Emerging Technology Zero Emission Vehicle Household Travel and Refueling Behavior

Emerging Technology Zero Emission

Vehicle Household Travel and

Refueling Behavior

(CARB Contract 16RD009)

April 19, 2021

Prepared By:

Gil Tal

Vaishnavi Chaitanya Karanam

Matthew P. Favetti

Katrina May Sutton

Jade Motayo Ogunmayin

Seshadri Srinivasa Raghavan

Christopher Nitta

Debapriya Chakraborty

Adam Davis

Dahlia Garas

Prep

ared For:

Table of Figures i

Table of Contents

Table of Figures ............................................................................................................................... iv

List of Tables ................................................................................................................................... ix

Abstract ........................................................................................................................................... xi

Preface ............................................................................................................................................ xi

Acknowledgements ........................................................................................................................ xii

Executive Summary ......................................................................................................................... 1

1 Introduction ............................................................................................................................. 3

2 Recruitment and Background Survey ...................................................................................... 8

Logger Installation Process ............................................................................................... 9

Data Collection and Limitations ..................................................................................... 12

Data Annualization ......................................................................................................... 13

Sampling of the Logged Participant Households ........................................................... 14

PHEV eVMT Calculation .................................................................................................. 18

3 Logger Data: Vehicle Level Analysis of PEVs .......................................................................... 21

Data Description ............................................................................................................. 21

Battery Electric Vehicles Driving .................................................................................... 26

PEVs Used for Commuting ............................................................................................. 36

Battery Electric Vehicle Charging ................................................................................... 37

Plug-in Hybrid Electric Vehicles (PHEVs) Driving ............................................................ 48

Plug-in Hybrid Electric Vehicle Charging ........................................................................ 57

Charging Distance Based on GPS Location of PEVs ........................................................ 63

4 Household Level Analysis of PEVs ......................................................................................... 67

Households with a BEV Only or BEV and ICEV ............................................................... 71

Households with a PHEV Only or PHEV and ICEV .......................................................... 76

Two-PEV Households: BEV and PHEV Mix ..................................................................... 85

Table of Figures ii

UF and GHG Profile ........................................................................................................ 89

Additional ICE Usage Metrics ......................................................................................... 94

5 PHEV Engine Starts Analysis ................................................................................................ 100

Cold Starts .................................................................................................................... 100

Proportion of Days with Engine Starts ......................................................................... 101

Engine Start Event Description ..................................................................................... 102

Travel Conditions at Engine Start ................................................................................. 103

Potential Emission Impacts of Engine Starts ................................................................ 112

Engine Starts Discussion ............................................................................................... 114

6 Fuel Cell Vehicle Analysis ..................................................................................................... 115

Fuel Cell Vehicle Driving ............................................................................................... 118

Fuel Cell Vehicle Refueling ........................................................................................... 123

Fuel Cell Household Analysis ........................................................................................ 128

7 Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews ........ 132

Introduction to Charging Management Strategies ...................................................... 133

Methods ....................................................................................................................... 133

Results .......................................................................................................................... 133

Interviews Discussion ................................................................................................... 142

Interviews Policy Discussion ......................................................................................... 143

8 Conclusions .......................................................................................................................... 144

Table of Figures iii

9 Glossary ............................................................................................................................... 146

10 References ........................................................................................................................... 148

11 Appendix A ........................................................................................................................... 153

Table of Figures iv

Table of Figures

Figure 1. California LDV Fleet Composition (2018) by Fuel Type .................................................... 3

Figure 2. Share of BEVs and PHEVs in New Vehicle Registration .................................................... 4

Figure 3. PEVs as share of all vehicles ............................................................................................. 5

Figure 4. Overview of Number of Logger Installations During Each Phase of the Project ........... 11

Figure 5. Home and Daytime Charging Locations 2015-2020....................................................... 12

Figure 6. Distribution of Household Income: Survey Respondents and Logged Households ...... 16

Figure 7. Distrubution of Household Size Among Survey Respondents and Logged Households 17

Figure 8. Average Annualized VMT by PEV Model ........................................................................ 24

Figure 9. BEVs: Percentage Share of Total VMT by Trip Speed (in mph) ....................................... 25

Figure 10. PHEVs: Percentage Share of Total VMT by Trip Speed (in mph) .................................. 26

Figure 11. Average and Maximum Trip Distance on Weekdays and Weekends by BEV Type ...... 28

Figure 12. Percentage of Trips by Trip Distance Bins (miles) and by BEV Type ............................. 28

Figure 13. Effect of Speed on Energy Consumption per Mile ....................................................... 29

Figure 14. Average Daily VMT of the Individual BEVs by BEV Model ........................................... 30

Figure 15. Proportion of annual VMT on days with and without routine destinations................ 32

Figure 16. Trip destinations of Tesla Model S-60_80 .................................................................... 33

Figure 17. Standard deviation ellipse of vehicle destinations plotted against annual VMT ......... 34

Figure 18. Percentage of Daily VMT by Distance Bins: Weekdays vs. Weekends ......................... 35

Figure 19. Share of VMT on LDT (50 miles or more) as Percentage of Total VMT by BEV Type ... 36

Figure 20. Number of PEVs Used for Commute Purposes by Type. .............................................. 37

Figure 21. Probability of Charging Within the Logging Window of Individual BEVs by BEV Type 39

Figure 22. Share of Charging Sessions by Charging Level and BEV Type ...................................... 40

Figure 23. Share of Charging kWh by Charging Level and BEV Type ............................................ 40

Figure 24. Charging Session Starting Time: Weekdays vs. Weekends .......................................... 41

Figure 25. Percentage of L1 Charging Start Times by Time of Day and BEV Type ........................ 42

Figure 26. Percentage of L2 Charging Start Times by Time of Day and BEV Type ........................ 42

Table of Figures v

Figure 27. Percentage of DCFC Charging Start Times by Time of Day and BEV Type ................... 43

Figure 28. Average L1 Charging kWh Charged and Charging Duration: Weekdays vs Weekends 44

Figure 29. Average L2 Charging kWh Charged and Charging Duration: Weekdays vs Weekends 45

Figure 30. Average DCFC Charging kWh and Duration: Weekdays vs Weekends ......................... 45

Figure 31. L1 Charging: Average Starting and Charged SOC ......................................................... 46

Figure 32. L2 Charging: Average Starting and Charged SOC ......................................................... 47

Figure 33. DCFC: Average Starting and Charged SOC .................................................................... 47

Figure 34. PHEV eVMT, gVMT, and VMT of Individual PHEVs by PHEV Type ................................ 49

Figure 35. Utility Factor (UF) for Each PHEV by Type .................................................................... 50

Figure 36. Average UF by PHEV Type ............................................................................................ 50

Figure 37. Average Trips per Day by Driving Mode ....................................................................... 51

Figure 38. Percentage of Total PHEV Trips by PHEV Driving Mode ............................................... 52

Figure 39. Share of Trips by Trip Distance Bins: Weekdays vs Weekends ..................................... 53

Figure 40. Daily Average VMT, eVMT, and gVMT Share by PHEV Type ......................................... 54

Figure 41. Share of ZE Days, CS Days and CDB/CS Days ................................................................ 55

Figure 42. Share of Long-Distance Travel (LDT; 50 miles+) Days: Weekdays vs Weekends .......... 56

Figure 43. Share of Long-Distance Travel (LDT; 100 miles+) Days: Weekdays vs Weekends ........ 56

Figure 44. Share of Daily VMT by Distance Bin: Weekdays vs Weekends ..................................... 57

Figure 45. Share of Charging Sessions Charged Energy by Charging Level ................................... 59

Figure 46. Share of Total Number of Sessions by Charging Level ................................................. 60

Figure 47. Average L1 and L2 Charging kWh/Session: Weekdays vs Weekends ........................... 61

Figure 48. Average L1 and L2 Charging Session Duration : Weekdays vs Weekends .................... 62

Figure 49. Percentage of Charging Sessions Starting Time (L1 and L2): Weekdays ...................... 62

Figure 50. Percentage of Charging Sessions Starting Time (L1 and L2): Weekends ..................... 63

Figure 51. Percentage of Charging Sessions More Than 1 Mile From Home ................................ 64

Figure 52. Percentage of DCFC Charging Sessions by BEV Type and Distance from Home .......... 65

Figure 53. Percentage of DCFC Charging Sessions by Distance from Last Night Location ............ 66

Table of Figures vi

Figure 54. Percentage of Level 2 Charging by BEV Type and Distance from Last Night Location . 67

Figure 55. Composition of Households Included in the Analysis .................................................. 69

Figure 56. Number of Households with only One BEV or PHEV Logged in the Study .................. 70

Figure 57. Household Utility Factor by BEV Type and Household Car Composition .................... 72

Figure 58. Average Daily VMT in HHs with BEVs, Showing the eVMT and gVMT Percentages .... 73

Figure 59. Percentage of BEV and ICEV trips in HHs with BEVs .................................................... 74

Figure 60. Number of Days/Year BEV was Used for Long Distance Travel (LDT) .......................... 75

Figure 61. Annualized Number of L1 or L2 charger (L1/L2), and DCFC Sessions in BEV HHs ....... 76

