Energy Poliy 52 (2013) 429 438 Contents lists available at SiVerse SieneDiret Energy Poliy journal homepage: www.elsevier.om/loate/enpol Cost-effetiveness of plug-in hybrid eletri vehile battery apaity and harging infrastruture investment for reduing US gasoline onsumption Sott B. Peterson a, Jeremy J. Mihalek a,b,n a Dept. of Engineering and Publi Poliy, Carnegie Mellon University, United States b Dept. of Mehanial Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, United States H I G H L I G H T S We ompare ost of PHEV batteries vs. harging infrastruture per gallon of gasoline saved. The lowest ost solution is to swith more drivers to low-apaity PHEVs and HEVs. If more gasoline savings is needed, batteries offer a better value than hargers. Extra batteries & hargers are both more ostly per gal than oil premium estimates. Current subsidies are misaligned with fuel savings. We disuss alternatives. artile info Artile history: Reeived 18 June 2012 Aepted 21 September 2012 Available online 22 Otober 2012 Keywords: Plug-in hybrid eletri vehile Charging infrastruture Battery size abstrat Federal eletri vehile (EV) poliies in the United States urrently inlude vehile purhase subsidies linked to EV battery apaity and subsidies for installing harging stations. We assess the osteffetiveness of inreased battery apaity vs. nondomesti harging infrastruture installation for plug-in hybrid eletri vehiles as alternate methods to redue gasoline onsumption for ars, truks, and SUVs in the US. We find aross a wide range of senarios that the least-ost solution is for more drivers to swith to low-apaity plug-in hybrid eletri vehiles (short eletri range with gasoline bakup for long trips) or gasoline-powered hybrid eletri vehiles. If more gasoline savings are needed per vehile, nondomesti harging infrastruture installation is substantially more expensive than inreased battery apaity per gallon saved, and both approahes have higher osts than US oil premium estimates. Cost effetiveness of all subsidies are lower under a binding fuel eonomy standard. Comparison of results to the struture of urrent federal subsidies shows that poliy is not aligned with fuel savings potential, and we disuss issues and alternatives. & 2012 Elsevier Ltd. All rights reserved. 1. Introdution US interest in federal poliy to redue gasoline onsumption in the transportation setor began in response to the Arab oil embargo of the 1970s. The 1975 Energy Poliy Conservation At reated orporate average fuel eonomy (CAFE) standards to mandate inreases in the effiieny of the vehile fleet by setting effiieny standards for passenger ars starting in 1978 (Energy Poliy and Conservation At, 1975). Effiieny requirements initially inreased annually until 1985, but standards for ars remained stati until further legislation was passed. In 2007 the Energy Independene and Seurity At set a target of 35 mpg for ombined ar/truk fuel eonomy by model year n Corresponding author at: Dept. of Mehanial Engineering, Carnegie Mellon University, 5000 Forbes Avenue, Pittsburgh, PA 15213, United States. Tel.: þ1 412 268 3765. E-mail addresses: speterson@mu.edu (S.B. Peterson), jmihalek@mu.edu (J.J. Mihalek). 2020 (whih has sine been moved to 2016 (EPA and NHTSA, 2011)), but it stipulated penalties only if manufaturers fell below 92% of the standard (Energy Independene and Seurity At of 2007, 2007). The standard allows manufaturers to trade redits, and it was implemented using formulas tied to vehile footprint, rather than diret averages of orporate fleets (Whitefoot and Skerlos, 2012). The at also provided loan guarantees for advaned battery researh and grant programs for plug-in hybrid eletri vehiles (PHEVs, whih use a mix of eletriity and gasoline) and battery eletri vehiles (BEVs, whih use eletriity only and have no gasoline bakup) to help make them eonomially feasible as mass market vehiles. At the end of 2010 General Motors introdued a PHEV alled the Volt, Nissan introdued a BEV alled the Leaf, and other automakers have also begun to offer PHEV and BEV models olletively referred to here as eletri vehiles (EVs). These types of vehiles use grid eletriity to displae gasoline. In August 2012 the National Highway Traffi Safety Administration (NHTSA) finalized rules pushing CAFE standards higher to 0301-4215/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.enpol.2012.09.059
430 S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 40.3 mpg by 2021 and 48.7 mpg by 2025 (EPA and NHTSA, 2012). In addition to CAFE standards the Environmental Protetion Ageny (EPA) released greenhouse gas (GHG) standards of 163 g/mile by 2025 that will math (these values orrespond with 54.5 mpg, but the EPA expets other improvements suh as a derease in oolant leakage to redue GHG emissions) (EPA and NHTSA, 2012). The EPA standards inlude a number of inentives to enourage the use of game hanging tehnologies (EPA and NHTSA, 2012). These inentives begin with off yle redits for model year (MY) 2012 2016 vehiles that implement speifi tehnologies. Alternative fuel vehiles suh as PHEVs are permitted a multiplier for model years 2017 2021 in the GHG alulations (EPA and NHTSA, 2012). The rule also treats emissions from eletri travel as 0 g/mile in model year 2022 2025. 1 Thereafter upstream fuel emissions will be taken into aount (EPA and NHTSA, 2012). While BEVs may seem to be the most omplete solution to displaing gasoline with eletriity, there are signifiant drawbaks. Among them are range and refueling onstraints. High rate harging for the Nissan Leaf allows the battery to be harged to 80% in 30 min at a speialized high-voltage harging station (Witkin, 2012). Despite being many times faster than a typial home harger, this rate is still slow ompared to refilling a gas tank and requires new infrastruture. On a long trip, this would mean stopping every 60 miles for approximately 30 min and would require hanges in infrastruture (suh as new transmission, sub-transmission, and distribution lines) (Lemoine et al., 2008; Hadley and Tsevetkova, 2008; Peterson et al., 2011; Lemoine et al., 2008). While BEVs may funtion well for some appliations, they are not an adequate replaement for most primary vehiles unless a seondary vehile is available to be used for longer trips. We fous on PHEVs beause they an more easily replae urrent vehiles and an benefit from availability of publi harge points without being dependent on them. PHEVs at like gasoline-powered hybrid eletri vehiles (HEV), suh as the Toyota Prius, by using a battery as a buffer to store braking energy and improve engine effiieny. But PHEVs have the additional ability to store eletriity from the grid onboard and use it to displae gasoline while driving. PHEV drivers will not need to hange their habits. Long trips an still be taken using gasoline, while short trips an be powered mostly or entirely using eletriity. These advantages do not ome