Zero Emission Vehicles: Forecasting Fleet Scenarios and their Emissions Implications (2019)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

25 CHAPTER 3 ANALYSIS SCENARIOS Findings from the literature review were used to develop analysis scenarios. Collectively, the scenarios bracket a broad range of potential future ZEV fleet changes that can be used to support transportation agencies in planning for potential outcomes beyond ZEV adoption rates in 2019. The scenarios require specification of key factors (e.g., net vehicle cost and availability of electric vehicle charging stations) linked to ZEV adoption rates. The feasibility of adequately representing quantitative changes to the vehicle fleet in each scenario was considered when selecting scenarios to carry forward for analysis of emissions reductions due to changes in ZEV adoption as estimated by MA3T for the scenarios. The Base Case is assumed to be a business as usual scenario with no changes in trends for policy, consumer, technology, and infrastructure factors that are assumed in MA3T V20190404. For this project, MA3T was calibrated to the 2019 AEO, which accounts for key laws and regulations in effect in 2019. The following sections describe (1) a comparison of future ZEV adoption from MA3T and two other data sources, (2) details about the MA3T model and justification for its use in the analysis, and (3) the ZEV adoption scenarios selected for subsequent analysis. 3.1 OVERVIEW OF THE MA3T MODEL MA3T is a comprehensive consumer choice model that can be used to estimate ATV populations. The model was created in Microsoft Excel® and its methodology is described in detail in Lin and Greene, 2009. Briefly, the model constructs its forecasts by integrating several behavioral models with certain known technology, infrastructure, and policy parameters. The model disaggregates the household market into thousands of segments representing geographic area, area type, attitudes towards technology risk, vehicle use intensity, and recharging access. Additional resources for the model can be found on the ORNL Transportation Energy Evolution Modeling (TEEM) website (https://teem.ornl.gov/ma3t.shtml). The latest version of the model is calibrated to total light-duty vehicle sales data obtained from the U.S. DOE Energy Information Administration’s 2019 AEO, which covers a projection period of 2019 to 2050 (U.S. EIA 2019a). The model forecasts populations of light-duty vehicles across 20 powertrain technologies (e.g., spark ignition conventional cars, fuel cell hybrid electric cars, and battery electric cars with 200-mile range) and five vehicle classes (i.e., cars, car sport utility vehicles [SUV], pickups, truck- SUVs, and vans). It also includes more than 40 input parameters that represent key factors affecting ZEV adoption, all of which can be specified to predict light-duty vehicle populations for various ZEV adoption scenarios. The parameters fit into five categories: • Attributes and behaviors of consumers • ATV policies and programs • Cost, characteristics, and performance of ATVs • Energy prices • Refueling and recharging availability MA3T has previously been used in analyses of future light-duty vehicle markets under different assumptions about future vehicle technologies and market conditions to support the DOE Office of Energy Efficiency and Renewable Energy (EERE) Vehicle Technologies Office. Work related to the development of MA3T includes the assessment of the impact of charging infrastructure, range optimization, DOE targets for hydrogen fuel cell vehicle adoption, and energy and power grid impacts.

