The shorter charging times enabled by Level 3 charging, while convenient and possibly economically motivated, introduce significantly more variance in emissions results. The authors of this work believe that much more attention to DCFC load growth and charge management in terms of research, financial justification, and policy planning is warranted. In Phase I, where we focused on Levels 1 and 2 charging rates, almost all the charging events lasted several hours which helped to smooth out the ebb and flow of marginal emissions resulting from the turning on and off of peak generation resources . Level 3 fast charging systems result in short, high-power charging events, which can yield counterintuitive emissions trends, as has now been initially revealed in this Phase II study. When a charging event facilitates the transfer of large amounts of energy over a short duration of time, the timing of the event becomes even more important in terms of emissions. This is because the entire battery may be charged with relatively high-CO2 electrical power if the charging occurs during a period when the marginal resource mix is dominated by coal as opposed to a period of a less carbon-intensive mix dominated by natural gas. A high-powered charging event such as those resulting from commercial vehicles with larger battery capacities is subject to highly erratic, bidirectional swings in emissions intensity.
In the near term, planting racks in terms of electric power generation capacity and quality, the argument that more renewables will help accelerate EV adoption while mitigating environmental impacts has not been fully reconciled against space and time considerations. The large and variable rates at which electric power is called for by DCFC events present a challenge to electricity infrastructure planners and developers in identifying optimal strategies; this includes the sizing of distribution equipment, its location, and decisions around the upstream generation mix. Case in point for renewable energy projects, there is often a mismatch in terms of nameplate capacity, capacity factor, and its effective power delivery rate relative to the demands of a fleet of EVs charged at Level 3 conditions. Technological research and policies to address this mismatch require more in-depth study. As the grid decarbonizes over the longer term , the potential adverse impacts of marginal emissions from predominantly fossil fuel resources will subside. However, there will remain a need for decision-support tools to assist at many stages of this transition to a future state. More attention to firmed renewables via long-duration storage or vehicle-to-grid approaches will also be valuable. Because Level 3 Fast Charging is more likely to experience a higher magnitude and variance of emissions, incorporating battery storage infrastructure, either at the utility-scale or as smaller, distributed units, may become an attractive solution for reducing the emissions impact of vacillations between different marginal emissions regimes.
At the utility scale, the grid operator can use battery storage as a reservoir of low-carbon electrical energy for use as a marginal peaking resource. It can charge batteries during periods of low CO2, like during off-peak hours or when renewable generation shares are highest, banking low-carbon electricity and relieving part of the burden currently borne by traditional fossil resources called on to meet marginal demands. This would reduce the fossil fuel dominance of currently observed marginal mixes and therefore reduce the magnitude of vacillations between regimes. Additionally, it may be that greater investments in distributed battery storage infrastructure coupled with EV charging infrastructure could be valuable. Distributed battery storage systems could, in principle, perform a similar function as utility-scale batteries. A key difference is that they would likely be privately owned or jointly owned, suggesting a need for coordination around command and control in order to yield social benefits. Another difference is that it would require many small projects on the distributed side, compared to a few larger utility-scale projects. This can have advantages, such as a reduced need for new transmission investments, but also challenges, such as the need to build out intelligent distribution infrastructure to support it.
Incentivizing private battery storage may be a cost effective method for shifting demand on the electrical power grid off of peak hours, improving the utility’s ability to effectively manage increasing load due to growing EV shares and activity. Further investigation is required to explore the operational nuances and cost components of battery storage systems as described here, but as far as this research can conclude it may be an effective strategy for tackling the fast-charging problem. A fourth, potentially critical, policy implication is the notion that vehicle electrification will not happen in a vacuum as far as the grid is concerned. Thus far in our study, we’ve exclusively looked at new heavy electric loads as additive to existing and future demands. Furthermore, it is also reasonable to assume that most of the electricity generation from renewables will essentially be fully consumed by so-called “baseline” demands for the foreseeable future. In other words, renewables still account for a modest enough share of the total mix that they will be consumed, with or without any EV growth at all. In fact, with the increasing pressure to retire coal plants, much new low-carbon generation is needed to simply offset those resources. However, if EVs could be deployed within a broader frame, that could go a long way toward reducing uncertainties raised by marginal emissions scenarios. Such a broader frame would manage demand, intelligently control EV charging, prioritize overall efficiency gains, and focus on conservation, avoidance, or substitution. In this way, the assumption of average hourly mix or even average daily mix might be more relevant than the uncertainty around a specific marginal resource assumed to meet the incremental kWh required by a particular EV charging session. While the study offers some suggestions for tools and next steps during the near-term and transition period, this broader framing seems to be a difficult hypothesis to test with certainty over the longer term. In addition, it seems there is talk of greater electrification, not less, when it comes to other sectors like residential heat pumps, data centers, industrial heat, or other energy-intensive processes that currently use thermal methods such as fossil fuels. While broader interactions across electric power use segments are beyond the scope of this study, the potential policy implications of those interactions and factors could be profound. For now, it may be reasonable to assert that as EVs are deployed, it is imperative to not only manage the EV charging events in time and space but also consider our latitude to control or influence other large loads on the grid in conjunction with EV deployment growth. Doing so can help ensure that the electrification of transportation results in meaningful decarbonization gains. This Phase II study has built upon a Phase I study that developed a systems methodology to explore important questions about EV growth relative to new vehicle categories and use cases. This report pursued “future work” that was identified during the first research investigation which focused exclusively on light-duty passenger vehicles. The now “present work” has specifically explored fleet and commercial use cases involving medium and heavy-duty vehicles and augmented the findings in depth and breadth. Our study has demonstrated the usefulness of the methodology developed in Phase I, pipp horticulture which integrated vehicle power train, charging profile and grid generation mix sub-systems. The present study investigates an enhanced understanding of MD/HD EV emissions and several promising scenarios and use cases that can help optimize charging schedules and minimize CO2 emissions.