without a penalty. PHEVs require both an internal ombustion engine (ICE) and a substantial battery pak. Beause of this, a PHEV designed for 40 miles of eletri range will weigh and ost more than a BEV with similar range or a onventional vehile with similar performane and interior spae. To help overome obstales to EV adoption, poliymakers have provided purhase inentives based on battery size and inentives for installation of harging infrastruture. The Amerian Reovery and Reinvestment At of 2009 provides a tax redit of $2500 per plug-in hybrid eletri vehile sold (requires at least 4 kwh battery apaity) and an additional $417 for eah additional kwh of battery apaity in exess of 4 kwh (apped at $7500 for vehiles with a gross vehile weight less than 14,000 lb) (The Amerian Reovery and Reinvestment At of 2009, 2009). This subsidy for a speifi manufaturer s vehiles delines to 50% then 25% in a phaseout period, whih begins in the seond alendar quarter after that manufaturer has sold 200,000 vehiles and lasts four alendar quarters (The Amerian Reovery and Reinvestment At of 2009, 2009). 1 This inentive is for PHEVs, BEVs, or fuel ell vehiles and overs the first 200,000 vehiles for all manufaturers and 600,000 vehiles for those that sell 300,000 suh vehiles in the 2019 2021 timeframe (EPA and NHTSA, 2012). The US Department of Energy (DOE) also granted $37 million for installing 4600 harge points in speifi markets around the nation (over $8000 per harge point) (Dogget, 2011) and granted $99.8 million to fund the EVProjet, whih is installing 14,000 level 2 (208 240 V) hargers and a variety of other infrastruture and monitoring equipment (Advaned Vehile Testing Ativity, 2011). While BEVs require a large number of harge points in order to at as primary vehiles, PHEVs an benefit from a smaller number: harge points ould help small-battery PHEVs displae a greater amount of gasoline. We examine the strategies of subsidizing EV battery apaity and harging infrastruture installation by assessing ost-effetiveness of eah approah for reduing US gasoline onsumption under a range of senarios. 2. Methodology To estimate osts and gasoline savings of eah approah, we first alulate gasoline and eletriity use by PHEVs of varying battery apaity under a range of harging senarios. Seond, we estimate the neessary harging infrastruture to enable eah harging senario. Finally we use estimates of ost and gasoline displaement to ompare aross options. We disuss eah in turn. 2.1. Fuel use model This work uses two main data soures. The first is the National Household Travel Survey 2009 (NHTS) whih inludes travel information from diaries of over 150,000 households (National Household Travel Survey, 2010). The NHTS day trip file lists trips taken by eah household on a randomly-assigned day. The seond data soure is the Department of Energy s Greenhouse Gasses, Regulated Emissions, and Energy Use in Transportation (GREET) model, version1.8d (The Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Model, 2010). The GREET model assesses energy use in transportation and inludes simulations for passenger ars, SUVs, and light truks with over 80 fuel systems and tehnologies. We use GREET 1.8d estimates of year 2015 vehile effiieny in harge depleting (CD) and harge sustaining mode (CS). Combining these effiieny numbers with vehile travel patterns from the NHTS day trip file allows the predition of estimated gasoline and eletriity onsumption from a set of PHEVs, with designs that span battery apaity values produing all-eletri range (AER) values from 5 to 60 miles. Base ase assumptions are desribed below and summarized in Table S.39 of the supporting information. The GREET model assumes that PHEVs with AER less than or equal to 25 miles utilize a split powertrain and blended ontrol strategy, while PHEVs with AER in exess of 25 miles utilize a serial hybrid design, whih is far less effiient in harge sustaining mode (the Chevy Volt with AER of 35 does not use a serial design (Markus, 2010)). This assumption auses the model to predit lower effiieny for vehiles with AER over 25 miles (as shown later in Fig. 3). To avoid the serial powertrain assumption, we onsider only AER of 25 miles or less. Our results shown later suggest that AER values above 30 miles are not ompetitive with the shorter AER ranges on a ost per gallon saved metri regardless of powertrain assumption aross all sensitivity senarios. The NHTS data were proessed following (Peterson et al., 2011) with several modifiations desribed below (Fig. 1). The daily vehile travel data were extrated and weighted aording to the vehile weights assigned in the sample. To estimate the timing of fuel savings, vehiles are partitioned by age and vehile lass beause their travel varies signifiantly along both dimensions (Figs. 2 3). They are further partitioned into trips taken on
S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 431 weekdays and weekends and then the results are weighted by the number of weekend (104) and weekdays (261) in the year (see Eq. (1)). The trip hain and the resulting total distane traveled in CD and CS mode for eah vehile under eah harging strategy senario is averaged over the vehiles within eah lass and age and within eah weekend or weekday group in the NHTS data set. This approah represents the driving patterns of an aggregate US vehile for eah lass and age, and we do not examine heterogeneity for different vehiles that have different driving patterns within eah lass-age group (Raykin et al., 2012; Shiau et al., 2009; Neubauer et al., 2012; Traut et al., 2012). Changes in effiieny and AER as vehiles age were ignored. In pratie, PHEV battery apaity will deline with age, whih would shorten AER; however, vehiles are typially designed to use only a portion of the battery s available energy, in part to delay redution in AER pereived by the onsumer. Whether PHEVs will enounter more or less engine wear than onventional vehiles (CVs) depends on how PHEVs use their engines. If the engine speed is partially deoupled from the wheel speed using the eletri motor to ompensate, this ould prolong engine life by enabling the engine to run mostly at steady state, but if the engine starts and stops Fig. 3. Gasoline onsumption over a 12 year life for vehiles of various AER using GREET 1.8d 2015 effiieny estimates. AER orresponds to GREET AER. Gasoline use inludes both CD and CS travel. The inrease at AER of 30 is due to the assumption in GREET that PHEVs with AER above 25 have a series drivetrain rather than a split drivetrain. Fig. 1. Basi fuel use model overview. Fig. 2. Change in annual VMT with vehile age as found from NHTS data. often and revs to follow vehile power demand then engine wear ould inrease. We onsider ars, SUVs, and truks (vans were ignored beause data were unavailable for effiieny and ost, but they make up the smallest portion of the lasses mentioned here). The