Figure 62. Individual HH UF by PHEV Type in PHEV HH ................................................................ 82

Figure 63. Average HH UF by PHEV Type; and Average HH UF by Number of Cars in the HH ...... 82

Figure 64. Average HH UF by Number of Cars per HH and PHEV Type ......................................... 83

Figure 65. Percentage of Household Trips Powered by Different PHEV Driving Modes or ICEVs . 84

Figure 66. Daily Average HH VMT and Percentage of PHEV eVMT, PHEV gVMT, and ICE gVMT .. 85

Figure 67. Daily Average HH VMT, and Percentage of eVMT and gVMT BEV-PHEV Households . 87

Figure 68. Average Annual HH VMT and Share of BEV eVMT and PHEV eVMT in BEV-PHEV HHs 88

Figure 69. Two car HHs VMT by Vehicle Type, PEV UF and HH UF ............................................... 90

Figure 70. Average GHG per Mile and Utility Factor ..................................................................... 91

Figure 71. Household Level GHG and Utility Factor per Mile ....................................................... 92

Figure 72. Ratio of PEV and ICE GHG/Mile to Total HH GHG/Mile ............................................... 93

Figure 73. Annualized ICE VMT in 2 Car HHs (Single ICE and Single PHEV or BEV) by PEV Type. . 95

Figure 74. Annualized ICE VMT in 3 Car HHs (Two ICEs and Single PHEV or BEV) by PEV Type ... 96

Figure 75. PHEV and ICE Use (Days/Year) for Long Distance Travel. ............................................. 97

Figure 76. BEV and ICE Use (Days/Year) for Long Distance Travel in 2 Car HHs. .......................... 98

Figure 77. PHEV and ICE Use (Days/Year) for Long Distance Travel in 3 Car HHs. ........................ 99

Figure 78. BEV and ICE Use (Days/Year) for Long Distance Travel in 3 Car HHs. ........................ 100

Figure 79. Share of Drive Days with No Engine Starts ................................................................ 102

Figure 80. Engine-on Time Trace ................................................................................................. 102

Table of Figures vii

Figure 81. SOC at First Engine Start ............................................................................................ 104

Figure 82. Maximum Power Requirement 5 Seconds before Engine Start ................................. 105

Figure 83. Catalyst Temperature at Engine Start ........................................................................ 106

Figure 84. Prius Plug-in-4.4 Soak Time by SOC at Engine Start ................................................... 107

Figure 85. C-max Energi Soak Time by SOC at Engine Start ........................................................ 107

Figure 86. Prius Prime-8.8 Soak Time by SOC at Engine Start .................................................... 108

Figure 87. Pacifica-16 Soak Time by SOC at Engine Start ............................................................ 108

Figure 88. Volt-16 Soak Time by SOC at Engine Start ................................................................. 109

Figure 89. Volt-18 Soak Time by SOC at Engine Start ................................................................. 109

Figure 90. ICE Soak Time for the Conventional Gasoline Vehicles in Households ...................... 110

Figure 91. Distance from Start of Day to First Engine Start of Day for all PHEVs ........................ 111

Figure 92. Distance from Start of Trip to First Engine Start of Trip for all PHEVs ........................ 112

Figure 93. Probability of Engine Start per Vehicle Model ........................................................... 113

Figure 94. Average Daily Cold Starts ........................................................................................... 114

Figure 95. Fuel Cell Vehicle Sales by Model, Country, and Model for California ........................ 116

Figure 96. Annualized VMT of Mirais .......................................................................................... 117

Figure 97. Mirai: Percentage Share of Total VMT by Trip Speed (in mph) .................................. 118

Figure 98. Average and Maximum Trip Distance on Weekdays and Weekends by Mirai ........... 119

Figure 99. Percentage of Trips per Vehicle by Trip Distance Bins (miles) ................................... 120

Figure 100. Average Monthly Vehicle VMT Across Deployments ............................................... 121

Figure 101. Average Daily FCV VMT per Vehicle ......................................................................... 122

Figure 102. Average FCV Energy Efficiency (miles/kg(H2) .......................................................... 122

Figure 103. Time (in minutes) Elapsed between Refuels ............................................................ 123

Figure 104. Distance Traveled Between Refuels ......................................................................... 124

Figure 105. Average Fuel in during Refuel Days per Vehicle ....................................................... 124

Figure 106. Number of Hydrogen Fueling Stations within 140 miles of a Block Group ............. 125

Figure 107. Areas within 5 and 10 Miles from a Refueling Station ............................................. 126

Table of Figures viii

Figure 108. Average Fuel pumped in during Refuel, per Station ................................................ 127

Figure 109. Refueling Share per Station ..................................................................................... 128

Figure 110. Household Utility Factor per Vehicle by Household Car Composition .................... 129

Figure 111. HH Average Daily VMT in HHs with FCVs, Showing eVMT and gVMT Percentages . 130

Figure 112. Percentage of FCV and ICEV trips in HHs with FCV .................................................. 130

Figure 113. Number of Days/Year FCV was Used for Long Distance Travel (LDT). ...................... 131

Figure 114. Percentage of FCV and ICEV LDT (>100 mi) Days over LDT Days in HHs with FCV ... 132

Figure 115. Interviewees reported home based charging (N=47) .............................................. 136

Figure 116. How often interviewees reported charging their vehicles in public (N=40). ........... 138

Figure 117. Break down of interviewee's work charging patterns and availability .................... 139

Figure 118. Three interviewees did not go into enough detail about their charging strategies 142

List of Tables ix

List of Tables

Table 1. Sociodemographics and Vehicle Types Among the Usable Surveyed Participants ......... 15

Table 2. Battery Capacity and Vehicle Types Among the Logged Participants ............................. 17

Table 3. BEV Driving Data Overview .............................................................................................. 22

Table 4. BEV Charging Data Overview ........................................................................................... 22

Table 5. PHEV Driving Data Overview ........................................................................................... 23

Table 6. PHEV Charging Data Overview ........................................................................................ 23

Table 7. FCV Driving Data Overview .............................................................................................. 24

Table 8. FCV Refueling Data Overview .......................................................................................... 24

Table 9. Annualized VMT by Vehicle Types ................................................................................... 25

Table 10. BEV Driving Trip Level Summaries (on days when the BEV was driven) ....................... 27

Table 11. Summary statistics for area of SDE activity space by vehicle type ................................ 34

Table 12. Charging Summaries on by BEV Type ............................................................................ 38

Table 13. PHEV VMT, eVMT, gVMT, Fuel and Energy Consumption by PHEV Type ....................... 48

Table 14. PHEV Charging Summary Statistics................................................................................ 58

Table 15. Double-PEV (1 BEV and 1 PHEV) With or Without an ICEV (N=11) ............................... 70

Table 16. (Average) Annualized Estimates of VMT and Energy Consumption in BEV HHs ........... 71

Table 17. (Average) Annualized Estimates of VMT and Energy Consumption in PHEV HHs ......... 77

Table 18. Annualized Estimates of PHEV VMT, HH VMT, and HH UF ............................................ 79

Table 19. Annualized Estimates of Charging Sessions and kWh Charged in PHEV HH ................. 80

Table 20. Annualized Estimates of Charging Sessions by PHEV Type in PHEV HHs ...................... 81

Table 21. Average Utility Factor (UF) of PHEVs by Model Year (MY) ............................................ 83

Table 22. Double-PEV (1 BEV and 1 PHEV) HHs With or Without ICEV(s) (N=9) .......................... 85

Table 23. Annualized Driving Metrics in BEV/PHEV HHs ............................................................... 88

Table 24. Annualized Charging Metrics in BEV/PHEV HHs ............................................................ 89

Table 25. FCV Driving Data Overview .......................................................................................... 116

Table 26. FCV Refueling Data Overview ...................................................................................... 117

List of Tables x

Table 27. Mirai Driving Trip Level Summaries (on days when the FCV was driven) ................... 118

Table 28. Refueling Summary...................................................................................................... 123

Table 29. (Average) Annualized Estimates of VMT and Energy Consumption on FCV HHs ........ 129

Table 30. Summary of Interviewee Information ......................................................................... 134

Abstract

Results from this report highlight how alternative fuel vehicles are used based on data collected

between 2015 and 2020. Alternative fuel vehicles include plug-in electric vehicles (PEVs), vehicles that

are either battery electric vehicles (BEVs) or plug-in hybrid electric vehicles (PHEVs), and fuel cell

vehicles (FCVs). This category of vehicle technologies is included in the California Air Resources Board’s

Zero Emission Vehicle regulations and is referred to as ZEV in this report. We explore the environmental

impacts of driving, charging behavior and infrastructure. In households with ZEVs, the data from surveys,

loggers, and interviews indicate that those vehicles are being used extensively. This report, which

combined the data collected in two consecutive studies between 2015-2020, includes first and second

generation PEVs popular in California between 2011-2018. The BEVs include the first-generation, short-

range Nissan Leaf and the long range BEVs such as the Chevrolet Bolt and Tesla Model S. The PHEVs

include short range sedans such as the Toyota Prius Plug-in and longer-range vehicles such as the Toyota

Prius Prime, Chevrolet Volt and Chrysler Pacifica. The FCVs include the most popular fuel cell vehicle, the

Toyota Mirai.