26 Numerous peer-reviewed articles on work using MA3T have been published (e.g., Lin and Greene 2011, Greene et al. 2013, Lin et al. 2013, Farzaneh et al. 2014, Podkaminer et al. 2017, Lin and Liu 2014). 3.2 MA3T AND PROJECTIONS OF FUTURE ATV SALES AND POPULATIONS To further investigate future ZEV adoption beyond the market share forecasts identified in the literature review, data was compiled from (1) the 2019 AEO (U.S. EIA 2019a), (2) the ATV Sales Dashboard, and (3) the MA3T model V20190404 using default values for model inputs. The assumptions applied to MA3T from the AEO forecast can be found in U.S. EIA 2019b and 2019c; assumptions applied in the MA3T model can be found in Lin and Greene 2009 and Lin and Greene 2012. An important aspect of the AEO forecast is the incorporation of California’s Zero-Emission Vehicles program into the transportation module. The Zero-Emissions Vehicles program was also adopted by the nine other MOU states. In the 2016 update to the ZEV program, the required percentage of credits earned by automobile manufacturers for BEV, PHEV and FCEV sales increases from 4.5% for model year 2018 to 22% for model years 2025 and beyond. The number of credits for each vehicle sale depends on the vehicle type and technology, where vehicles with longer battery range count for more credits. In addition, the AEO forecast accounts for the California corporate average fuel economy (CAFE) standards (also adopted by several other states), which (as of 2019) extend until 2025 with its current requirements. These aspects of the AEO forecast are carried over to the MA3T model, which is calibrated to the AEO with respect to total light-duty vehicle sales. However, the sales of ATVs estimated by MA3T are not limited by the number of sales in the AEO forecast because of several key feedback assumptions: for any year, ATV sales in MA3T are affected by sales from the previous year through (1) supply constraint parameters, (2) learning by doing parameters, (3) make and model availability parameters, and (4) consumer attitude parameters, such as perceived technological risk. Furthermore, sales in any year affect the availability and power of charging infrastructure in the following year. As Figure 3 shows, MA3T predicts fewer annual BEV sales than the AEO does between 2018 and 2029, but much greater sales between 2030 and 2050. The large increase in sales from MA3T is likely a result of the internal feedback loops. A few features of the growth in BEV sales for both forecasts reflect some key assumptions, effects of MA3T’s feedback loops, and real-world milestones: • The AEO forecast shows a “leveling off” of the rate of increase in BEV sales in 2025 o The California ZEV program requirement of a 22% credit percentage does not change after 2025, and o The CAFE standards (as of 2019) extend until 2025, at which time they will be revaluated. • MA3T assumes no availability of public charging until 2017, and roughly a doubling of availability between 2017 and 2050, although the sales seem slow to respond to the initial public charging availability. • MA3T predicts relatively slow growth in BEV sales between 2017 and 2025, during which period the model assumes no change in home and workplace charging availability. • The MA3T model shows growth in sales of BEVs becoming relatively steep beginning around 2025 (2025 is the first of two “Year Points” in MA3T, between which home charging power doubles).

27 • MA3T estimates that the growth in sales will begin to slow in 2030, which happens to be the target year for a goal of 5 million PHEVs, BEVs, and FCEVs on the road in California. This goal spurred the ZEV program, though it is not clear if the goal is accounted for in the AEO forecast and carried over to MA3T. • The rate of increase in BEV sales forecasted by the two models is similar between 2035 and 2045, at which time the growth in sales estimated by MA3T seems to reach an inflection point. • In general, by 2050, MA3T assumes that home, workplace, and public charging availability and power reach levels much greater than what is assumed for the first year of its modeling period (i.e., 2005), and the feedback loops appear to maximize BEV sales between 2045 and 2050, the end of the model’s forecast period. Figure 3. Forecasted Annual BEV sales from MA3T V20190404 and the 2019 AEO. Data Sources: MA3T V20190404 and EIA 2019 Annual Energy Outlook (U.S. EIA 2019a). Figure 4 shows historical ZEV populations in 2017 and estimatesii of ZEV populations in 2040 using the three data sources. The 2017 historical ZEV populations are roughly the same across the three data sources. Based on the data from these three sources, the total ZEV on-road light-duty vehicle population is forecasted to reach approximately 22 to 25 million vehicles by 2040. Using cumulative sales to project future sales based on ATV historical annual sales (as tracked in the ATV Sales Dashboard) produces approximately 1.5 million more ZEVs than is forecast by the 2019 AEO, but the cumulative sales approach may be overestimating the population because vehicle scrappage is not accounted for. MA3T estimates a population of 3.7 million more ZEVs in 2040 than estimated in the AEO, and 2.3 million more ZEVs in 2040 than estimated from the ATV Sales Dashboard data. 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 5,000 20 17 20 19 20 21 20 23 20 25 20 27 20 29 20 31 20 33 20 35 20 37 20 39 20 41 20 43 20 45 20 47 20 49 BE V Sa le s ( th ou sa nd s o f v eh ic le s) MA3T AEO