energy onsumption of eah vehile in the sample is alulated assuming vehiles began eah day fully harged and harge ompletely after the last trip of the day. Several harging senarios were inluded to determine how muh additional gasoline ould be displaed with eletriity by also harging vehiles at work or also at every loation where the vehile parked for at least 30 min. Table 4 desribes these harging senarios. Speifially, the total eletriity use f ELEC and gasoline use f GAS are alulated for eah vehile lass, vehile age a (years), vehile AER b (miles), and harging senario g. For eah ase, we ompute average distanes traveled in CD and CS modes separately for weekday (WD) vs. weekend (WE) data, and we use these averages to ompute annual and total distanes, as shown in Eq. (1), where L is the vehile life (12 years base ase), j indexes the vehile driving profiles taken from the NHTS day trip file, J a,,we is the set of NHTS vehile profiles of age a and lass surveyed on a weekend, J a,,wd refers to those surveyed on a weekday, 9J a,,we 9 is the number of NHTS vehiles of age a and lass surveyed on a weekend, 9J a,,wd 9 is the number of vehiles of age a and lass surveyed on a weekday, and Z CD-E, Z CD-G, and Z CS-G are the vehile s eletrial effiieny in CD mode (mi/kwh), gasoline effiieny in CD mode (mi/gal, for blended operation), and gasoline effiieny in CS mode (mi/gal) as estimated by GREET (shown in tables Tables 1 3 below). These GREET values are estimates for effiieny of future vehiles as indiated by year. Note that the PHEVs run blended operation in CD mode, meaning that they onsume both gasoline and eletriity. f ELEC a f GAS a P ¼ 104 0 P ¼ 104 @ jej a,,we d CD j J a,,we Z CDE jej a,,we d CD j J a,,we Z CDG þ261 þ PjeJ a,,wd d CD j J a,,wd Z CDE PjeJ a,,we d CS 1 j A J a,,we Z CSG
432 S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 Table 1 Z CD-E in mi/kwh for 2015 vehiles from GREET 1.8d. f ELEC f GAS 0 þ261@ X L,L ¼ a ¼ 1 X L,L ¼ PHEV5 PHEV10 PHEV15 PHEV20 PHEV25 Car 5.2 5.2 5.3 5.3 5.3 SUV 4.5 4.5 4.5 4.5 4.5 Truk 4.9 4.9 4.9 4.8 4.8 Table 2 Z CD-G in mi/gallon for 2015 vehiles from GREET 1.8d (blended operation). PHEV5 PHEV10 PHEV15 PHEV20 PHEV25 Car 74 74 78 82 82 SUV 54 54 56 58 58 Truk 41 41 42 42 42 Table 3 Z CS-G in mi/gallon for 2015 vehiles from GREET 1.8d. CV HEV PHEV5 PHEV10 PHEV15 PHEV20 PHEV25 Car 27 38 43 43 43 42 42 SUV 20 28 28 28 28 28 28 Truk 18 24 25 25 25 25 25 Table 4 Charging senarios. Charging senario Brief desription HomeEve Vehile harges after arriving home on last trip of the day HomeAll Vehile harges anytime it is parked at home for at least 30 min WorkHomeEve Vehile harges when it first arrives at work and is parked for at least 30 min and at home after last trip of the day WorkHomeAll Vehile harges anytime it is parked at either home or work for at least 30 min AllStops Vehile harges anytime it is parked anywhere for at least 30 min a ¼ 1 þ PjeJ a,,wd d CD j J a,,wd Z CDG f ELEC a f GAS a 1 A PjeJ a,,wd d CS j J a,,wd Z CSG The funtions d CD j and d CS j use the NHTS data to ompute the distane that a vehile with AER of b under harging senario g traveling on vehile day trip profile j would travel in CD mode and CS mode, respetively. We examine ba{5,10,15,20,25} miles and ga{homeeve, HomeAll, WorkHomeEve, WorkHomeAll, AllStops}, where the harging senarios are desribed in Table 4. The proedure for omputing d CD and d CS for eah vehile in the set of a given age, AER, harging senario, lass, and weekday or weekend starts by assuming eah vehile begins the day fully harged. The vehile is traked through all reported trips and it is assumed it operates first in CD mode, where it onsumes both eletriity and gasoline, swithes to CS mode one the battery drops to its target state of harge (SOC) (40% of battery energy ð1þ remaining aording to GREET), and fully reharges after the last trip of the day. We use trip distanes and times speified in the NHTS dataset and assume a onstant effiieny per VMT from GREET (Z CD-E, Z CD-G, and Z CS-G ), ignoring differenes in effiieny for different drive yles (Patil et al., 2009). For eah trip if a vehile s battery is above the target SOC then the total battery energy required to omplete a trip is alulated. If the battery has enough energy, the SOC is deremented by the energy requirements of the trip. If the SOC is too low to omplete the trip in CD mode then the SOC is deremented to the target SOC and the portion of the trip not traveled in CD mode is traveled in CS mode. If the vehile battery is at the target SOC at the beginning of a trip the entire trip is traveled in CS mode. When a vehile parks, the time between trips is alulated. If it is greater than or equal to 30 min then the vehile an harge if the designated harging senario allows harging at the loation where the vehile is parked. GREET also aounts for redution in real-world effiieny ompared to test yle effiieny, where the AER is rated. This means that simulated AER may be shorter than rated AER (but this is primarily pronouned for the serial hybrid onfiguration that is not inluded in this analysis). Vehiles that are not driven on the survey day have a d CD and d CS of zero but are inluded so that the average total mileage found when simulating the trips (found by adding d CD and d CS of all vehiles in a set and dividing by the total number in that set) and fuel use estimates properly reflet all vehiles in NHTS for eah lass. The resulting daily onsumption is multiplied by either 104 for weekends or 261 for weekdays and summed to onvert to annual onsumption for a given age, AER, and harging senario. The NHTS file does not speify if travel was on a holiday, nor are vaation days speified, so suh days are inluded in omputing average weekday or weekend travel. Calulating fuel use by CVs and HEVs is aomplished in the same manner, but total mileage is used instead of traking d CD and d CS separately, sine these vehiles do not use multiple energy soures. The results of total distane traveled annually by all vehiles in a set of given age and lass divided by the number of vehiles (both those driven and not driven on the day surveyed) in that set is shown in Fig. 2. To simulate the life of a vehile it was assumed that it was driven in a manner onsistent with reported NHTS data for a vehile of its age and lass. Thus, for the base ase of a 12 year vehile lifetime we assume a ar drives roughly 14,000 miles in the first year, 13,000 in the next year and so on for eah year until it reahes year 12 (Fig. 2). Calulation of fuel onsumption treats eah vehile age and lass separately so any hanges in travel patterns as vehiles age are aounted for. It was found that a signifiant reason that vehile miles traveled (VMT) deline with age is that older vehiles are less likely to be driven on a given day. This an be seen when looking at figures Fig. S.25 and Fig. S.26 in supporting information. The total onsumption numbers for eah AER, vehile lass, and vehile age are reported in the supporting information for the base ase. 2.2. Infrastruture estimates Beause the amount of shared harging infrastruture required per PHEV to enable the WorkHome and the AllStops harging senarios will vary with the number of PHEVs in operation, we estimate an optimisti infrastruture ase that is favorable to harging points. The first assumption is that harging infrastruture is based on widespread PHEV adoption and harger installation; thus a new harge point would not need to be installed every time a person moves, hanges jobs, goes on a different errand and so on. To make these optimisti estimates the number of vehile harges for eah vehile and eah harging strategy was