In replacing household gasoline miles with electric miles, the results of this study suggest a significant

difference of this driving behavior according to vehicle range and battery size. While we cannot say that

this driving behavior is directly influenced by vehicle characteristics such as range, size, and

performance, we can, however, observe the trends in driving behavior of participants who own the

same vehicles. For example, it is important to note that longer-range PHEV users in this study had the

tendency to plug in their vehicle more and achieve higher electric vehicle miles traveled (eVMT). It is

also plausible that similarities in driving behavior between users who own similar vehicles are

coincidental, since infrastructure availability and other variables aside from vehicle characteristics could

be the main variables in vehicle performance.

Overall, the results suggest, as expected, that longer-range PEVs have more electrified miles than those

of shorter range PEVs. However, to maximize the impact of PEVs, a full set of policies is needed to

address charging behavior and vehicle purchase. FCVs in our sample are being used for long commutes,

but not on long road trips, based on local refueling infrastructure deployment. The results in this report

point to factors that affect the environmental impact of ZEVs, including charging behavior, household

fleet composition, vehicle usage, and more. Further research is necessary to shape policies that lead to

more sustainable transportation and ZEV usage. The household analysis suggests that longer-range ZEVs

can reduce the environmental impact of transportation, however future households may move to two

ZEVs; combining BEVs or FCVs with PHEVs, or short-range BEVs with long-range BEVs, which would

significantly increase the electrification of miles at the household level. The report’s main limitation is

the sample size of logged households.

Preface

The purpose of this report is to understand, under real world conditions, the emission potential of zero

emission vehicles (ZEVs), as defined by CARB regulation, to highlight benefits and challenges, and to

present needs for improving and regulating future vehicles. This report includes data from three distinct

zero emission vehicle types; plug-in electric vehicles (PEVs), which includes both battery electric vehicles

(BEVs) and plug-in hybrid electric vehicles (PHEVs), and fuel cell vehicles (FCVs). This report covers data

collected from the original Advanced Plug-in Electric Vehicle Travel and Charging Behavior report (CARB

Acknowledgements

xii

Contract 12-319

) and updates it with data collected from additional models between 2018-2020 and a

new section on fuel cell vehicles (CARB Contract 16RD009). We also updated the cold-start section that

was part of the original report. This Emerging Technology Zero Emission Vehicle Household Travel and

Refueling Behavior report allows us to monitor how new PEVs are being used on a day-to-day and

month-to-month basis within a household travel context, by placing data monitoring devices (loggers) in

all vehicles in participant households for a period of one year. The combined projects provide a common

basis to evaluate technologies comparatively and in a consistent way. The data was collected from all

the vehicles in the households, including ICEVs and the ones in the original report, with a focus on the

larger group of two vehicle households. The report includes the five years of data collection from

different vehicles that have no standardized protocol for data reporting. Over these five years, loggers

were installed on about 800 vehicles, including ZEVs and ICEVs. The result is the collection of 7 million

miles of data, including 4.3 million miles that were collected from alternative fuel vehicles.

The main additions to this report are: (1) updating the sample size of the original report to get a better

representation of vehicle usage in California, (2) updating the household level analysis by adding a larger

sample of two and three vehicle households, (3) adding battery electric vehicles with longer ranges,

lower priced BEVs, and larger platform PHEVs such as the Chrysler Pacifica minivan, (4) adding fuel cell

vehicle analysis, (5) updating the engine cold-start analysis for plug-in hybrids, and (6) conducting a new

set of interviews.

Acknowledgements

The authors would like to thank the California Air Resoures Board and California Energy Commission for

their funding of these aformentioned projects. The California Air Resources Board staff were essential

and thoughtful collaborators in executing these complex projects. The first phase of this study was made

possible with the additional prelimary support of the utilities in California for pre-testing, and the data

logger supplier FleetCarma. In addition to the authors, there were many staff and students who

supported and contributed to this project in many ways, and without which this project would not have

succeeded. Thank you to Eric Racadag, Nicole Tsugawa, Andrew Ferguson, Corey Hobbs, Tommy Garcia,

Marcelo Steinkemper, Alex Tang, Nicholas Magnasco, Valerie Silva, Alexa Monserret, Noh Kahsay,

Melissa Ng, Gene Uehara, Devin Ciriaco, Nathaniel Kong, Jonathan Gordon, Thomas Bradas, Ethan Khoe,

Scott Begneski, Courtney Carroux, Mae Moninghoff, Mounika Bodagala, Daniel Reif, and Mike Nicholas.

The statements and conclusions presented are those of the contractor and not necessarily those of the

California Air Resources Board or the California Energy Commission. The mention of commercial

products, their source, or their use in connection with material reported herein is not to be construed as

actual or implied endorsement of such products

https://ww2.arb.ca.gov/sites/default/files/2020-06/12-319.pdf

Executive Summary

This Emerging Technology Zero Emission Vehicle Household Travel and Refueling Behavior Report

monitors how new plug-in electric vehicles (PEVs) and fuel cell vehicles (FCVs) are being used on a day-

to-day and month-to-month basis within a household travel context. This was done by surveying owners

and placing data monitoring devices in all vehicles in participant households for approximately one year,

including conventional or non-plug-in hybrid internal combustion engine vehicles (ICEVs). Over five years

of data collection, loggers were installed on about 800 vehicles, ZEVs and ICEVs alike, that had no

standardized protocol for on-board data reporting. The result is the collection of 7 million miles of data,

including 4.3 million miles that were collected from alternative fuel vehicles. The first project began with

studying several models of plug-in vehicles. PHEVs included the following: (1) the Toyota Prius Plug-in

(Model Years [MY] 2012–2016), (2) the first-generation Chevrolet Volt (MY 2010–2015), (3) the Ford C-

Max Energi PHEV (MY 2014-2016), (4) the Ford Fusion Energi PHEV (MY 2014-2016), and (5) the second

generation Volt (MY 2016). The BEVs included the following: (1) the first generation Nissan Leaf (MY

2010–2016), (2) the second generation Leaf with a 30kWh pack (MY 2014–2016), (3) the Tesla Model S

with battery size of 60-100kwh (MY 2013-2017), and (4) the Toyota RAV4 EV with battery size of

41.8kwh (MY 2012-2014). Over time, the project expanded to include new additions to the market,

including the Prius Prime-8.8, the Pacifica-16, the Bolt-66, and the Toyota Mirai FCV.

The data collected shows that longer-range BEVs and PHEVs, vehicles with larger batteries, had a greater

substitution of gasoline miles with electric miles. The same results are true for both vehicle level and

household level analyses. While exploring vehicle usage, we learn that short range BEVs were not used

for long road trips or long freeway speed trips. As battery price drops and driver preferences and needs

are clearer, the future supply is not anticipated to include BEVs with a range lower than 100 miles. The

analysis found significant differences between the use of the longer-range BEVs and that of shorter-

range vehicles. In addition, there were notable differences between long-range BEVs, the Bolt-60 and

the Tesla Model S, in our study. The Bolt-60, for example, recorded 6.8% of miles with speeds higher

than 75 mph whereas the Tesla sample population logged more than 12% of their miles as high-speed

miles. Tesla drivers also used DC fast charging further away from home than other BEVs, which

illustrates another difference in usage between long-range BEVs.

Our survey shows that there is a greater share of alternative vehicles used for commuting in comparison

to the California fleet. On average, the BEVs in our study charge less than once a day, including days

when the vehicle was not used as expected. DC fast charging is still used mostly around home, within a

radius of less than half the vehicle range from their home location. Only Tesla vehicles use DC fast

charging for longer trips in a substantial way. Level 1 charging was also significant for all vehicles’ long,

overnight trips. All the FCVs in our sample were used for commuting, similar to short range BEVs, and

other small vehicles in our sample and were not used for long road trips. In households with two

vehicles, the FCV accounted for 50%-70% of the total household annual miles.

While most modeling and early assumptions hypothesized that electric vehicle drivers would plug-in

every night and start each day with a full battery, our results show that charging every other night is

more common for longer range BEVs or when driving less, while charging more than once a day is

common for PHEVs who drive more than their electric mileage range. The interviews that were

conducted show the importance of charging policy and charging management in the workplace, which

can inform the optimization of charging infrastructure. The interviews explore what could cause a low

frequency of workplace charging such as charger congestion, convenience, and dependability.

Executive Summary

We had a small sample of Toyota Mirai FCVs in our study that were used for an average of 10,700 miles

per year including long commute days, but a small number of long road trips. In households with two

vehicles, the FCVs account for 50%-70% of the VMT.

For PHEVs, our study presents lower utility factor (share of electric miles driven over total miles driven)

values than those of the EPA (Environmental Protection Agency) results, primarily from driving more

than the expected mileage used for estimating the utility factor, but also from driving at higher speeds

than simulated. We also learned that PHEV drivers with larger batteries charge their vehicles more when

needed, achieve higher utility factors, and have many days with no engine starts at all, compared to

PHEV drivers with smaller batteries.

