MONDAY STUDY: Electricity Needs For Electric Vehicles

Electric Vehicles at Scale – Phase 1 Analysis: High EV Adoption Impacts on the Western U.S. Power Grid

Kintner-Meyer, Sarah Davis, Dhruv Bhatnagar, Sid Sridhar, Malini Ghosal, and Shant Mahserejian, July 2020 (Pacific Northwest National Laboratory)

Executive Summary

The use of electric vehicles (EVs) in the United States has grown significantly during the last decade. So much so that the U.S. Department of Energy (DOE) asked Pacific Northwest National Laboratory (PNNL) to perform an authoritative study of the impacts of EVs at scale on the electric grid. “At scale” was defined by high-penetration scenarios performed earlier by the Electric Power Research Institute (EPRI) and the International Energy Agency (IEA). During the discussion of scope with DOE, it became clear that EVs at scale affect the electric infrastructure fundamentally in two different ways: (1) EVs affect the electric infrastructure at the point of common coupling, which for most EV charging stations (also referred to as EV Supply Equipment) is a connection to the distribution system, either at home, at a workplace, or at a public charging station; and (2) EVs at scale affect the bulk power system as an aggregated new load.

This Phase I study focuses on the bulk power electricity impacts; distribution system analysis is left for the follow-on Phase II study. Because of a sense of urgency related to performing the analysis and publishing the results, PNNL recommended that the study focus on the Western grid (i.e., the Western Electricity Coordinating Council [WECC]). The WECC already has a commonly agreed-upon data set for a future grid scenario—the WECC 2028. By using the WECC 2028 scenario, PNNL used the best available future grid scenario definition that included load growth assumptions, generation retirements and additions, as well as transmission expansions. The analysis was based on a production cost modeling approach using the ASEA Brown Boveri Gridview tool.

click to enlarge

This EV-at-scale Phase I analysis addressed the following two key questions of interest to DOE related to the impacts of EV at the bulk power level at the time when EVs are deployed at scale:

1. Are there sufficient resources in the U.S. bulk power grid to provide the electricity for charging a growing EV fleet? This question addresses the system adequacy.

2. What are the likely operational changes necessary to accommodate a growing EV fleet?

This question addresses changes in

• generation mix

• production cost

• challenges and benefits of accommodating the new EV loads.

click to enlarge

The study is unique because it represented for the first time not only the projection for light-duty but also for medium-duty and heavy-duty electric vehicles (LDVs, MDVs, and HDVs). It should be noted that this study did not include a capacity expansion analysis that searches for costoptimal investments of new grid infrastructure given the new EV loads. Instead this study focused on the resource adequacy question of high EV adoption as the WECC grid planners defined the evolution of the bulk power system to the year 2028.

Key Outcomes of the Study

Assumptions About the Penetration of EVs

This analysis applied for the following penetration assumptions for 2028 expressed as a national figure: LDVs: 24 million, MDVs: 200,000, HDVs: 150,000. The national figures were applied to the WECC footprint by a 0.4 scaling factor.

click to enlarge

Modeling of Load Profiles of LDVs, MDVs, HDVs for the 2028 Scenario

The following load profiles were established. LDV load profiles were generated by National Renewable Energy Laboratory using the EVI-Pro tool. MDV and HDV load profiles were modeled by PNNL. Load profiles are shown in the figures below.

Major Findings

2028 resource adequacy is likely to be sufficient for high EV penetration assumption.

• Under a high-penetration scenario with national electric fleets of ~24 million LDVs, 200,000 MDVs, 150,000 HDVs for a 2028 time frame, we are not expecting resource adequacy issues in the WECC under normal operating conditions (normal system, weather, and water conditions). The corresponding electric fleet sizes for the WECC footprint are 9 million LDVs, 70,000 MDVs and 94 HDV charging stations. EV resource adequacy can be doubled with managed charging strategies.