S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 433 tabulated when simulating NHTS trips (as desribed above). The total number of harges per vehile driven in the WorkHomeEve harging ase was ompared to the HomeEve harging ase to determine how many additional harges are needed when vehiles are parked at work. Similarly the total number of harges per vehile in the AllStops harging ase was ompared to the Work- HomeAll ase to determine the number of additional harges at loations other than home or work. Sine every trip uses some energy, the number of harges does not hange with AER. The average additional harges for eah NHTS data point (vehile lass and age) were weighted by the number of vehiles of a given age and lass in the sample to determine the number of additional harge points used to estimate harging infrastruture. The NHTS inludes only data that generally desribes the loation, suh as home, work, plae of worship, shopping and so on. It is aknowledged that this lak of information ould lead to errors in the estimated number of harge points needed. Over-ounting ould result when a vehile parks at the same loation twie in the day (if it is not work or home) or when the same harge point ould serve more than one vehile if they parked at different times in the day. Underounting will our beause averaging does not aount for peak demand: a vehile surveyed on a given day might not travel to work (on a holiday for example), but the next day that same vehile may travel to work and require a harger. We are likely substantially underounting infrastruture needs to fully enable eah harging senario, and thus results are purposefully favorable toward infrastruture. Table 5 summarizes our estimates. The weekday or weekend set of vehiles was used depending on whih resulted in a greater demand for harge points, resulting in 1.3 hargers per vehile to enable WorkHome harging and 1.9 hargers per vehile to enable AllStops harging. 2.3. Cost estimates We estimate lifetime ost by ombining available ost estimates for vehiles, fuel, and infrastruture. The osts of harging infrastruture are varied over a range shown in Table 6. These osts inlude installation and equipment osts and are lower in the base ase than has been observed thus far for harging away from home (Dogget. 2011; Morrow et al., 2008), maintaining our optimisti stane on infrastruture. Installation osts an vary substantially. Plugging into an existing outlet requires no infrastruture investment, while installing a new iruit or a new eletrial panel with potential arpentry and onrete work or installing a publi harger that an aept payment and withstand exposure to the elements while dealing with potential vandalism is more expensive. Publi harge points would onsume some amount of eletriity even when not harging and would need maintenane, but neither of these osts is inluded in the model. Although the reently announed DC quik harger from Nissan osts far less than in the past ($9900), it requires three phase AC input (Nishimoto, 2011). The atual installation osts of suh units are likely to be substantial. Co-loating fast hargers to save on trenhing and other installation osts would inrease the likelihood of transformer upgrades and other ostly hanges to the distribution system. The base ase harging rates are assumed to be 1.4 and (optimistially) 7.7 kw for home and nondomesti (work and publi) harging, respetively. Higher rates have small effets on gasoline and eletriity onsumption. In partiular, we find that under the HomeAll harging strategy, inreasing home harge rates from 1.4 to 3.8 kw results in a maximum inrease of 2.5% in CD miles (for AER of 15 miles). Inreasing from 1.4 to 7.7 kw for home harging results in a maximum inrease of 3% in CD miles traveled (for AER of 20 miles) (see supporting information). We ignore any additional harging effiieny losses assoiated with ohmi heating at higher harge rates. Lifetime ost premium estimates for different options are found as follows. Vehile osts are taken from the 2015 average ase estimated by Argonne National Labs (ANL) in their 2011 report on potential of tehnologies in the light duty vehile fleet to redue petroleum onsumption (US Department of Energy, 2011b). ANL estimates the manufaturing ost premium of several plug-in vehiles ompared to a referene CV, and we multiply osts by 1.5 to aount for markup in retail priing (Plotkin and Singh, 2009). Any differenes in vehile maintenane osts are ignored, and we assume that the tration battery lasts the lifetime of the vehile for PHEVs (alternative assumptions are examined in the supplemental material). Lifetime gasoline and eletriity use is estimated from the fuel use model desribed previously. Fuel osts are taken from the EIA Annual Energy Outlook (2011) (AEO) that lists retail pries inluding taxes (Annual Energy Outlook, 2011). We adopt the AEO traditional high oil prie ase as our base ase and inlude other ases in supporting information. Vehile ost and fuel osts are taken starting in the year 2015. Fuel osts our over time, so we ompute the hange in net present value (NPV) ompared to a referene CV (lifetime ost premium) for eah AER and vehile lass. Negative lifetime ost premium values indiate lifetime savings. Lifetime Cost Premium ¼ C NPV PHEV, CNPV CV, Table 6 Charging infrastruture ost estimates (Dogget, 2011). Low ($) Base ase ($) High ($) Home 1.4 kw 25 75 550 Home 7.7 kw 500 1125 4000 Away 1.4 kw 1050 3000 9000 Away 7.7 kw 2500 5000 15,000 Away 38.4 kw 11,000 20,000 50,000 Table 5 Optimisti estimates of harging infrastruture requirements for eah harging senario. Home harging only Home and work harging All stops harging Weekday Weekend Weekday Weekend Number of nondomesti hargers required per vehile for vehiles driven on the day surveyed 0 0.47 0.12 1.13 1.5 (exluding one home harger) Portion of vehiles driven on the day surveyed 67% 60% 67% 60% Number of nondomesti hargers required per vehile (exluding one home harger) a 0 0.3 0.07 0.76 0.9 Total number of hargers required per vehile 1 1.3 1.1 1.8 1.9 a Assumes optimistially that vehiles not driven on the survey day do not need harge points other than home.
