Professor Lisa Aultman-Hall and Jonathan Dowds April 2019
University of Vermont, Transportation Research Center
In partnership with Dr. Paul Hines
This project will model hourly electricity demand for plug-in electric vehicle (PEV) charging using survey-derived trip-making data for a 7-state region comprised of New York and New England. Different PEV demand scenarios will be used in a regional electric power dispatch model to estimate generation costs and emissions. Trip lengths as well as duration at destinations will be used to model both at-home and away-from-home charging to generate hourly electricity demand from a future passenger vehicle fleet. Prior research on the grid impacts of vehicle electrification (conducted between 2006 and 2010) concluded that overnight vehicle charging, coinciding with periods of low existing electricity demand, offered significant benefits for management of the electric grid. However, in the last decade, the rapid deployment of solar generation has resulted in substantial changes in net load (electricity demand less wind and solar generation) throughout the day, resulting in idle generating capacity during daytime hours and increasing the rate at which fossil fueled electricity plants may need to ramp up generation through the late afternoon and evening. Mid-day generation from solar sources may represent a need to increase daytime vehicle charging that was not previously expected and which may require significantly more away-from-home charging than previously assumed.
Many early efforts to reduce greenhouse gas emissions (GHG) focused on achieving nearer term emissions goals in the electricity sector while assuming that significant emissions reductions in other sectors, especially the transportation sector, would require a longer timeframe to achieve. Reflecting this thinking, the first GHG cap-and-trade system in the United States, the Regional Greenhouse Gas Initiative (RGGI), was initiated in 2009 in ten northeast states, and applied exclusively to electric power generation. The first phase of the Western Climate Initiative (WCI) cap-and-trade program, now including California and Quebec, was initiated in 2013, and was limited to power generation and combustion at industrial facilities. In more recent years, a greater emphasis has been placed on achieving GHG emissions reductions in the transportation sector as its relative share of total emissions has increased to 28%. Moving forward, achieving significant GHG reductions in the transportation sector is widely expected to require significant adoption of plug-in electric vehicles (PEV), which made up 3% of new vehicle sales in the United States in 2018. Previous emissions modeling analyses focused on the electric power sector has historically failed to incorporate details of the transportation system, diverse travel patterns, as well as temporal details of when during the day vehicles are being driven and when they can be charged. Given temporal peaks in non-transportation demand for electricity and the increasing role of renewable electricity generation including its variable availability, linking models of the electricity demand and generation to more disaggregate transportation demand models is critical.
Travel demand models (Figure 1) are used to forecast key travel behaviors and resultant system performance such as vehicle miles of travel (VMT) by road link, congestion, travel speeds, delays and emissions at hourly, peak period or daily timeframes. Travel demand models require data about the spatial distribution of land uses and thus origins and destinations, but also household trip-making rates, mode choices, and trip length distributions as inputs. Information describing trip-making behaviors have typically been derived from travel survey data such as that collected by the National Household Travel Survey (NHTS), although use of mobile device-derived trip data and other “big data” sources are seeing increased use. Travel demand models are typically undertaken for metropolitan regions and in some cases entire states.
Figure 1: Research Context: Integrating Travel and Power Models
Electric power sector economic dispatch models most often operate at the jurisdictional boundaries of Independent System Operators (ISOs), the entities that manage regional grids and wholesale electricity markets. ISOs do not necessarily align with state boundaries and frequently include multiple states. ISO-NE and NYISO operate the grids and energy markets for New England and New York respectively. The dispatch models allocate power generation at a given time to a set of available generating facilities to ensure that power supply matches power demand at all times at minimum possible cost, subject to a variety of technical and regulatory constraints. Inputs for dispatch models include the set of generating facilities (and in some cases options for adding generating capacity) available in the region, emissions caps if applicable, key characteristics of the facilities such as fuel type, and combustion efficiency as well as the time series of electricity demand in the region. Dispatch models output the generation by each plant for each time period (often hourly) as well as total and marginal generating costs, including the cost of GHG emissions when these emissions are capped.
