2A: Tranportation Moderator: Margaret Taylor, Lawrence Berkeley National Lab

John Anderson, Center for Sustainable Energy Identifying areas with high proclivity to adopt electric vehicles Researchers focused on clean transportation have produced a large body of knowledge about factors that predict the likelihood of (EV) adoption. The literature routinely informs policy aimed at reducing barriers to EV adoption, as well as for targeting and tailoring outreach and education aimed at “adding fuel to the fire,” or increasing adoption among consumers already predisposed to adopting electric vehicles. Consumer attributes thought to be related to higher likelihood of adoption include single-family home ownership, multiple vehicles in a household, higher income levels, high levels of EV adoption among neighbors, and prior adoption of clean technologies like solar. The authors build on these identified attributes, and use data from a variety of sources — including California’s Clean Vehicle Rebate Project (CVRP) Consumer Survey — to identify census tracts with high proclivity for electric vehicle adoption. These results will inform an interactive tool for use by EV stakeholders. Results will be useful for identifying areas with high potential for EV adoption, especially in areas where adoption levels are presently low. These results will also provide CVRP outreach implementation teams with a tool for identifying areas where their outreach efforts might have high impacts on increasing adoption.

Alec Beall, The University of British Columbia A carbon price by another name may seem sweeter: Consumers prefer upstream offsets to equivalent downstream taxes Consumers are influenced not only by prices, but also by how those prices are labelled. Prices for carbon emissions can be framed in a variety of ways, such as carbon "taxes", "permits", or "offsets." Furthermore, the emissions can be regulated at many different points in the production and usage system: "upstream" regulations are applied to the extraction and importation of fossil fuels, while "downstream" regulations are applied to the sale of products and services. From a conventional economic standpoint, these points of regulation should have roughly equivalent impacts on carbon emissions. However, the impact of "upstream" vs "downstream" frames on consumer perceptions and preferences is largely unknown. This talk presents data from three studies examining U.S. consumer preferences in the airline industry (N = 1097). In all three studies, participants were presented with several scenarios in which they were asked to choose between two ostensibly identical flights for purchase (e.g., two flights to little-known Caribbean islands): One was a flight that carried a $14.00 carbon fee and the other was a flight that did not carry a $14.00 carbon fee. Across all three studies, consumers reported being more likely to purchase a flight that carried a $14.00 carbon fee when that fee was labelled as an “upstream offset” (a “carbon offset on aviation fuel production and importation”) than when it was framed in other ways (i.e., as a “carbon tax” or “carbon permit”) or if the point of regulation of the fee was downstream (i.e., “…on airplane travel”). Strikingly, individuals in the “upstream offset” labelling condition were actually no less likely to prefer flights carrying a carbon fee when compared to a control condition in which no description was given and the $14.00 carbon fee was not even applied. These framing differences were moderated by political ideology, such that Republicans show a particular distaste for downstream taxes. Countries or states that wish to enact a carbon fee may want to use the "upstream offset" frame, especially if competing with other countries without a carbon fee. Our results suggest that consumers may in fact prefer airline flights with an upstream carbon offset; this preference may be strong enough to counteract any additional cost to the country or state that implements it; and the implementing country could potentially realize further benefits if the offset investment helps finance sustainable low- carbon development in that country. Furthermore, aviation consumers might be more accepting of "upstream offset" regulation than "downstream tax" regulation. In fact, our findings suggest that customers may be more willing to purchase tickets that include appropriately described carbon offsets, even if the cost is higher. Other implications and future directions for this work are discussed.