434 S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 C NPV XL f PHEV, ¼ Dþ PMT ðc D,i,L a ¼ 1 ð f PMT ðc D,i,LÞ¼ C DÞi 1ð1þiÞ L P! C NPV CV, ¼ XL j A J a, d j p GAS a ð1þrþ a a ¼ 1 J a, Z CV þp GAS a f GAS a ð1þrþ a þc CH Þþp ELEC a f ELEC a where D is the vehile down payment (100% of the ost of vehile of lass in our base ase); f PMT is the annual vehile loan payment; C is the additional ost of a plug-in vehile of lass (inluding home harging infrastruture) over a CV of lass ; i is the vehile loan s interest rate; L is loan period in years; p ELEC a is the prie of eletriity a years after 2015; p GAS a is the prie of gasoline a years after 2015; r is the disount rate; C CH is the ost of harging infrastruture away from home (see Table 5 for summary); d j is the total distane that NHTS vehile profile j drove on the day surveyed; and Z CV is the effiieny of the onventional vehile in miles per gallon. We onsider two main ases for disounting. In our base ase we use a normative onsumer disount rate of r¼5%. The seond ase attempts to reflet onsumer behavior using a higher disount rate for vehiles and fuels to reflet observed onsumer behavior. Consumers often exhibit surprisingly high impliit disount rates and sometimes gravitate toward purhasing whatever osts less at the point of purhase regardless of lifetime osts (Meier and Whittier, 1983). Studies onduted using surveys have estimated onsumer disount rates for alternative vehile purhases in the range of 20% 50% with the most likely value loser to 20% (Mau et al., 2008; Horne et al., 2005). One problem with suh surveys is that most onsumers are unlikely to understand a NPV alulation (Kurani and Turrentine, 2004). It has been posited that when making an atual purhase a onsumer might seek out expert information regarding lifetime osts Table 7 Summary of base ase and onsumer ase. Base ase (%) Consumer ase (%) Disount rate 5 20 Down payment 100 31 Loan interest rate 8.7 ð2þ (Santini and Vyas, 2005). This idea is supported by findings suggesting that impliit onsumer disount rates deline as purhase prie inreases. A study onduted looking at refrigerator purhases found that onsumers exhibit impliit disount rates of about 45% (Meier and Whittier, 1983). There is a possibility that some of these refrigerators were purhased by landlords, or others that were not paying the utility osts and therefore had little inentive to purhase an effiient model (prinipal-agent problem). Other studies fousing on retirement plans instead of appliane purhases found lower disount rates ranging from 1% 26% whih may be attributed to the greater value (perhaps supporting the idea that an individual thinks more arefully about a finanial deision of larger amount) (Gilman, 1976; Matthew, 1984; Steven et al., 1982; Warner and Pleeter, 2001). These studies also found that in general those with higher inomes and eduation levels exhibited lower disount rates. It is possible that studies fousing on retirement deisions are biased toward higher inome households who exhibit lower disount rates. The newer of these studies examined the military drawdown of the early 1990s and the deision servie members faed about aepting either a lump sum payment or annuity. It found that disount rates varied onsiderably among servie members depending on whether they were enlisted or offiers (Warner and Pleeter, 2001). It was found that offiers had a disount rate of 12% and enlisted members had a disount rate of 26%. In aggregate the disount rate was 18%. Given that surveys regarding purhase of alternative vehile found a likely implied disount rate near 20% and that the atual deision regarding retirement resulted in an implied rate of 18%, we use r¼20% to represent our onsumer ase. It has been reported that over 80% of new vehiles were purhased using a loan (between 1998 and 2003) and that the median loan had a period of 60 months and rate of 8.7% with down payment of 14% (Agarwal et al., 2008). Twenty perent of new vehiles are not purhased on a loan. Our onsumer behavior ase assumes an intermediate ondition, where onsumers take a 60 month loan at 8.7% with a 31% down payment. Table 7 summarizes the base ase and onsumer ase. 3. Results Fig. 4 summarizes results for the base ase, and Fig. 5 shows results for the onsumer behavior ase. Eah ombination of Fig. 4. Base ase results. Lifetime gasoline onsumption vs. NPV of vehile, hargers, and fuel osts ompared to CV given a 12 year vehile life, AEO traditional high oil prie, GREET 1.8d 2015 effiieny, 2015 average vehile osts from ANL (2011) with 50% markup, and a 5% disount rate. Vehiles and hargers are purhased outright. PHEV AER values inrease in 5 mile inrements from 5 to 25 and are labeled on the AllStops harging senario for larity.
S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 435 Fig. 5. Consumer behavior ase. Disount rate at 20% for vehile and fuel; 31% down payment; 8.7% loan interest rate. Other values the same as Fig. 4. vehile lass, vehile type (inluding CV, HEV, and PHEVs of varying AER), and harging senario is plotted in two dimensions: lifetime ost premium (vs. the ost of the CV) and lifetime gasoline onsumption. For larity, PHEV AER is labeled only on the AllStops harging senario, but in all ases inreasing AER from 5 to 25 miles auses a derease in gasoline onsumption, so AER inreases from top to bottom in all graphs. Also for larity, the HomeEve and HomeAll harging senarios are averaged to represent the home harging senario (shown with diamonds), and error bars indiate variation under HomeEve and HomeAll senarios. The WorkHomeEve and WorkHomeAll senarios were averaged in the same way (shown with squares). Given these two objetives, points on the lower left are preferred, and points toward the upper right are dominated. For example, in the ase of ars, both the CV and HEV are dominated by the PHEV5 with home harging, sine it offers more gasoline savings at lower lifetime ost. Generally, the PHEVs with home harging are Pareto optimal, and additional harging infrastruture provides slightly more gasoline savings at signifiantly higher ost. If the market were to adopt the alternative with lowest lifetime ost, it would be PHEV5 with home harging only for ars and truks and HEV for SUVs. Purhase of additional AER or harging infrastruture represents a tradeoff of additional ost vs. additional fuel savings. For example, in the base ase for ars, shown in Fig. 4, the PHEV10 is $275 more expensive than the PHEV5 over the vehile s lifetime and saves 165 gallons of gasoline. This implies a value of $1.61 per gallon saved (if gasoline savings is the only benefit of interest). In ontrast, paying for additional harging infrastruture requires $9.99 per gallon saved for workplae infrastruture and $19.50 per gallon saved for AllStops infrastruture. For omparison, Mihalek et al. (2011) summarize estimates of the US oil premium, inluding inreases in supply disruption osts assoiated with inreased US oil onsumption of $0.10/gal ($0 $0.30/gal) (Brown, 2010), inreases in payment of world oil pries due to additional US onsumption of $0.24/gal ($0.08 $0.49/gal) (Leiby, 2007), and inreases in military osts due to US oil onsumption of $0.03/gal ($0 $0.17/gal) (Deluhi and Murphy, 2008), resulting in a total oil premium of $0.37/gal ($0.08 $0.96/ gal) in 2015 dollars (using a projeted GDP prie deflator (US Department of Commere, 2010)). This implies that publi spending to inrease vehile AER would not pay bak in oil premium osts, even in an optimisti senario, and publi spending to inrease harging infrastruture is orders of magnitude more expensive than oil premium osts. The onsumer behavior ase is shown in Fig. 5. Using a higher disount rate, similar results are found for relative benefits of inreased AER and harging infrastruture, however not all PHEV options result in lifetime savings. In the ase of truks only HEV and PHEV5 derease lifetime ost ompared to CVs. HEVs are the lowest lifetime ost options for SUVs and truks. We summarize the results presented in Figs. 4 and 5 as follows. First, HEVs and PHEVs save gasoline over onventional vehiles in all ases. Seond, HEVs and some PHEVs an save both gasoline and total lifetime osts over onventional vehiles both at normative and observed impliit disount rates. Third, the additional ost per gallon saved of alternatives other than the least-ost option in eah ase is higher than oil premium estimates, and harging infrastruture is orders of magnitude more expensive per gallon saved, even with our optimisti assumptions for harging infrastruture. These findings are robust aross a wide range of sensitivity senarios examined in the supplemental information. 