Overall, the results indicate that longer-range PEVs have more electrified miles than their shorter-range

counterparts, resulting in a reduced greenhouse gas (GHG) footprint. The results of this study address

possible factors that affect the environmental impact of ZEVs. This study focuses on ZEV performance

and household performance but did not collect data to compare those households to the general

population and ICEV-only households. This nature of the data collected for this study introduces two key

limitations: first, no data was collected on ICEV-only households, which limits our ability to extrapolate

from these results to the general population or new-car-buying households; second, the data collection

for this study spans 5 years and may include behavioral changes in society as a whole or among ZEV

drivers. To maximize the benefits of PEVs, a full set of policies is needed to address charging behaviors

and vehicle purchases. As those factors continue to change over time, ongoing research is necessary to

better shape the policies that lead to more sustainable transportation and efficient ZEV usage.

Introduction 3

1 Introduction

Road transportation accounted for 21% of global energy consumption (Contestabile, Alajaji et al. 2017)

and it will increase unless the share of carbon intensive transportation fuels is substituted by cleaner

sources. Plug-in electric vehicles (PEVs), which include full battery electric vehicles (BEVs) as well as plug-

in hybrid electric vehicles (PHEVs), and hydrogen fuel cell vehicles (FCVs) are promising alternatives to

conventional internal combustion engine vehicles (ICEVs) because of their energy conversion efficiency

and reduced tail-pipe emissions. Over 1.5 million electric vehicles have been sold in the United States,

and over half of those were sold in California. Nevertheless, gasoline-powered vehicles and other

conventional-fuel vehicles still constitute 98% of the light duty vehicle (LDV) fleet (Figure 1). For this

report, conventional vehicle powertrains such as gasoline, gasoline-hybrid (HEV), flex-fuel, diesel, and

diesel hybrid vehicles will all be included under the umbrella term of ICEVs, since they are not the focus

of this report.

Figure 1. California LDV Fleet Composition (2018) by Fuel Type (CA Department of Motor Vehicles,

published 2019)

The market share of BEVs and PHEVs has been increasing over the past decade. The share of hydrogen

fuel cell vehicles (FCVs) has been considerably lower than the share of BEVs and PHEVs as illustrated in

Figure 1. According to the California DMV and data reported by the California New Car Dealers

Association, the share of PEVs in total new vehicle sales/registration went up from 3% in 2014 to 8% in

2019 (Figure 2). In addition, the share of PEVs in the total LDV stock of California increased from 0.4% in

2014 to 1.43% in 2018.

Introduction 4

Figure 2. Share of BEVs and PHEVs in New Vehicle Registration (Source: California New Car Dealers

Association)

The deployment of vehicles so far is not evenly distributed; areas with higher income population and

more total vehicles have a higher share of electric vehicles. (Figure 3)

Introduction

Figure 3. PEVs as share of all vehicles

When it comes to BEVs, a large share of the rebates in the past couple of years have gone to Tesla

buyers while the share of Nissan Leaf rebates has dropped among first-time BEV adopters. In the case of

PHEVs, adopters of the Chevrolet Volt, Toyota Prius Prime, and the PHEVs offered by Ford, like the

Fusion and the C-Max Energi, have claimed the majority of CVRP (California Vehicle Rebate Project)

rebates. These trends in CVRP application data based on vehicle make reveal the composition of the PEV

fleet in California in terms of vehicle characteristics such as electric range and battery technology.

PEV usage, including charging and driving behavior have a major impact on the energy and

environmental benefits of these vehicles. There are many variables that discourage the adoption of

PEVs, including limited electric driving range, household access to charging locations with various

capabilities, costs, and charger access rights as well as behavioral variables, such as the habits and

desires of households for using these new types of vehicles. PHEVs may charge more or less frequently,

resulting in a higher or lower percentage of electric powered VMT (eVMT)—where the proper

denominator for calculating the percentage is the household’s total VMT, not merely the total VMT of

the PHEV. Such complexities can complicate attempts to predict and calculate the impact of recent

technologies on emissions in the coming decades. This study identifies and begins to measure these new

patterns.

Consumer’s perceptions on PEVs’ ability in meeting daily mobility needs compared to ICEVs, higher

upfront capital cost compared to ICEVs, range anxiety, and reliable access to charging infrastructure

continue to be major barriers to large-scale PEV adoption (Dimitropoulos, Rietveld et al. 2013, Liao,

Molin et al. 2017, Lutsey, Meszler et al. 2017, Hardman, Jenn et al. 2018). These barriers create

Introduction

uncertainties in the evolution of the PEV market. Heterogeneities in daily driving patterns and needs

across various sociodemographic indicators and geographical locations, further compound these

uncertainties. Since PEVs are uniquely positioned to interact with the energy and the transportation

sector, uncertainties in the evolution of the PEV market pose many problems for policy makers, auto

manufacturers, electric utility companies, and charging infrastructure developers (Wietschel, Plötz et al.

2013). Policy makers have to continually fine tune existing incentives (financial and/or non-financial) or

introduce new incentives to encourage the adoption of PEVs. Understanding daily driving needs is

crucial for auto manufactures for optimal PEV design and model choice offerings. Charging

infrastructure developers have to ensure that electric vehicle supply equipment (EVSE) is efficiently

located and managed to alleviate concerns about range anxiety and accessibility to EVSE. Utility

companies are particularly concerned about PEV charging patterns as it has the potential to create

localized hot spots when not managed properly, necessitating network upgrade or expansion (Muratori

2018). Utility companies also would have to design their PEV specific rates keeping in mind when and

where PEVs are charged.

The decision to own a PEV or FCV will have long-term consequences on the user from a total cost of

ownership (TCO), value proposition, and life-time GHG reduction potential perspective. The user’s daily

driving and charging behavior will have short-term impacts on planning charging infrastructure roll out

and effectively managing the incremental demand imposed by PEV charging. To better understand the

impacts of PEVs across varying timescales given their negligible global share and the scarcity of their

usage data compared to ICEVs, studies have relied on existing data to model their behavior. Modeling

PEV driving behavior will offer qualitative and quantitative insights into the feasibility of PEVs in

replacing a conventional vehicle. The daily and long-term energy, emissions, and economics of

PEVs/FCVs are related to the extent in which prospective and current PEV/FCV owners perceive the daily

driving utility of these vehicles when compared to a conventional vehicle. The charging demand

imposed specifically by PEVs is affected by their daily driving distance and dependent on their trip start

times, end times, and dwelling times.

Given the relative scarcity of actual ZEV usage data, researchers and policymakers create scenarios by

combining various sources of travel data and superimposing a set of preconceived expectations about

ZEV driving and charging/refueling needs. There has been an increase in efforts to analyze data from the

real-world operation of PEVs to estimate eVMT, since it is the most widely adopted metric to determine

the potential of electricity as a transportation fuel. The scope of such efforts has expanded recently to

estimate the zero emission VMT, or zVMT, which is the miles traveled on electricity only. For BEVs and

FCVs, VMT, eVMT, and zVMT are the same. However, for the PHEVs, due to their dual modes of

operation, zVMT is lower than eVMT. Information about ZEV usage based either on assumptions or from

real-world operations have direct consequences on not only their VMT, eVMT, zVMT energy

consumption (electricity, hydrogen, and gasoline) and emissions (from driving and charging), but also on

specific policies that rely on them, such as credit allocation under the ZEV mandate (CARB 2017) and PEV

infrastructure projections and investments (Wood, Rames et al. 2018, Brecht and Orenberg 2019).

There are many variables that are presumed to interfere with the assumption of a ZEV for a previous

conventionally fueled vehicle, including limited electric driving range, household access to

charging/refueling locations with various capabilities, costs, and charger access rights as well as

behavioral variables such as the habits and desires of households for using these new types of vehicles.

Depending on travel needs, desires, fuel costs, charging opportunities, and how much drivers like or

dislike their ZEV, they may end up using their new ZEV for more or less vehicle miles traveled (VMT) than

they had for a previous vehicle. With PHEVs, they may charge more or less frequently, resulting in a

higher or lower percentage of electric powered VMT (eVMT)—where the proper denominator for

Introduction

calculating the percentage is the household’s total VMT, not merely the total VMT of the PHEV. This

report identifies and begins to measure these new patterns.

FCVs have even lower adoption rates than PEVs despite having high driving ranges, with the Toyota

Mirai surpassing 300 miles of range and refueling times equivalent to that of ICEVs. The main barriers

impeding the adoption of FCVs include scarcity of hydrogen infrastructure, relatively high purchase

price, inability to refuel at home and safety concerns (Hardman, Shiu, Steinberger-Wilckens, &

Turrentine, 2017). In California, there are 41 active hydrogen fueling stations that are mostly

concentrated in and around dense metropolitan areas such as Los Angeles and San Francisco. The sparse

and selective spatial distribution of refueling stations can serve to discourage potential adopters from

purchasing FCVs due to refueling inconvenience. The high purchase price owed to the high cost of fuel

cells and hydrogen tanks coupled with the lack of economies of scale also drives potential buyers away

from FCVs (Hardman et al., 2017). The strategic, concurrent adoption of FCVs and PEVs can reduce a

significant proportion of transportation related carbon emissions. Despite FCVs having lower well to

wheel efficiencies than many PEVs, their usage can still have positive environmental benefits from an

emissions perspective if they are specifically adopted in regions where the electricity grid is largely coal

based (Hardman et al., 2017). As it is for PEVs, there is little empirical research on FCV usage. This study

analyzes vehicle usage data collected from a small number of households with Toyota Mirai to uncover

usage patterns that can potentially be indicative of FCV impacts on carbon emissions in the coming

decades.