Mersiha McClaren, Research Into Action, Inc. Race to the Bottom: Using Advanced Analytics, Operator Training, and Feedback to Improve Electric Bus Fuel Economy Fleet operators can incorporate a few electric buses with little impact to daily operations; there are capital costs, but for the most part it’s “business as usual.” Transitioning an entire fleet to electric buses, however, is disruptive – for both the transit agency and the utility. A transit authority in California has committed to electrifying its 80-bus fleet by the end of 2018. For their electric buses to be cost-effective compared to internal combustion, the average fuel economy needs to be 2.0 kWh/mile. The fuel economy of their two currently deployed buses for trips along the same route range from 1.2 kWh/mile to 5 kWh/mile, representing significant risk to the transit authority due to highly variable operating costs. The transit agency has partnered with a team to design and implement an E-Bus Operator Training and Feedback Program (E-Bus Program) that maximizes E-Bus fuel economy by coupling education and training for E-bus operators with advanced analytics and proven behavior modification strategies. The data model for the E-Bus Program includes processing data from: 1) The E-Bus onboard telemetry system; 2) The transit agency’s administrative system that tracks driver assignments to buses and routes, tracks the location of the bus during its route (using GPS), and forecasts estimated departure times; 3) Traffic data to identify slow zones (e.g., hospitals, schools, etc.) and development density factors that may impact E-Bus operational efficiency; and 4) Existing driving habits (good and bad) and training/guidance provided to operators on driving best practice. These data will help the transit authority to better understand key drivers of fuel economy (including both good and bad habits) and inform the development of the E-Bus operator training model. The agency will also use feedback mechanisms and other behavioral modification strategies to encourage driving techniques and practices that improve fuel economy. The authors of this paper will discuss the design and evaluation of this effort, including: how the program will develop strategies that are scalable and fleet-wide; ways to assess and optimize operational parameters and operator performance to maximize fuel economy and minimize operating costs; and how to establish a transit agency outreach, education, and technical assistance platform to scale-up E-Fleet tools, programs, lessons learned and best practices. The authors will also discuss the benefits of and need for developing an evaluation plan at the initial stage of the design so that the data model is comprehensive. Lastly, this paper will highlight the truly disruptive nature of fleet electrification, and how the E-Bus Program uniquely addresses four key dimensions of organizational systems change: technology, people, process, and policy

Shiqi Ou, Oak Ridge National Laboratory Quantitative estimation for residential vehicle parking rate in China and its potential influence on PEV purchasing China has become the largest vehicle market in the world since 2009, and is ambitious to expand the population of the plug-in electric vehicles (PEVs) to 5 million units by 2020. Accompanied with the rapid urbanization and motorization, the residential vehicle parking and home PEV charging issues have been causing concerns. This study probes the residential parking rates by provinces in China and projects the residential vehicle parking rates in years, and adopted the discrete choice model to simulate the potential influence of the residential parking rates on the plug-in electric vehicles (PEVs) purchasing. The residential parking rates in China are firstly quantitatively revealed with limited data resources. By data mining in several major real estate trading network platforms in China, this study obtained the raw information on the residential communities in 31 provinces (areas) in mainland China, including household numbers, residential parking numbers, price, building ages etc. By quantitatively estimating the housing lifetime and urbanization rate in China, this study calculated the average residential vehicle parking rates in metropolitan, suburb, and rural areas in every province (area) in China for years in 2005, 2015, 2025, and 2050. These 372 residential parking rates comprehensively present a full picture of the residential parking level in China, which shows an inequality in areas in China: varies by provinces, by regions with different economic levels, and by urban types. Shown by the distributions of the results for the residential parking rates, the development of the residential parking rates is positive related to the economic development, urbanization rate, and urban planning. The results also reveal that the residential parking rates vary in the same metropolitan cities with different urban planning. These values directly reflect the contemporary urbanization changes and are reasonably explained by integrated considering the economic level, urban types. Meanwhile, with the quantitative model, the relations between the residential parking rates and PEV sales are revealed in this study. By adopting the consumer discrete choice model and considering the observed utilities (or the use costs, including vehicle price, energy cost, recharging inconvenience cost, government subsidy, range anxiety cost etc.) for each type of vehicle, the study quantitatively discussed the impacts of the charging accessibility (or residential parking rates) to the PEV purchases in Beijing, China. The model results show that the residential parking rates have very positive impacts to the PEV sales. Specifically, the consumer’s tendency to choose the PEVs would significantly grow up when the residential parking rates are improved to be more than 50 - 60 %. Due to data limitation, the approach taken in estimating the national residential parking rates and developing the model is to create a framework for integrating data and consumer behavioral models at an appropriate level of detail, whether or not the data are fully available or the consumer behaviors are fully understood at the present time. The presentation will give the full picture of the development of the Chinese residential parking estimated rates, and show how the improved residential parking rate promote the PEV purchasing.

