three or four charge posts, each charge post had a 250kW max. But these multiple cabinets could be connected by a common DC bus, meaning that in practice the power was shared between all posts at the site, not just between the ones connected to a particular cabinet. To play this scenario out, an eight-stall site could have charge posts that would each be capable of delivering 250kW reliably, but instead of having a connected load of 2mW as would be required if those were individual 250kW chargers, only 700kW would be required from the utility. This allows for a smaller transformer, as well as smaller switchgear and conduit. This also means that the utility is more likely to be able to provide the necessary power without major (sometimes infeasible) modifications of the existing electrical distribution infrastructure. Put another way, with this power-sharing set-up, for a given electrical load you could install roughly three times the number of chargers that in practice would be about as fast (for all the battery-chemistry reasons stated above), on the same electrical connection.
The same general principle applies at a more bare-bones, NEVI-compliant, four-port site. NEVI requires a charge capacity of at least 600kW at every site for those four 150kW ports. A site with the same charge capacity but five or more ports is not compliant under the NEVI and therefore ineligible for federal funds. That’s a shame, because a few additional ports would allow more vehicles to be serviced at a given time, without significantly slowing down charging for any particular vehicle.
The NEVI-mandated installation of DC fast chargers that do not share power (at least below the quite-high threshold of 150kW) impacts more than the number of EVs that can be served or the size of transformers, switchgear and conduit. It impacts the cost of electricity itself. That’s because two potential cost-drivers on a utility bill are the capacity and demand charges. Put simply, capacity is the maximum power you have the ability to use, demand is the maximum instantaneous power you actually use. Having dedicated 150kW chargers necessarily maximizes capacity charges, and has the potential to increase demand charges for very little benefit. This means that even for the same amount of electricity, used at roughly the same time of day, in a power-sharing scenario the electricity bill could be significantly lower. These savings could be passed on to the consumer.
During the NEVI notice and comment period, commenters highlighted the many potential benefits of accommodating power sharing. But the official response, reflected in the final rule, was that “the requirement that each DCFC must simultaneously deliver up to 150 kW, as requested by an EV, was retained as a minimum requirement to provide a standard, reasonably high level of charging service.” This stance is particularly frustrating considering that, as I covered in my first article, that “high level” of charging service is not guaranteed for the majority of EVs by the NEVI standards because of the agency’s decision to ignore certain actually critical determinants of charging speed, like cord amperage.
And so we find ourselves in a scenario where 150kW of dedicated power has become the de facto standard for the building of fast chargers, despite being of dubious value in practice–that is, as experienced by current EV owners who will use that dedicated power for a fraction of their charging session (if at all). Allowing more dynamic power sharing would allow for more chargers to be built more quickly on available power, and would reduce not only construction costs but operating costs as well. Sharing available power over a larger number of chargers also helps alleviate drivers’ biggest complaint with the charging experience and the subject of my next NEVI article: reliability.