4. Limitations and assumptions Our base ase and onsumer behavior ase analysis is based on several assumptions that should be understood to interpret results appropriately. Where possible, impliations of assumptions are examined expliitly in the supplemental material, inluding sensitivity analysis on future fuel pries, vehile retail ost and markup, disount rate, vehile life, harging infrastruture osts, nondomesti harging infrastruture requirements, battery replaement, vehile effiieny, harging rate, and vehile loan down payment. Our key findings are robust to variation in these assumptions (see supplemental information). Individual driving behavior an vary from the aggregate survey data, and PHEVs will benefit some drivers more than others. Shiau et al. (2010) found that aounting for individual driving distanes when alloating vehiles is of seondary importane for overall ost and emissions; however, variations in driving onditions (drive yle, terrain, temperature, et.) may have a signifiant effet on relative benefits (Karabasoglu and Mihalek, in review). Additional limitations involve impliations of having
436 S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 different buyers for vehiles vs. harging infrastruture, short vs. long term infrastruture requirements, non-ost reasons for vehile adoption, soial benefits of plug-in vehiles other than gasoline savings, and indiret effets. We desribe eah of these in turn. First, we do not address how differenes in who pays for vehile, harger, and fuel osts will influene inentives or how infrastruture investment osts will be passed down to vehile owners who purhase eletriity. Not all vehile owners have aess to off street parking or a private garage. While vehile owners who are home owners with a garage are likely to pay for home harge points themselves, landlords are less likely to invest in harging infrastruture for tenant use, and vehile owners are unlikely to pay for infrastruture on property they do not own. The likely buyers of harge points are property owners where the harge point is installed, eletri utilities, or harge point network operators. Our estimates of nondomesti infrastruture requirements are intended to be optimisti. However, if the typial life of harging infrastruture equipment is longer or shorter than the life of vehiles (onsidering design life as well as vandalism, aidents, maintenane, and obsolesene), it will affet the number of harge points needed per vehile in the long run and thus osts. Investment in harging infrastruture does not neessarily imply use, and some infrastruture is likely to be underutilized due to poor site loation, unertainty, demographi hanges, finanial inentives, or inonveniene. These fators are expeted to make harging infrastruture less ompetitive, strengthening our onlusions. Seondly, we do not address other fators that affet vehile adoption besides lifetime osts. For example, HEVs were not initially ost ompetitive with CVs, yet individuals adopted them for other reasons, inluding symbolism and image (Lave and MaLean, 2002; Heffner et al., 2007). We also assume a single owner purhases a vehile and drives it until end of life with no salvage value. Given a onsumer disount rate of 20%, a salvage value as high as $3000 in year 12 is worth only $335 at purhase, so we assume salvage value of the vehile and batteries is negligible. We onsider only gasoline savings and no other benefits of plug-in vehiles. Other benefits inlude hanges in negative externalities from life yle air emissions, ground emissions, and noise as well as positive externalities assoiated with knowledge spillover in tehnology advanement (Mihalek et al., 2011). Differenes in these negative externalities among vehiles and harging senarios is relatively small on average, and we expet that positive externalities assoiated with advanement of harger tehnology and large-battery plug-in vehiles would not substantially exeed those assoiated with small-battery plug-in vehiles, if at all. Finally, we onsider only diret effets and ignore indiret effets. One indiret effet of subsidizing EVs is to aelerate tehnology development, potentially leading to an earlier breakthrough of mass-marketed EVs and resulting gasoline savings. This effet is unertain, but if urrent subsidies result in earlier mainstream adoption of fuel-effiient vehiles, ost effetiveness ould be higher than reported here. However, other indiret effets ould redue or eliminate the net effetiveness of vehile and harger subsidies. In partiular, the Congressional Budget Offie estimates that beause CAFE standards are high enough to be binding for automakers, subsidies spent to enourage adoption of EVs will have pratially zero effet on total gasoline onsumption (CBO, 2012). The inentivized sale of gasoline-saving vehiles will simply allow automakers to sell more gas-guzzling vehiles under CAFE regulation, ahieving the same net average fuel onsumption to omply with the law. If this is the ase, then the ost effetiveness of EV subsidies may be redued to zero. 5. Summary and onlusions Using assumptions strongly favorable to harging infrastruture, we find that paying for harging infrastruture results in lower gasoline savings per dollar spent than paying for inreased PHEV battery apaity, and both approahes are more ostly per gallon saved than US oil premium estimates. Comparing the subsidy neessary to ahieve lifetime ost parity with the least ost option for eah vehile lass in the base ase, we find that the maximum ost per gallon saved for inreased AER is 5% 40% less than the minimum ost per gallon saved when installing harging infrastruture, depending on vehile lass. Looking forward as battery pries derease and the AER resulting in maximum lifetime ost savings inreases, the relative value of plugging in multiple times throughout the day will also deline. Nondomesti harging infrastruture is generally not neessary for operation of PHEVs, and substantial gasoline displaement an be ahieved solely with home harging. In ontrast, the limited range of BEVs make nondomesti harging infrastruture more ritial if the vehiles are to be used as primary vehiles. But publi investment in either large-battery vehiles or harging infrastruture generally produes fewer benefits per dollar spent than investment in small-battery PHEVs (Mihalek et al., 2011), suggesting that subsidizing sales of BEVs and installation of harging infrastruture are not the most effiient use of limited publi funds. If the purpose of existing federal PHEV subsidies is to redue gasoline onsumption, this implies that the poliy subsidizes 4 kwh battery PHEVs at $1.25 per gallon saved while subsidizing 16 kwh battery PHEVs at roughly $4.50 per gallon saved (Fig. 6), ignoring indiret effets. It is lear that federal subsidies are not urrently aligned with the goal of dereased gasoline onsumption in a onsistent and effiient manner. Other relevant poliy objetives, inluding redution of emissions externalities, enouragement of tehnology development, and job reation do not show lear benefits of favoring large battery paks over small battery paks (Mihalek et al., 2011). Fig. 6. Comparison of urrent federal subsidy to base ase assumptions showing lifetime fuel savings (HomeEve harging senario). An EPA estimate based on the Chevy Volt s reported effiieny is also inluded for omparison (Download Fuel Eonomy Data. US Environmental Protetion Ageny, 2011). The federal subsidy signifiantly favors larger battery paks to a stronger degree than their potential for additional gasoline savings.