Kalai Ramea, University of California, Davis Transitions to Alternative Fueled Vehicles Using a Spatial Consumer Choice and Fueling Infrastructure Model The consumer choice models developed so far operate on a rather spatially aggregated level, either at a state level. Detailed spatial models have been developed mainly to analyze the travel time or convenience impacts of siting refueling stations, but none of these models developed so far establish the relationship between vehicle purchase choices of consumers vs. their attributes based on geographical location on a fine level. These models include factors other the technology-related costs that goes into decision-making. This is especially important when it comes to adoption of new technologies. A spatially refined consumer choice model provides a robust layout to analyze various station siting locations for optimal adoption of vehicles. This is the first of its kind modeling effort where consumer choice is represented at a detailed spatial level. This work will be of interest to a wide range of stakeholders (including automotive companies and fuel providers, as well as government and NGOs) as this work will focus very specifically on the important interactions between deployment of fueling infrastructure, consumer reaction and spatial adoption of alternative fueled vehicles. This work will help broaden our understanding of the role of infrastructure deployment, consumer demographics, spatial and geographic factors, all of which will influence vehicle adoption. This model framework answers the following questions: - How do the locations and numbers of hydrogen stations and electric vehicle public chargers influence the alternative vehicle purchases at a local level? - How might sales for alternative fueled vehicles vary spatially as a result of geographic and demographic factors and infrastructure deployment? - What are likely sales and fleet shares of hydrogen fuel cell vehicles and plug-in electric vehicles in study regions out to 2030? The model segments consumers into various categories based on their geographical location, driving behavior, attitude towards risk, and access to infrastructure. For each consumer group, “disutility costs” are estimated based on these factors. The technology specific direct costs (vehicle and fuel costs) are combined with disutility costs to determine the purchase probability of a consumer group for that vehicle technology through a nested multinomial logit approach. Preliminary results show that hydrogen station infrastructure is extremely important for adoption in the neighborhood, and work charging is more important than public charging when it comes to electric vehicle adoption. Other factors, such as, early adopter population in the neighborhood also plays a role in identifying the starter hubs for the new vehicle technologies.

Angela Sanguinetti, UC Davis Findings from a Meta-analysis of In-vehicle Eco-driving Feedback The Monroney sticker on every new vehicle sold in the U.S. provides a fuel economy estimate with the caveat, “Actual results will vary for many reasons, including driving conditions and how you drive and maintain your vehicle.” Eco-driving is taking advantage of this variability to reduce actual on-road fuel economy. Eco-driving has been highlighted as a significant opportunity toward reaching goals for carbon dioxide emissions reductions in the transportation sector. The most common strategy to promote eco-driving is the provision of in-vehicle feedback, conveying the effects of driving behaviors on fuel economy. There is a growing body of evidence that in-vehicle feedback is an effective strategy to promote eco-driving and resultant fuel savings and emissions reductions. However, results are widely variable (from no savings to over 20%). Some of the variation in eco-driving feedback effectiveness is undoubtedly due to the extremely wide variation in feedback provided, e.g., from real-time numeric fuel economy gauges, to haptic feedback (counter-pressure from accelerator pedal), to more complex and combined feedback systems. The relatively few studies that have compared multiple types of in-vehicle eco- driving feedback suggest it is more effective when it aligns with drivers’ values and goals (e.g., saving money versus reducing emissions) and when it is adaptive (consisting of graduated challenges), but these are each solitary findings. Additionally, past studies are highly inconsistent in terms of how they measure outcomes. As a result of these limitations, we do not yet understand the most promising eco-driving behaviors to target and the most effective types of feedback to promote those behaviors. To help address this gap, we conducted a statistical meta-analysis of in-vehicle eco-driving feedback studies. Meta- analysis enables two general outcomes. First, it enables a pooled estimate of an effect. In this case, meta-analysis allows us to derive an estimate of the effect of eco-driving feedback on fuel economy that is closer to the truth than the effect observed in any individual study. Second, meta-analysis enables the identification of variables that mediate or moderate an effect. In our case, this means we can identify how characteristics of feedback studies, such as the behaviors targeted and the types of feedback provided, influence effects on fuel economy. This presentation will discuss our findings regarding the overall effectiveness of eco-driving feedback, the characteristics and types of eco-feedback that are more or less effective for particular eco-driving behaviors, and how targeting particular eco-driving behaviors may lead to more or less fuel savings and emissions reductions. We will also discuss implications for designing and evaluating in-vehicle feedback