The overall goal of this project is to utilize detailed travel behavior data to calculate time-specific regional demand for vehicle charging and input this high-resolution PEV electricity demand into a dispatch model for New York and New England. This integration will facilitate the assessment of the combined impact of increased PEV charging and renewable electricity generation on the power sector as well as the total GHG emissions. Results for a range of PEV penetration levels and charging behaviors will be unique in that they span across both the transportation and electricity sectors.
Policy decisions about the level of investment to make in charging infrastructure and how to target these investments requires integrated modeling of both travel behavior and electric power generation to realistically capture the impact of vehicle charging on grid operations, GHG emissions and costs. The team will sample NHTS vehicle and trip data from a northeast study region consisting of the states of Connecticut, Maine, Massachusetts, New Hampshire, New York, Rhode Island, and Vermont. Vehicle type, daily trip sequences and household attributes will be used to estimate the relative probability that a PEV would be used for the observed daily travel profile in the future. In other words, we assume travel demand or patterns remain constant and predict which travel may be undertaken with a PEV when PEV penetration is higher. Trip data will be considered as a series of stops that represent the times of day when the vehicle might charge based on the probability of charging availability as a function of trip purpose and destination location. Time of day charging demand corresponding to several PEV penetration scenarios will be aggregated as input in the regional electricity dispatch model. The outputs of the dispatch model will allow for the calculation of electricity generation costs and net GHG emissions changes across the transportation and electricity sectors.
In the first year of this project, the research team will develop a disaggregate charging demand model from the vehicle travel data in the NHTS. The timing of vehicle charging in the year 1 model is intended to reflect charging behavior with increased PEV market penetration but in the absence of pricing or policy constraints and thus will be considered the “unmanaged charging scenarios”. This charging demand model will allow the research team to address the following research questions:
- What is the expected time-resolved daily electricity demand from electric vehicle charging in New England and New York given different PEV penetration rates?
- How is the time-specific electricity demand for vehicle charging impacted by different levels of daytime, away-from-home PEV charging and levels of charging infrastructure availability?
The research team will also incorporate solar generation into an existing, hourly electric dispatch model for New England and New York and run the dispatch model using baseline electricity demand for the region plus the time-resolved charging demand described above. Dispatch modeling in year 1 will be conducted using only an electric-power sector GHG cap-and-trade constraint modeled on the RGGI program. Output from the electric grid dispatch model will enable the research team to answer the following research question:
- How does unmanaged PEV charging impact the costs and GHG emissions for the NY and New England electric power system given solar generating capacity and different PEV penetration levels with various charging scenarios?
Output measures will include the total cost of electricity generation, the marginal cost of electricity generation (which is a reasonable proxy for wholesale electricity prices), and the cost of complying with GHG cap-and-trade relative to a baseline with only home-based charging of PEVs.
In year 2 of the project, the modeling will be expanded to include optimization of the timing of vehicle charging endogenously in the dispatch model, subject to the constraint that all vehicles are charged in a manner compatible with the user’s travel needs. Modeling in year 2 will also incorporate a representation of a multi-sector GHG cap-and-trade program covering both the transportation and electric power sectors and enable the research team to answer the following research questions:
- How does managed PEV charging impact the electric power generation and marginal costs as well as the GHG emissions for different levels of PEV penetration and different levels of solar generating capacity?
- How does a single-sector versus multi-sector cap-and-trade system affect the costs of complying with GHG caps?
Funding: In 2019, the preliminary scenarios described above are supported by the National Center for Sustainable Transportation (NCST), a USDOT UTC program at the University of California Davis as well as a grant from the Social Science and Humanities Research Council of Canada to the Joint Clean Climate Transport Research Partnership (JCCTRP) at UMAQ. Part of this work forms the basis of the Master’s thesis of Ms. Sarah Howerter.