In both phases of our study, we have proposed and investigated rigorous approaches that estimate the variability associated with CO2 and other emissions involving electric vehicles. This required a simulation framework that explored multiple parameters concurrently to yield broad comparisons. In this way, we explore a growing set of electric vehicles as compared to a baseline case . We explore a representative and diverse suite of use cases, driving cycles, and charging profiles for a range of users, including individual commuters, fleets, small businesses, as well as municipal transit and services. The simulations estimate CO2 emissions under a range of scenarios useful to inform decisions, investment, and policy. As noted, prior studies often utilize annualized averages for grid-level C02 emissions to simplify the analysis. In our research review of other tools and dashboards , we observed that a very basic algorithm is typically utilized that does not consider time of day or seasons of the year. We acknowledge such traditional approaches provide a kind of first-order, initial estimation that can be useful to some audiences in some contexts. However, it is imperative to recognize and explain the limitations of this approach, and the risk of relying too heavily on average emissions estimates. The reason is that such estimates are subject to change in the future, and also subject to variability during the present on multiple timescales . This Phase II effort emphasizes the need to focus on Level 3 Fast Charging because this subcategory of charging stands to incur higher rates and potential uncertainty. Not only will better assumptions be needed to estimate emissions resulting from Level 3 charging, but they will also be imperative to inform infrastructure siting to build out charging networks and inform resource planning for the grid at large.Next steps should consider the benefits, tradeoffs, assumptions, and limitations associated with the methodology, practicality, and intent of related research. The authors believe continued attention, in particular during the near-term transition period, can facilitate more direct comparisons of EVs and use cases to other technologies as penetration rates grow. Several summary statements emerge from this body of work. It is clear that at certain modest levels of EV deployment, a weighted average mix of resources may not be illogical or inaccurate in estimating CO2 impacts. However, it can be stated that with significant increases in EV charging, in particular at certain hours of the day and seasons of the year, the assumption of weighted mixes breaks down. The study demonstrates that the breakdown can be quite pronounced for use cases involving greater vehicle miles traveled, for charging sessions occurring in the field, and for charging occurring during periods of peak grid demand. The breakdown also seems pronounced for scenarios involving Level 3 charging. These findings reveal that managing charging events throughout the 24 hours of the day and across LD, MD, and HD use cases in distinct ways should merit greater attention. Furthermore, charge management alone will likely be inadequate as EV shares grow to much greater levels. Future study is anticipated to further inform decision-making around near and intermediate term scenarios including new interactive methods of forecasting both EV demand and grid resources. Historical approaches to dispatch are already underway toward predictive forecasting approaches. Ideally, scenarios will be developed that can better simulate future resources both fossil and non-fossil in order to meet load growth to support electric transportation, as well as additional demand growth from the electrification of other sectors like data centers, heating and cooling, and industrial processes.Together with its Phase I counterpart, this Phase II study explores a systems-of-systems methodology to simulate viable grid-charging-vehicle scenarios of increasing interest for planning and policy-making. Collectively, our team has considered a range of vehicle categories and use cases . The Phase II effort in particular has refined the methods introduced in Phase I and deepened our understanding of several potentially compelling EV applications including electric pickup trucks used by small businesses in service-oriented urban applications, medium-duty fleets, and other specialized uses where vehicles have predictable routes and return to base on a regular basis. The study’s simulation reveals that the CO2 emissions intensity of a battery-electric light truck traveling 20 miles per day could vary dramatically depending on the charging schedule used. Under the study’s marginal resource X grid condition where a specific marginal resource is needed on an hourly basis to meet a particular EV charging event, estimated CO2 emissions could be as much as 42% lower than a baseline ICEV, or as much as 24% higher than the same baseline. This large variance is purely a function of when and how quickly the vehicle is recharged. Thus, this example suggests such tools will be important to ensure environmental benefits are realized. While the simulated use cases yield valuable guidance in their own right, the collective work reveals that such modeling and simulation-based comparisons are generalizable and extendable.