Kipp Searles, Center for Sustainable Energy Outcomes from a Northern California Electric Vehicle Incentive Program from the Lead Agency, Administrator, and Dealer Perspectives

Gil Tal, UC Davis Assessing Long-distance Trip Length Distributions for Improving the Modeling of Plugin Vehicle Market Demand and EV Travel Performance Measures of long-distance passenger vehicle travel patterns are important for the metrics needed for many policy and planning applications. Most obvious among recent research topics is the measurement long trip distance and frequency and its relationship to the demand for battery electric vehicles (BEVs). In most cases, BEVs cannot travel long-distance without charging stops and modeling the performance of plugin hybrid electric vehicles (PHEVs) who performed as electric for a limited range before shifting to traditional fuel. Long-distance trip length frequency distributions will be important input for understanding vehicle purchase decisions as well as mode and vehicle choice on a trip by trip basis and the results electric vehicle miles traveled as share of total miles. TRB Special Report 320 (2016) on Interregional Travel recently acknowledged the gap in knowledge of long-distance travel, stating that “long-distance travel is not nearly as well understood as local travel.” The last national long-distance travel dataset for the United States was collected in 1995, the American Travel Survey (ATS). Long- distance trips may be correlated with the household vehicle fleet but they are difficult to survey and have low frequency and therefore remain under-reported and underestimated in studies estimating plugin vehicle adoption and use. The data and knowledge on long-distance road trips and other long distance and overnight trips are very limited. An early study by this research team focused on the longest road trip in the last year of plugin vehicle owners in California. The project used a new method to survey long-distance road trips, a web map survey based on google maps interface. We studied only one trip, the longest vehicle trip in the last 12 month. For 40% of the households in our study one long trip accounts for more than 5% of the annual gas used. For 20%, one long trip accounts for more than 5% of yearly miles. We show that long-distance road trips constitute a significant share of some household’s VMT, and this share is correlated with socio-demographic, location, travel behavior, and vehicle ownership characteristics (Tal and Nicholas, 2015). More simple studies on long road trips and electric vehicle acceptance using GPS traces to evaluate the number of day per year over a certain threshold may lack behavioral motivation and trip purposes but these data do provide relative mileage and trip length information. Tamor et al. (2013) for example, use a sample of 133 vehicles augmented with doors based on the national household travel survey (NHTS) to estimate vehicle acceptance. The purpose of this project is to combine different datasets that include long road trips to better understand long-distance trip length distributions. We will compare the results of a long-distance web survey that looks at the longest trip the household perform in the last year, one week of the CHTS GPS data from all the vehicles in 1,627 households collected in California in 2012, 1 to 3 years of in-vehicle data collected by manufacturers and shared with UC Davis from 6500 new plug-in hybrid vehicles in California and 27 other states between 2011 and 2015. In addition to the trip length frequency data that will be of direct us to planners, we intend that the comparison of data sources can feed future travel survey and data collection designs. Results will inform our understanding electric vehicle acceptance based on internal combustion engine (ICE) households and electric vehicles owner households. Our results will improve our knowledge answering two main questions. First, comparison of alternative data collection techniques will allow recommendations of future data collection methods for long-distance travel. Second, we will create the first robust trip length frequency distributions since the 1995 American Travel Survey (ATS), the last comprehensive long-distance travel survey conducted in the US.

Eric Cahill, Plug In America Report from the Dealer Trenches: Selling EVs in San Diego

EVs create a number of pain points for dealers. They are more complicated to sell and customers need more handholding, which requires more time and attention from sales staff. This can cost dealers sales of more profitable vehicles. Unlike conventional cars, most sales staff have little or no behind-the-wheel experience with a plug-in. Many struggle to convey the unique value of electric driving. A lack of EV-focused sales training and high turnover further erodes EV expertise at the dealership. With little or no commissions and uncertain profits, many dealers simply cannot justify diverting resources toward EVs when they comprise a small fraction of monthly sales in the early years of sales. To address this challenge, Plug In America has launched a comprehensive plug-in dealer initiative to bolster dealers selling electric vehicles. The program partners with new dealers to grow EV sales by providing additional training, tools and resources to area auto dealers and consumers. Our first pilot program, which launches later this month, targets the San Diego metro area and is funded through a partnership with San Diego Gas & Electric. The presentation will discuss the challenges faced by dealers, the design and implementation of the program and will report on initial outcomes and lessons learned.