The EV resource adequacy for the entire WECC interconnection was estimated for a likely unmanaged charging scenario under which most LDVs were charging at home starting in the evening (Home High power No Delay: HHND). Unmanaged charging is predicated on arrival time at home in the evening, when we assumed that the charging process begins. The maximum number of LDVs when projected to the national fleet was about 30 million (national value) or 9 million for the WECC footprint. Alternatively, if managed charging was applied by hypothesizing a price-minimization scheme, the EV resource adequacy could be expanded to 65 million (national fleet number) or 19.6 million for the WECC. This suggests a significant opportunity to substitute additional generation and transmission requirements with smart charging strategies and much better utilization of the existing grid. Figure S.5 shows the limited resource adequacy for unmanaged and managed charging. Note, the resource adequacy limit is set when the unserved energy becomes greater than zero.

• At the maximum number of LDVs, the authors found transmission congestion to be the limiting factor, which means that there are some available power plants in the WECC but the electric power could not be delivered to the load centers because of transmission limitations. The largest transmission congestions were in California (Paths 15, 26).

click to enlarge

Operational changes can be made to accommodate EVs.

• The additional generation for charging EVs is likely to be provided by natural gas combined cycle plants and combustion turbine s predominantly throughout the WECC (85%–89% of all new generation). See hourly marginal generation for WECC by technology for two different charging scenarios in Figures S.6 and S.7.

• Storage is used in California to meet the peaks set by EVs. Hydropower generation in Washington State is redispatched to resemble a commonly observed charging/discharge cycle of an energy storage technology. No new hydropower generation is expected because hydropower generation is energy limited—no more water is expected in the Columbia River system.

• All EV charging load is likely to reduce renewable curtailments between 25% and 75% based on when EVs are charged. Managed charging could reduce the curtailment the most by an additional 16%.

• The production cost implications due to the additional load varied from 3% in Arizona, where there is some available coal generation, to 23% in California, where combustion turbines are required to meet the peak load set by EVs. It should be noted that all cost estimates were done keeping the generation capacity constant. It is likely that capacity expansion in anticipation of additional load may mitigate the cost increase, particularly, if the additional generation is renewable generation resources.

• Managed charging has significant operational benefits in solar-rich areas such as California. It reduced the duck curve in two ways: (1) it reduced the coincident peak (duck height) and (2) it reduced the ramp requirements in the evening when the sun sets (steepness of the duck’s neck).

In addition, the authors analyzed EV-at-scale impacts for Washington State using the WECC results. The results of this higher resolution analysis are as follows:

• Unmanaged EV resource adequacy for Washington State is approximately 1 million LDVs and 4,600 MDVs under normal system, weather, and water conditions. With managed (smart) charging, the resource adequacy can be increased to 2.7 million LDVs.

• Washington State hydropower resources may need to be redispatched to accommodate unmanaged EV load.

• The average production cost implications of high LDV penetration are minor and vary between 4% and 9% based on the generation mix of the utility organization.

• The authors recognize congestion in the transmission system that already exists during high loading in the winter. Congestion is likely to be exacerbated with new EV loading with unmanaged charging, because of transfers from Canada to Washington, Washington to Oregon, and eastern Washington to western Washington under normal system, weather, and water conditions.

click to enlarge

The bulk power analysis had inherent limitations.

The employed production cost modeling approach using a grid data set that represents a 2028 future grid realization comes with some inherent limitations.

The production cost model solves the generator unit commitment and economic dispatch problem given the available generators, the transmission system, and hourly loads in the WECC. It can identify system inadequacy by revealing sufficient generation and/or transmission capability to serve loads. It also provides insights into production costs and locational marginal prices on an hourly bases. However, this approach does not consider the evolution of the grid infrastructure as new investments are made. As a consequence, the production cost tend to increase with load additions as more expensive generators are being dispatched. In this analysis, the authors added new EV load beyond what the WECC members estimated in the 2028 data set to test the system adequacy question. Thus, the production cost implications are expected to be higher than if grid evolutions had been considered.

click to enlarge

The illustrative distribution system analysis offered insightful results for Phase II analysis.

An illustrative distribution system analysis was presented that demonstrated the mechanism of how to perform a distribution system analysis and what the expected results and outcomes are. This illustrative example indicated the following:

• Factors most likely to limit the additional growth of EVs are thermal overloading or reaching the rated capacity of grid assets in the distribution system under fast charging conditions.

• Voltage violations may occur under fast charging conditions that feature high ramping loads during fast charging events…

Tweet