Travel behavior researchers have known that the household is the critical unit to study because

activities are often allocated among a fleet of household vehicles on a trip-by-trip basis. Previous studies

of household vehicle travel have been for short periods or have not used data loggers. However, this

project studied the use of vehicles by the household, instrumenting all their vehicles with GPS enabled

logging devices, to accurately measure the trip allocation and activity route formation of the whole

household across a year.

This research is designed to investigate these household travel patterns and lifestyle activity spaces in

response to PEVs/FCVs across a large set of households.

The overarching objective of this research project is to collect and analyze longitudinal, spatial, and in-

use vehicle data, including eVMT, from a variety of plug-in electric vehicles (PEV models). PEVs are

imperative in achieving California’s long-term air quality and climate stabilization goals. This means

measuring the travel and fueling of all vehicles within a PEV-owning household is important. Usage and

charging habits of PEV owners remain ambiguous due to the diversity of PEV designs, technologies,

electric ranges, and the failure to account for other travel within households. However, these behaviors

will have significant implications for statewide emissions, energy consumption, and electrical grid

management based on the miles these vehicles travel using off-board electricity sources. The secondary

objective of this project is to collect and analyze in-use vehicle data from FCVs, specifically the Toyota

Mirais. FCVs can also serve to further California’s air quality goals, especially when used in regions with

high-carbon electricity grids. Objectives include:

1. Determining the share of PEV miles traveled, powered exclusively by off-board electricity (eVMT), and

how emissions profiles might differ between the several types of PEVs.

2. Learning the allocation patterns between household vehicles for daily, weekly, seasonal, and

infrequent trips. Knowing these reasons will assist CARB and others in creating policies to increase eVMT

in the future and better estimate current eVMT.

Recruitment and Background Survey

3. Learning recharging patterns of PEVs in a household context. These patterns can assist CARB and

other State partners in developing the charging network in ways that might help households maximize

their eVMT. Additionally, knowing the locations and times of charging events could help CARB and

partners to assess the time-of-day emissions impacts, and perhaps influence the recharging of PEVs in a

way to reduce emissions and optimize the use of the grid across time and seasons. The same data can

also assist utilities and their regulators to understand grid impacts from PEV charging, rate impacts on

charging behavior, and the need for public infrastructure. Temporal and spatial data would provide a

better picture of when and where PEVs are charging, which informs upstream emissions estimates.

4. Understanding how any measure of eVMT develops within the overall travel of households because of

systematic variation caused by, for example, household self-selection of different types of PEVs.

Furthermore, while an individual PEV may have a high share of eVMT, total transportation-related

emissions from the household, it will also depend on the activity and usage of all other vehicles in the

household fleet, as well as other modes of transportation used, such as public transportation.

5. Characterizing the engine start activity profiles of blended PHEVs. In the 2017 market, many PHEVs

are “blended” in the sense that an internal combustion engine (ICE) can start to help power the vehicle

before the battery is depleted. These ICE starts occur when the electric drivetrain is not sufficient to

meet immediate high torque demand, regardless of the battery state of charge. These ICE starts occur

under high power demand scenarios and are distinct from cold starts for conventional vehicles, which

typically occur with the vehicle stopped, in park/neutral, and with a low immediate torque demand.

PHEVs likely have a different distribution of engine-on events compared to conventional vehicles and

these can occur due to battery depletion as well as high-torque demand events. The results of this study

can be used to improve the emission inventory model (EMFAC) in estimating PHEV start emissions. The

results can also be used to guide the development of future clean car standards.

6. Studying the driving and refueling behavior of FCV vehicles. This includes understanding the allocation

patterns between household vehicles for daily, weekly, seasonal, and infrequent trips. It also involves

determining the location of and duration between refueling events. The metrics from these analyses can

help CARB and other state partners determine optimal locations for hydrogen fueling stations in order

to promote the adoption of these environmentally clean vehicles.

2 Recruitment and Background Survey

This report seeks to collect the data that can answer essential questions about future travel and

charging behavior and refueling of ZEV owners in California households and the benefits that are likely

to result. What are the environmental benefits of these vehicles? How much travel can and will be

shifted to PEVs, and specifically to BEVs and to PHEVs, per vehicle and for the household fleet? What

kind of charging network is needed?

The funds for this project cover collection, cleaning, and basic analysis of the data, but not the analysis

aimed at understanding the interaction between the data factors collected and potential correlations.

This study uses data from three main sources: 1) survey data of PEV households, 2) vehicle-level data

collected from 303 households through loggers connected to the vehicle telematic system, and 3)

interviews of 40 PEV users that participated in the logging component. This research helps identify ways

to facilitate increased use of zero-emission vehicles (ZEVs) by Californians. Also, longitudinal, temporal,

and spatial data provide a picture of when and where PEVs are charging, as well as the electric- and

gasoline-vehicle miles traveled by PEVs and other vehicles in the household.

A detailed, 30-minute recruitment survey of ZEV owners/lessees (hereafter referred to as owners for

simplicity) was conducted to determine how many participants would be needed for each region and

Recruitment and Background Survey

sociodemographic group so that the results would be representative of statewide ZEV owning

households—i.e., so the results could be generalized to the wider population. The survey included eight

categories of questions: travel behavior, driving behavior, vehicle performance (MPG), vehicle

characteristics, response to ZEV related incentives, vehicle purchase history, current household vehicle

fleet, sociodemographic characteristics, PEV charging behavior and FCV refueling behavior. The survey

targeted owners of all ZEV models in the market at the time of the survey. The initial survey also was the

first step in recruitment, asking whether respondents would be willing to participate in the second part

of the study by having a logger installed in their vehicle. The information also helped determine whether

household vehicles were suitable for participation based on logger limitations and vehicle usage

(appropriate mileage, accessible OBD port, household with vehicles newer than 1996). In addition, the

surveys allowed us to capture information about the households such as commute location, charger

access, sensitivity to price, demographics, etc. We invited participants to take the internet-based survey

in three different methods. First, CARB sent email invitations to ZEV owners who had applied for the

California Vehicle Rebate Project (CVRP); second, CARB sent postcards to a random selection of persons

who had a ZEV registered based on the DMV records but did not apply for CVRP; and third, CARB sent

postcards to a random selection of owners of used PEVs based on DMV records. The result of the

surveys and the total response rate were estimated, based on the DMV records, for the total number of

light duty vehicles and plug-in vehicles by region and year, based on model year for vehicles model

2010-2017. Updated VMT records allow for re-evaluation of the survey data and logger data

representation of the state light duty fleet, comprised of light duty vehicles, plug in vehicles, and new

and used vehicles. The DMV data allow us to estimate the total number of ZEV on the road each year, in

addition to the information about new vehicles sold based on the CVRP application.

21,000 new PEV owners and lessees started our survey between May 2015 and August 2018 in addition

to 680 used PEV owners. Of those surveyed, 12,396 PEV households and 470 FCV households had

enough information, answered all parts of the survey, and indicated that we could contact them for the

logging phase, but this number included surveys with missing information for some survey part based on

our skip logic or households that owned a vehicle that was incompatible with the loggers. The overall

response rate to the surveys was 18%, and 82% of these respondents completed the survey. However,

this 82% included persons who were not eligible for the logging study because they utilized their PEV for

business purposes, no longer owned a PEV, and similar cases.

Logger Installation Process

The combined project design called for a straightforward process. After identifying potential households

for the logging part of the study, we emailed those households to reaffirm their interest, that they still

had the PEV, and that they planned to have it for the next 12 months. Of the households we invited, 15–

25% agreed to participate and moved to the next phase. The overall rate of recruitment was 1 logger

installation for every 300 households that received the initial survey.

The project was budgeted to allow two visits to each household, one to install loggers on all household

vehicles and another to remove them. The initial plan also called for the project team to make one trip

per region to do all installations in that region and a second trip to do all the removals. The regions

included areas from San Diego in the south to Crescent City in the north, and the project team was

based in Davis. The installation-removal team included the project researchers, a full-time project

manager, part time staff member, and 16 undergraduate students.

The loggers for this project were obtained from FleetCarma, a vendor selected by a bid conducted by UC

Davis. FleetCarma’s OBD loggers (C2 and C5) are configured to collect real-time data with automatic

data upload via the cellular network and contain internal GPS, making them the ideal data loggers for

Recruitment and Background Survey 10

capturing vehicle data. Each logger had to be programed to a specific ZEV model, a process that was

done manually at the beginning of the project and through the logger’s internet connection later. The

data collected by the loggers was analyzed and then sent by cellular connection to the vendor servers

and from there to UC Davis servers.