S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 437 Alternatives to the urrent federal subsidy struture inlude measures to redesign or replae the urrent poliy. A poliy redesign should onsider the following issues: Current federal subsidies are tied to total battery apaity rather than usable battery apaity or AER, whih inentivizes use of larger battery paks. The Chevy Volt, for example, uses only about 65% of its 16 kwh apaity in order to improve safety and battery life (Peterson et al., 2010; US Department of Energy, 2011a). Subsidizing usable apaity, rather than total apaity, would remove the disinentive for automakers to figure out how to use a larger portion of the battery (Shiau et al., 2010). Alternatively, subsidizing based on AER (as measured in a standardized test) would also enourage automakers to make vehiles more effiient, and removing the exlusion for lower-apaity lower-range vehiles would be more onsistent with potential benefits. PHEVs have diminishing returns in gasoline savings as battery apaity inreases. Subsidies intended to generate gasoline savings would be better if tied to estimated gasoline savings rather than battery apaity or AER, and subsidies that are tied to battery apaity or AER should avoid a fixed rate per kwh or per mile and instead reflet the struture of diminishing returns. However, methods for estimating gasoline savings may be ontroversial, and depending on what referene point is used, subsidies tied to gasoline savings ould have unintended onsequenes, suh as the potential for separate referene points in eah vehile lass enouraging onsumers to purhase larger vehile lasses. The urrent subsidy of $2500 for 4 kwh ($1.25/gal saved) and $7500 for 16 kwh ( $4.50/gal saved) pays pries substantially higher than US oil premium estimates of $0.37/gal ($0.08 $0.96/gal). Subsidies intended to generate gasoline savings would preferably be omparable to the soial value of gasoline savings (and the value of other soial benefits). To the extent that larger subsidies are able to kik-start adoption and sustainable market aeptane of plug-in tehnologies that would not otherwise be adopted, temporary larger subsidies may be warranted. But the magnitude or duration of this dynami effet remains highly unertain. More effiient poliies generally target the poliy goal, suh as gasoline displaement, diretly rather than a proxy, suh as battery size. On eonomi effiieny grounds, subsidies are justified insofar as they orret for positive externalities, suh as innovation knowledge spillover, and researh funding is an alternative to subsidizing sales for ahieving this effet. A more effiient way to address negative externalities is to apply Pigovian taxes, whih would inrease the prie of gasoline and make plug-in vehiles more ompetitive in the marketplae while enouraging the most effiient responses to reduing externalities, inluding not only alternative powertrains but also effiieny improvements and inentives to drive less and purhase smaller vehiles (as well as to make hanges in other setors of the eonomy). Revenue from suh taxes ould be used to redue taxes elsewhere on outomes we prefer to enourage, like employment. We aknowledge the politial hallenge of inreasing or reating a tax. Finally, onsidering the presene of binding CAFE standards, it remains in question whether EV subsidies will provide any net gasoline savings for the foreseeable future (CBO 2012). Ignoring interations with CAFE poliy, HEVs and PHEVs with low AER and only home harging generally provide the largest diret gasoline savings per dollar spent, offering both lower osts and lower gasoline onsumption than CVs, depending on the onsumer s disount rate. It is therefore possible that inentivizing a larger number of onsumers to purhase HEVs or low-aer PHEVs would save more gasoline under a fixed poliy budget than inentivizing a relatively smaller number of onsumers to purhase high-aer PHEVs (Mihalek et al., 2011). However, given a fixed market of eletrified vehile adopters, if more gasoline savings is needed than what an be ahieved with HEVs and low-aer PHEVs, additional savings an be ahieved more effiiently by paying for additional AER than by paying for extra harging infrastruture. Aknowledgments The authors thank Jay Apt and H. Sott Matthews for useful disussions. This work was supported in part by the Carnegie Mellon Philip and Marsha Dowd Engineering Seed Fund, Toyota Motor Corporation, the Eletri Power Researh Institute under grants to the Carnegie Mellon Eletriity Industry Center (CEIC), and the National Siene Foundation CAREER Grant 0747911. The views expressed are those of the authors and do not neessarily represent the views of the sponsors. We thank the anonymous reviewers for helpful suggestions. Appendix A. Supporting information Supplementary data assoiated with this artile an be found in the online version at http://dx.doi.org/10.1016/j.enpol.2012.09.059. Referenes Advaned Vehile Testing Ativity, 2011. US Department of EnergyWashington, DC. /http://www1.eere.energy.gov/vehilesandfuels/avta/light_duty/m/arra_ vehile_rpts.htmls (aessed 14.12.11). Agarwal, S., Ambrose, B.W., Chomsisenghet, S., 2008. Eonomi Perspetives 32, 17 28. Annual Energy Outlook, 2011. US Energy Information Administration, Washington DC. Brown, H.H., 2010. Estimating US Oil Seurity Premiums. Resoures for the Future, Washington, DC. Congressional Budget Offie, 2012. Effets of federal tax redits for the purhase of eletri vehiles. In: Proeedings of the Congress of the United States, September 2012. Deluhi, M.A., Murphy, J.J., 2008. Energy Poliy 36, 2253. Dogget, S., 2011. Coulomb Gets $37 Million from Energy Dept. to Deploy 4600 ev- Charging Stations. /http://blogs.edmunds.om/greenaradvisor/2010/06/ou lomb-gets-37-million-from-energy-dept-to-deploy-4600-ev-harging-sta tions.htmls (aessed 20.02.11). Download Fuel Eonomy Data. US Environmental Protetion Ageny, 2011. Washington, DC. /http://www.fueleonomy.gov/feg/download.shtmls. EPA and NHTSA, 2010. Finalize Histori National Program to Redue Greenhouse Gases and Improve Fuel Eonomy for Cars and Truks. US Environmental Protetion Ageny, Washington DC. /http://www.epa.gov/otaq/limate/regula tions/420f10014.htms (aessed 11.05.11). Energy Independene and Seurity At of 2007, 2007. Publi Law. 110 140, Congress. /http://www.gpo.gov/fdsys/pkg/plaw-110publ140/html/plaw-110publ140. htms (aessed 26.11.11). Energy Poliy and Conservation At, 1975. Publi Law. 94 163. Congress. Gilman, H.J., 1976. Determinants of Impliit Disount Rates: An Empirial Examination of the Pattern of Voluntary Pension Contributions of Employees in Four Firms. Center for Naval Analyses, Arlington, VA. Hadley, S.W., Tsevetkova, A., 2008. Potential Impats of Plug-in Hybrid Eletri Vehiles on Regional Power Generation. US Department of Energy, Washington DC. Heffner, R.R., Kurani, K.S., Turrentine, T.S., 2007. Transportation Researh Part D: Transport and Environment 12, 396. Horne, M., Jaard, M., Tiedemann, K., 2005. Energy Eonomis 27, 59 77. Karabasoglu, O., Mihalek, J.J. In preparation. Influene of driving patterns on life yle benefits of hybrid and plug-in eletri vehiles. Energy Poliy, in review. Kurani, K.S., Turrentine, T.S., 2004. Automobile Buyer Deisions About Fuel Eonomy and Fuel Effiieny. University of California Institute of Transportation Studies. Lave, L.B., MaLean, H.L., 2002. Transportation Researh Part D: Transport and Environment 7, 155 162. Leiby, P.N., 2007. Estimating the Energy Seurity Benefits of Redued US Oil Imports. Oak Ridge National Lab, Washington, DC. Lemoine, D.M., Kammen, D.M., Farrell, A.E., 2008. Environmental Researh Letters 3, 1 10.