28 These differences are likely a result of the different assumptions used in estimating the future ZEV populations. These differences include (1) the AEO forecast includes the ZEV regulations adopted by California and the other MOU states; (2) MA3T includes feedback loops, such as the previous year’s sales volumes by technology affecting the number of makes and models available for each technology, and a survival rate parameter for each technology representing scrappage of older vehicles; and (3) the cumulative sales estimated from the ATV Sales Dashboard does not include vehicle scrappage. The default input values in the MA3T model represent the assumption that no FCEVs are on the market during the modeling period (2005 through 2050), while the AEO projects a population of nearly 1 million FCEVs in 2040. Nevertheless, MA3T produces an estimate of BEV population that outweighs the lack of FCEVs when compared to the total AEO population of ZEVs. The populations estimated by MA3T differ from the AEO projection by 17% and from the ATV Dashboard estimate by 9%. These relatively small differences support confidence in using the MA3T model for estimating ZEV populations for the analysis scenarios and subsequent estimations of emissions reductions. Figure 4. Historical (2017) and projected (2040) ZEV populations using three different data sources. Data Sources: EIA 2019 Annual Energy Outlook (U.S. EIA 2019a); Alliance of Automobile Manufacturers 2019; MA3T V20190404. 3.3 ZEV FLEET ASSUMPTIONS IN THE MOVES MODEL While MOVES2014iii can model electric vehicles, the market share of electric vehicles was sufficiently small when the model was released (2018) that the default electric vehicle population is effectively zero in the model. MOVES2014 allows users to model an EV population by providing market share information for EVs. EVs do not have exhaust emissions, but MOVES2014 does have energy consumption, brake wear, and tire wear rates for EVs (EV brake wear and tire wear rates are assumed to be the same as those of gasoline vehicles). With customized market share information, MOVES2014 can model energy consumption and brake wear and tire wear emissions from EVs (U.S. EPA 2014). 3.4 SUMMARY OF ZEV ADOPTION SCENARIOS The ZEV adoption scenarios are foundational to the emissions analysis work in this project. As found during the literature review, ZEV adoption varies as a result of a variety of factors, including the cost of 0 1 2 3 4 BEV FCEV Total ZEV 2017 U.S. ZEV Population (millions) AEO Auto Alliance MA3T 0 10 20 30 40 BEV FCEV Total ZEV 2040 Projected U.S. ZEV Population (millions) AEO Auto Alliance MA3T

29 technology, access to HOV lanes, availability of charging infrastructure, and availability of incentives. Four key scenarios were developed where changes in the underlying factors would likely result in the largest changes in ZEV adoption. These four scenarios were used in order to estimate (1) future ZEV populations using the MA3T model V20190404, and (2) the corresponding reduction in exhaust emissions as a result of the increase in ZEVs. The selected scenarios are summarized in Table 7, which includes the scenario IDs and a brief name and description for each scenario. In addition to the Base Case scenario, used as a reference point, three alternative scenarios were developed to reflect the key factors affecting ZEV adoption that were identified in the literature review. Those three scenarios represent a combination of both likely scenarios (e.g., accelerated achievement of cost parity) and scenarios that may be more directly influenced by transportation agencies (e.g., substantial expansion of infrastructure). Although AEO forecasts, projections from the ATV Sales Dashboard, and MA3T estimates use different assumptions in estimating ZEV populations, the relative similarity of ZEV populations in 2017 and 2040 between the three data sources provides confidence in use of the MA3T model for this study to estimate future ZEV populations for further analysis in this study. The various parameters in the MA3T model that were adjusted, and the adjustment values used in each simulation are summarized in Chapter 4. Table 7. ZEV adoption scenarios for modeling in MA3T and estimating emissions reductions. Scenario ID Scenario Name and Key Concepts B Base Case Business as usual. I Substantial Expansion of Infrastructure Expansion of EVSE beyond recent and pending improvements. P Advanced Use of Incentives/Policy Wide implementation of high-impact incentives policies/programs. C Accelerated Achievement of Cost Parity Accelerated reduction of vehicle costs and increased fuel costs.