By the end of the project we had to make many more trips to each region than what was originally

planned for and budgeted. The main reason for this difference between the planned and actual

execution was the difficulty of scheduling installations during weekdays, when people had their vehicle

or vehicles away from home (e.g., at work). As a result, evenings and weekends were often the only

times when we could install, and later remove, loggers in all the household vehicles at once. Other than

these limitations on workable time windows, we underestimated the number of additional visits that

would be necessary beyond the initial installation and final removal of the loggers. Over the project

period we had to replace more than 30 faulty loggers or data cables, we had to remove loggers from

vehicles owned by households who chose to leave the study, sold the vehicle, had an accident, moved

out of the state, etc. In many cases, we recruited an additional household to maintain the total sample

size. We had to make approximately 25 trips to Los Angeles, 20 to the San Diego area, 200 to the Bay

Area, 50 to the Sacramento area, and 25 to the regions north and east of Davis.

The participation incentive was $350 split between the installation ($150) and completion of the data

collection and return of the data logger ($200). Overall, we had to pay incentives to about 80% more

households than the number used in this final report and visit each household 4–6 times instead of the 2

times planned. We recruited 424 Households and installed loggers at 402 vehicles, including 14 FCVs,

388 PEVs, and about 300 ICEVs. We dropped vehicles with data collected for less than 90 days, all with a

sample of less than 4 vehicles of the same model.

Figure 4 presents the number of installations along the study timeline, including information on the

model of ZEVs that had loggers installed. To reduce the project cost, we reinstalled the loggers from

phase 1.0 in phase 2.0 vehicles, those from phase 1.5 in phase 2.5 vehicles, and those from phase 3 in

phase 4 vehicles. Therefore, this final project report includes data collected between June 2015 and May

2020 for some part of the analysis. As Of July 2020, we continue to collect data from about 30

households as we are unable to uninstall the loggers due to the stay at home order resulting from the

COVID-19 global pandemic. We hope that this data will be used in a future project to study travel

behavior of electric vehicle owners in the time of Coronavirus.

Recruitment and Background Survey

Figure 4. Overview of Number of Logger Installations During Each Phase of the Project, Classified by

Number and Type (New or Used) of Vehicles per Household. MUD= multi-unit dwelling

We planned the recruitment to cover the main vehicle models at the time of each phase and to cover

the shift from buyers of new PEVs to buyers of used PEVs and households with two PEVs. We also

covered all main electric utilities in California. However, due to the extended period of data collection

2015–2020, the relatively small sample, and logger problems that forced us to drop some of the

samples, we experienced a limitation in having statistically significant results in every category needed

to fully represent the changing ZEV owner population.

Recruitment and Background Survey

Figure 5. Home and Daytime Charging Locations 2015-2020

Data Collection and Limitations

As for all social science studies, this work suffers from self-selection bias resulting from the recruitment

being based on survey takers who volunteered for the logging phase. We controlled for basic social

demographic characteristics, but we could not control for any correlation between the probability to

take a survey and volunteer for the study and specific driving and charging behavior. Another important

bias in the household selection and the results presented is the fact that no participants were chosen

who did not plug in their PEV on a regular basis. On the other hand, the self-selection of vehicle model

by the households, for example users with lower daily miles select short range vehicles, is not a bias but

a representation of the vehicle owners’ population. To explore the potential bias of our selection please

refer to a paper we published based on the servey response (Chakraborty, et. al 2020). For both the

PEVs and the ICEs in the study, not all logger parameters were available on all vehicle models and the

parameters collected changed over time with changes made by the logger vendor to the dataset design,

the logger hardware, and the vehicle software. In general, the relevant set of parameters collected

varied between the ICEs, the BEVs and the PHEVs due to differences in fuel sources. Other than speed

and GPS metrics which were collected for all vehicles, the ICE data includes vehicle parameters such as

fuel rate, mass air flow, etc. to calculate fuel consumption, the BEV data includes attributes such as

battery voltage and power to estimate electrical enery consumption and the PHEV data contains both

fuel and electrical energy determining metrics.

Recruitment and Background Survey

The data transferred from the logger includes raw data from the vehicles and calculated data based on

algorithms programmed in the loggers. Some parameters, such as miles per gallon (MPG), are derived

from multiple parameters such as revolutions per minute (RPM), engine load, mass air flow, and intake

air temperature. Other parameters, such as distance, were derived from speed and time. Most

parameters were collected every second but others, such as GPS and State of Charge (SOC), were

collected every 10 seconds. Battery capacity was derived for each vehicle individually based on the

average value, over the data collection period of SOC change and energy used. We did not use the

manufacturer stated capacity or EPA range in our analysis.

One of the most important limitations of the data is that if one of the parameters being recorded

changed, a new row would be generated in the dataset/spreadsheet and values for all the parameters

would be populated in that row. However, because different parameters were recorded at different

rates, a parameter that had the same value between adjacent rows may have been updated and had

truly stayed the same over two collection times, or it may not yet have been updated and the program

had populated the cell with the last recorded value from the previous row. In summary, although we

attempted to resolve limitations within our study, due to the unreliability of certain data, it was not

possible to distinguish whether an unchanged parameter was copied from the last collection time or

recollected but had the same value.

Another limitation in the data collection was that data from ICEVs. First, our ICEVs sample includes only

vehicles in households with one PEV or FCV that are, in most cases, newer than the ICEV and therefore

are not a representative sample for ICEVs in California. Second, ICEVs within a given household that

were estimated to be driven less than 1000 miles per year did not have loggers installed. Thus, logger

data was not collected from these vehicles. However, the VMT on these ICEVs was recorded manually

from odometer readings with only one vehicle exceeding 1000 miles.

We developed four different methods to estimate energy consumption from PHEVs (and ICEVs) based

on the data reported for each vehicle. Overall, we identified a wide variation of results based on trip

distance, speed, etc.

Data Annualization

While collecting vehicle data, the vehicle loggers sometimes malfunctioned due to internal software

issues or external mishandling, leading to gaps in the time series data we received. If we analyzed the

data without accounting for these gaps, we would be underestimating the annualized values of key

vehicle metrics. Therefore, we developed a process to annualize vehicle metrics, while minimizing the

miscalculation associated with this incomplete data issue to the best of our ability. For annualizing

vehicle-level metrics for a given car, we first condensed the vehicle’s events into days – this includes all

days between the vehicle’s first logging event to its last logging event. Next, we identified major gaps

(more than 7 days) within the data collected by locating the dates and times when the vehicle logger

didn’t record any data. We then contacted the driver(s) of the car to gauge If the car was indeed not

used during those identified gap days or if the logger simply malfunctioned. We marked the vehicle’s

days for which we knew the logger malfunctioned as invalid, given those days do not represent real

travel behavior. Finally, we scaled the vehicle’s totalized metrics to 365 days based on the number of

valid logged days to attain annualized estimates of those metrics.

Similarly, for annualizing the metrics of a given household, we first condensed the events from each

vehicle within that household into days (separately). Next, we located gaps within the data collected for

each vehicle in the household. We then contacted the driver(s) of the vehicles to gauge If the vehicles

were not being used during their identified gap days or if their loggers were simply malfunctioning. We

Recruitment and Background Survey 14

marked all the vehicles’ days for which we knew at least one vehicle’s logger malfunctioned as invalid,

since those days do not accurately capture real household travel behavior. Lastly, we merged the days

data of all the vehicles and scaled the household’s totalized metrics to 365 days based on the number of

valid households logged days to attain annualized estimates of those metrics. In many cases, not all the

vehicles within a household began or ended logging data at the same time; to handle data gaps

introduced by mismatched start/end logging dates, we decided that the start date for the household

days data is the start date of the vehicle in the household that started logging the latest, while the end

date for the days data is the end date of the vehicle in the household that ended logging first. We did

not directly account for weekday/weekend differences or seasonal variation while annualizing both

vehicle and household metrics as we log data from all four seasons, for around 365 days for most of the

household vehicles in our sample.

Sampling of the Logged Participant Households

The distribution of households was selected by electric utility and generally follows the market for

electric vehicles with most participants being in one of the four largest metropolitan regions in

California: San Francisco, Sacramento, Los Angeles, and San Diego. Some participants were in

exceptional locations, such as in the mountains or along the coast, where isolation or temperature may

have had an impact on how they used their vehicles compared to those in major metropolitan regions.

Although the sample size is small in those cases, interacting with them and observing their behavior

presents the possibility for additional learning from the project. All the results presented in the report

are based on the relevant sample and are not weighted, as we focused on the impact of different

technology types and did not estimate total impact.

This survey’s participants—PEV households who purchased or leased their vehicle in the last 4 years—

differ from average Californian households. For the general population, less than one-third of

households buy a new car every 3-5 years, according to the 2012 California Household Travel Survey

(CHTS) (CalTrans 2013). To compare PEV buyers to the general population (based on the CHTS 2012), we

combined the income distribution by vehicle type and purchase year.

Considering the market penetration of alternative fuel vehicles, many of the current ZEV owners are

early adopters of the technology. As observed in cases of other technologies, early adopters may have

unique characteristics compared to other new car buyers—age group, education level, and technology

awareness, among others.

Table 1 presents the statistics on sociodemographics and vehicle models among the survey participants.