438 S.B. Peterson, J.J. Mihalek / Energy Poliy 52 (2013) 429 438 Markus, F., 2010. Unbolting the Chevy Volt to See How it Tiks. Motor Trend. Matthew, B., 1984. Personal Disount Rates: Estimates for the Military Population. US Department of Defense, Washington, DC. Mau, P., Eyzaguirre, J., Jaard, M., Collins-Dodd, C., Tiedemann, K., 2008. Eologial Eonomis, 504 516. Meier, A.K., Whittier, J., 1983. Energy 8, 957 962. Mihalek, J.J., Chester, M., Jaramillo, P., Samaras, C., Shiau, C.S.N., Lave, L.B., 2011. Proeedings of the National Aademy of Sienes of the United States of Ameria 108 (40), 16554 16558. Morrow, K., Karner, D., Franfort, J., 2008. Plug-in Hybrid Eletri Vehile Charging Infrastruture Review. Department of Energy, Washington DC. National Household Travel Survey, 2010. US Department of Transportation. Washington DC. Neubauer, J., Brooker, A., Wood, E., 2012. Journal of Power Soures 209, 269 277. Nishimoto A., 2011. Nissan Announes Low-Cost DC Quik Charger for US. Motor Trend. Patil, R., Adornato, B., Filipi, Z., 2009. Impat of Naturalisti Driving Patterns on PHEV Performane and System Design. SAE International, Warrendale, PA. Peterson, S.B., Apt, J., Whitare, J.F., 2010. Journal of Power Soures 195, 2385 2392. Peterson, S.B., Whitare, J.F., Apt, J., 2011. Environmental Siene & Tehnology 45, 1792 1797. Plotkin, S., Singh, M., 2009. Multi-Path Transportation Futures Study: Vehile Charaterization and Senario Analyses. US Department of Energy, Washington, DC. Raykin, L., Roorda, M.J., MaLean, H.L., 2012. Transportation Researh Part D. Transport and Environment 17, 243 250. Santini, D.J., Vyas, A.D., 2005. Suggestions for New Vehile Choie Model Simulating Advaned Vehiles Introdution Deisions (AVID): Struture and Coeffiients. US Department of Energy, Washington, DC. Shiau, C.-S.N., Samaras, C., Hauffe, R., Mihalek, J.J., 2009. Energy Poliy 37, 2653 2663. Shiau, C.-S.N., Kaushal, N., Hendrikson, C.T., Peterson, S.B., Whitare, J.F., Mihalek, J.J., 2010. ASME Journal of Mehanial Design 132 (091013), 1 11. Steven, C., Goldberg, M.S., Hogan, P.F., Mairs, L., 1982. The Personal Disount Rate: Evidene from Military Career Deisions, US Department of the Navy, Washington, DC. The Amerian Reovery and Reinvestment At of 2009, 2009. Publi Law. 111 5, Congress. /http://frwebgate.aess.gpo.gov/gi-bin/getdo.gi?dbname¼111_ ong_bills&doid¼f:h1enr.pdfs (aessed 26.11.10). The Greenhouse Gases, Regulated Emissions, and Energy Use in Transportation (GREET) Model, 2010. version1.8d. US Department of Energy, Washington DC. Traut E.C., Hendrikson C.T., Klampfl Y.L., Mihalek J.J. Optimal design and alloation of eletrified vehiles and dediated harging infrastruture for minimum life yle greenhouse gas emissions and ost. Energy Poliy. http://dx.doi.org/10.1016/j.enpol.2012.08.061, in press. US Department of Commere, 2010. Impliit Prie Deflator, 1949 2009. US Department of Commere, Washington, DC. US Department of Energy, 2011a. Fuel Eonomy Guide: Model Year, 2012. US Department of Energy, Washington, DC. /www.fueleonomy.gov/s. US Department of Energy, 2011b. Potential of Tehnologies for Displaing Gasoline Consumption by Light-duty Vehiles Through 2045. Washington DC. US Environmental Protetion Ageny (EPA) and the US National Highway Traffi Safety Administration (NHTSA), 2012. 2017 and Later Model Year Light-Duty Vehile Greenhouse Gas Emissions and Corporate Average Fuel Eonomy Standards, 2012. /http://www.nhtsa.;gov/statifiles/rulemaking/pdf/afe/2017-25_cafe_final_rule.pdfs (aessed 10.09.12). Warner, J.T., Pleeter, S., 2001. The Amerian Eonomi Review 91, 33 53. Whitefoot, K.S., Skerlos, S.J., 2012. Energy Poliy 41, 402 411. Witkin, J., 2012. In California, a Fast-Charging First. New York Times, April 10, 2012.