The sample was stratified by income to represent the income of the larger survey sample. More than

80% of households had an income higher than the median income in California ($67,739 according to

the Census American Community Survey 1-year survey) and the percentage of people with graduate or

professional degrees was 48.7% (California statewide 12.3%). In our dataset, males tended to drive the

PEV more than females in a household, and slightly more BEVs were driven than PHEVs. More than 80%

of respondents owned their houses, and more than 80% lived in detached units. About 50% of

respondents had the Chevrolet Volt, Tesla (Model S-60_80 or Model S-80_100), or the Nissan Leaf, and a

considerable number of the rest used the Prius Plug-in-4.4 or the Bolt-60.

Recruitment and Background Survey

Table 1. Sociodemographics and Vehicle Types Among the Usable Surveyed Participants

Income Age Education

<50K 217 10-19 years old 11 High school 995

50-99K 1,064 20-29 years old 326 College 3,247

100-149K 1,686 30-39 years old 1,767 Post-graduate 4,059

150-199K 1,529 40-49 years old 2,166 Gender

200-249K 1013 50-59 years old 1,932 Male 6,201

250-299K 667 60-69 years old 1,428 Female 2,052

300-350K 368 70-79 years old 554 Decline to state 77

350-399K 206 > 80 years old 74

Household size

400-449K 154 Missing 79 1 person 868

450-499K 105

2 persons 3,236

> 500K 363

3 persons 1,523

4 persons 2,021

5+ 706

Recruitment and Background Survey 16

Number of Vehicles Types of ZEV Model

1 1000 Battery 4,230 500e 160

2 4,303 Plug-in Hybrid 3,749 Bolt-60 748

3 2,049 Fuel cell vehicle 375 C-Max Energi 480

4 694 Purchase or Lease e-Golf 472

5+ 290 Purchased 3,812 Fusion Energi 377

Number of drivers Leased 4,202 i3 590

1 1,101

Housing types

Leaf 1,175

2 2,692 Own houses 6,998 Prius Plug-in 792

3 971 Rent or others 1,337 Tesla 1,384

4 486 Detached housing

Volt 1,442

5+ 86 Detached 6,751 Others 391

Others 1,585 Mirai 334

Table 1 presents the distributions of income, household size, and number of vehicles per household

among the survey population. We tried to select households for logging that would reflect the

geographic distribution and sociodemographic distribution of ZEV households as reflected in the initial

survey.

Figure 6. Distribution of Household Income Among Survey Respondents and Logged Households

Less than

$50,000

$50,000 to

$99,999

$100,000

$149,999

$150,000

$199,999

$200,000

$249,999

$250,000

$299,999

$300,000

$349,999

$350,000

$399,999

$400,000

$449,999

$450,000

$499,999

$500,000

or more

I prefer

not to

answer

Survey Logged

Recruitment and Background Survey

Overall, the logged households are similar to the surveyed households, other than having a minor

oversampling of households with incomes of $50k-$100k and households with two vehicles.

Figure 7. Distrubution of Household Size Among Survey Respondents and Logged Households

The main difference between the logged households and the survey and general populations that is not

reflected in the sampling methods is the exclusion of PHEV users who are not plugging in their vehicles.

Our 2014 research article suggests that short-range PHEVs are more likely to be used as conventional

hybrids (Tal et al. 2014). A more recent study suggests that about a third of the short-range secondary

PHEV owners who finished the survey are using the vehicle as a hybrid only without pluging in.

(Turrentine, Tal, and Rapson 2018)

For the Toyota Mirai we had to recruit three households for every working loggers becouse of loggers

rialbility problems and becouse of the small sample we did not aim for a reprasentative sample.

Table 2. Battery Capacity and Vehicle Types Among the Logged Participants

1 2 3 4 5 or more

Household size

Survey Logged

Make Model and Battery Capacity

Vehicle

Type

Number

Vehicles

Major data collection period

Nissan Leaf-24 kwh BEV 29 2015-2017

Nissan Leaf-30 kwh BEV 28 2016-2018

Toyota RAV4 EV-42 kwh BEV 5 2016-2017

Chevrolet Bolt-60 kwh BEV 27 2017-2020

Tesla Model S-60_80 kwh BEV 23 2016-2019

Tesla Model S-80_100 kwh BEV 25 2016-2019

Recruitment and Background Survey

Make Model and Battery Capacity

Vehicle

Type

Number

Vehicles

Major data collection period

Toyota Prius Plug-in-4.4 kwh PHEV 22 2015-2016

Ford C-Max/Fusion-7.6 kwh PHEV 60 2015-2018

Toyota Prius Prime-8.8 kwh PHEV 27 2017-2020

Chrysler Pacifica-16 kwh PHEV 28 2019-2020

Chevrolet Volt-16 kwh PHEV 44 2015-2017

Chevrolet Volt-18 kwh PHEV 40 2017-2019

Toyota Mirai FCV 11 2018-2020

Total - 369 -

Throughout this report, letters are used in figures to identify significant differences between the average

usage characteristics of different vehicle models among logged vehicles. These significance tests are

performed using the Tukey-Kramer range test for groups of unequal sizes (Tukey, 1977); this method is

commonly used to identify pairwise differences between groups when a one-way ANOVA test confirms

that a significant difference exists in the data. In charts where these letters are used to indicate

significance, each vehicle model will have one or more letters. If two models share any of the same

letters, the test does not identify a significant difference between them for the variable in question; if

not, the test does find a significant difference. For example, consider a chart showing the mean

efficiency for three vehicle models: if model 1 has the lowest mean and is marked with the letter a,

model 2 has the middle mean and is marked with the letters ab, and model 3 has the highest mean and

is market with the letter b, then the pairwise mean comparison finds a significant difference between

models 1 and 3 but not between model 2 and either of the other models.

PHEV eVMT Calculation

Attributing vehicle miles travelled (VMT) to either electricity (eVMT) or gasoline (gVMT) in an ICEV or

BEV or FCV is trivial, all the VMT fall into either one or the other category; however, PHEVs have two

energy sources and correctly tracking the energy can be challenging when both sources are used during

a trip. The following sections describe the methodology used to calculate eVMT for PHEVs.

Need for Energy Efficiency Ratio

One obvious way of calculating the portion of VMT that should be attributed to eVMT would be to

calculate the ratio of total electrical energy consumed to the total energy consumed for both gasoline

and electric, and multiply this ratio by the total VMT. The problem with this approach is that energy

consumption for the two sources does not yield the same number of miles. For example, the 2011 Chevy

Volt has an EPA rated 37 MPG on gasoline and a 93 MPGe when running purely electric. That means that

Recruitment and Background Survey 19

for every kWh of electricity the Volt can travel over 2.5 times as far as with the equivalent energy in

gasoline.

To correct for this, an Energy Efficiency Ratio (EER) needed to be calculated for comparing the electrical

and gasoline usage of energy. Ideally the EER would be calculated for every operating condition of the

vehicle (i.e., every combination of vehicle speed, engine speed, engine torque, motor speed, motor

torque, battery SOC, etc.). However, since this approach is not practical, a single EER was calculated

based upon the vehicle type. The combined fuel economy numbers from fueleconomy.gov was used for

calculating the EER. The EER was calculated by dividing the all-electric fuel economy in MPGe by the

gasoline-only fuel economy in MPG. For example, the 2011 Chevy Volt described previously would have

an EER of 2.5 (93 MPGe / 37MPG). The calculated EER was used to adjust the electrical energy

consumed by the vehicle before calculating the ratio of electrical energy consumed to total (gas and

electric).

Equation 1 shows the calculation of the EER; Equation 2, the calculation of the gasoline equivalent

electrical energy consumption; and Equation 3, the calculation of eVMT.

Equation 1. EER Equation

 =









where MPGe

EPA

is the EPA electric only fuel economy and MPG

EPA

is the EPA combined highway and city

fuel economy for the vehicle using gasoline only.

Equation 2. Electrical Energy Consumption to Gasoline Equivalent





=   



where E

Elec

is the measured electric energy consumption and E

ElecGE

is the gasoline equivalent electrical

energy consumption.

Equation 3. eVMT Calculation

 = 









+



where E

ElecGE

is the value calculated from Equation 2 and E

Gas

is the measured gasoline energy

consumption.

Adjusting for Battery Efficiency

The eVMT calculated using Equation 3 is dependent upon the calculation of E

ElecGE

, which in turn is

dependent upon the measurement (or calculation) of E

Elec

. One may intuitively think that the E

Elec

value

should not be calculated, but directly measured by integrating the power in and out of the battery.

However, this approach would not be correct because batteries are not 100% efficient. Energy is lost

when it is either put into or taken out of the battery. To correct for this, the energy consumed (energy

taken from the battery) and energy produced (energy put into the battery) are maintained separately

and an efficiency factor is applied to the energy produced.

Equation 4 is the equation for calculating the electrical energy consumed. Ideally the battery efficiency

should be determined by testing each individual vehicle, and will vary with temperature, rate of power

Recruitment and Background Survey 20

draw, age of the battery, etc. Since this approach would not be practical, a 90% battery efficiency was

used for all vehicles. The 90% efficiency was based on a linear fit of data analyzed for energy consumed,

energy produced, and delta SOC.

Equation 4. Electrical energy consumption calculation





= 



 



 



where E

BattCon

and E

BattProd

are the energy consumed and produced measured at the battery, and Eff

Batt

the battery efficiency.

eVMT Before Engine On

The eVMT calculation for the equations provided thus far apply a fraction of the VMT to eVMT on a trip

basis. While this approach is valid, further improvements can be made to increase the accuracy of the

calculations by addressing other variables that could influence eVMT. For example, one such variable is

that during a single trip the driving conditions (as well as vehicle efficiency) may vary dramatically and

therefore the use of energy consumption alone may not accurately attribute VMT to gasoline or electric.

It was observed that all the miles traveled prior to the first engine-on event were eVMT, where the miles

travelled after the first engine-on were a blend of gVMT and eVMT. It was this observation that

prompted the change to Equation 3. Equation 5 is the updated eVMT equation (Equation 3) that

attributes 100% of miles traveled to eVMT prior to the first engine-on event, and the fraction of the

miles after to eVMT based upon the fraction of energy.

Equation 5. Updated eVMT calculation

 = 





  













+



where VMT

EngOn

is the VMT at the first engine-on, and E

ElecGEAEO

is the gasoline equivalent electrical

energy consumption after the engine is first turned on.

Adjusting for Kinetic Energy

The initial eVMT equation provided assumed that it was on a trip basis, so the vehicle both starts and

ends at rest. However, the starting point of the hybrid mode may not be at rest, therefore, in the

updated eVMT equation (Equation 5), the E

ElecGEAEO

accounts for the kinetic energy of the vehicle. The

kinetic energy of the vehicle, when it is moving and the engine is on, will carry the vehicle some further

distance. One may wonder if this energy is significant or not. Consider a 2011 Chevy Volt with a curb

weight of 3,781 lbs carrying 200 lbs (passenger and cargo) at 80 mph, the kinetic energy in the vehicle

would be 1.15MJ or 0.32kWh which is equivalent to approximately 3% of the 10.9kWh of usable battery

capacity. This amount of energy would propel the vehicle 0.89 mi according to the EPA all-electric fuel

economy for the Volt (before any adjustments for battery and motor efficiencies). Considering that the

measured median trip distance for Volts was 5.35 mi the 0.89 mi represents approximately 16.6% of the

median trip length. PHEVs with smaller battery packs will potentially have a higher percentage of the

usable battery capacity converted into kinetic energy. This is because the kinetic energy of a vehicle is

related to the mass of the vehicle, and there is not a 1:1 scaling of vehicle mass to battery capacity. A

doubling in battery capacity will roughly double the mass of the battery pack, but this will not double

the mass of the vehicle.

Equation 6 is the equation for kinetic energy.

Logger Data: Vehicle Level Analysis of PEVs 21

Equation 7 is the calculation for the gasoline equivalent electric energy at engine-on. The kinetic energy

is divided by Eff

Motor

which is the assumed motor efficiency of 90%. The 90% motor effieciency was

chosen as it provided a simple round number that was in line with published motor efficiencies and fit

the data that had been collected. The energy consumed at the battery would be higher than the output

of the electric motor and must be accounted for.

Equation 6. Kinetic Energy Calculation









where m is the mass of the vehicle (assumed to be curb weight plus 200lbs), and v

AEO

is the velocity of

the vehicle at engine-on.

Equation 7. Electric Energy Gasoline Equivalent at Engine-On





=   











 ,

where Eff

Motor

is the motor efficiency, which was assumed to be 90%.

3 Logger Data: Vehicle Level Analysis of PEVs

In this section, we present our results and observations on PEV usage at the vehicle level using data

collected from the loggers. In total, 137 BEVs, 221 PHEVs, and 11 FCVs. 23 BMW i3 REX had trouble

acquiring data and were dropped from our analysis. There was one Kia Soul (111 mile range) and one

Fiat 500e (84 mile range), which were also dropped for our analysis due to exceptionally low sample

size. The other vehicles have reliable data for most parameters, have reliable data for longer than 120

days, and can be considered for the analysis. They are considered in the vehicle level analysis presented

in this section. To take advantage of the wealth of vehicle-usage information, all the remaining PEVs

were considered in the vehicle-level analysis. The sample size of vehicles and households used in the

household level analysis is provided in the following section. All the descriptive analyses and related

summary statistics summarized in Table 3 to Table 14

are from the loggers. Likewise, the descriptive

analyses and related summary statistics depicted in Figure 8 to Figure 54 are from the loggers.

Data Description

Table 3-Table 29 summarize, respectively, the data collected on BEV driving, BEV charging, PHEV driving,

PHEV charging, FCV driving and FCV refueling from the loggers. The summaries include vehicle data,

spanning approximately 1 year for each logged car, collected over the five year course of this study.

More granular breakdowns of the PEV driving and charging summaries are provided in following

chapters. From the raw data, which includes noticeably short trip events of zero to a few hundred yards,

we used a filtering criteria of 1 km to denote a valid trip for PHEVs, BEVs and FCVs. The filtering criteria

of 1 km is based on filtering out GPS noise and extemely short trips registered at the loggers with no

energy use and the rule of thumb values for acceptable walking distances (Smith and Butcher 2008;

Yang and Diez-Roux 2012). For the charging sessions, a cutoff of 1 kWh for the BEVs and 0.25 kWh for

the PHEVs was used. In addition, we filtered out trips and charging sessions that did not report variables

that are usually included for this type of vehicle such as: battery SOC, distance traveled, energy charged,

and driving energy (electrical and gasoline) consumed. Overall, 99.8% of charging energy (PHEVs and

BEVs), 99.7% of BEV VMT, 99.6% of PHEV VMT and 88% of FCV VMT were still retained after filtering.

We further classified the BEVs into 6 types based on the battery capacity: Leaf-24, Leaf-30, RAV4 EV-42,

Logger Data: Vehicle Level Analysis of PEVs

Bolt-60, Tesla Model S-60_80, and Tesla Model S-80_100. For the PHEVs, we adopted a similar approach

and classified the PHEVs into 6 types: Prius Plug-in-4.4, C-Max/Fusion-7.6, Prius Prime-8.8, Pacifica-16,

Volt-16, and Volt-18. Since both the Ford C-Max Energi and Fusion Energi 7.6 have the same battery

capacity, we combined them together as C-Max/Fusion-7.6. Chevy Volts model year (MY) 2016 or later

have bigger batteries than those of earlier model years and are classified as Volt-18. The rest of the Volts

were classified as Volt-16.

Table 3. BEV Driving Data Overview

BEV Type

Number

Vehicles

Trips-

Raw Data

Total

VMT-

Raw Data

Trips-

Filtered

Data

VMT-

Filtered

Data

Average

Driving

Days/Vehicle-

Filtered Data

Leaf-24 29 40714 263645 34061 262210 264

Leaf-30 28 38488 268780 33435 267535 266

RAV4 EV-42 5 8775 60163 7716 60005 344

Bolt-60 27 47760 382603 39479 381032 295

Model S-60_80 23 24295 375573 21057 374908 279

Model S-80_100 25 22066 285046 20032 284682 241

All BEVs 137 182098 1635810 155780 1630372 272

Table 4. BEV Charging Data Overview

BEV Type

Number

Vehicles

Charging

Sessions-

Raw Data

Total

kWh-Raw

Data

Charging

Sessions-

Filtered

Data

Total

kWh-

Filtered

Data

Average

Charging

Days/Vehicle-

Filtered Data

Leaf-24 29 9211 57739 8707 57638 219

Leaf-30 28 6880 62836 6744 62804 185

RAV4 EV-42 5 1527 20003 1482 19993 253

Bolt-60 27 9434 100612 8351 100535 176

Model S-60_80 23 6994 139844 6737 139777 217

Model S-80_100 25 5079 106753 4902 106710 152

All BEVs 137 39125 487787 36923 487456 192

Logger Data: Vehicle Level Analysis of PEVs

Table 5. PHEV Driving Data Overview

PHEV Type

Number

Vehicles

Trips-

Raw

Data

Total

VMT-

Raw

Data

Trips-

Filtered

Data

VMT-

Filtered

Data

Average

Driving

Days/Vehicle-

Filtered Data

Prius Plug-in-4.4 22 36915 315465 31473 314231 312

C-Max/Fusion-7.6 60 88796 729091 74971 726389 273

Prius Prime-8.8 27 42548 337533 37205 336465 301

Pacifica-16 28 39602 273916 34661 272970 238

Volt-16 44 60830 568379 51419 566354 291

Volt-18 40 56957 460974 49946 459257 299

All PHEVs 221 325648 2685359 279675 2675665 284

Table 6. PHEV Charging Data Overview

PHEV Type

Number

Vehicles

Charging

Sessions-

Raw Data

Total kWh-

Raw Data

Charging

Sessions-

Filtered

Data

Total kWh-

Filtered Data

Average

Charging

Days/Vehicle-

Filtered Data

Prius Plug-in-4.4 22 8101 17803 7700 17774 232

C-Max/Fusion-7.6 60 25267 76506 21727 76192 213

Prius Prime-8.8 27 9587 29975 8629 29948 226

Pacifica-16 28 10342 70338 9936 70301 199

Volt-16 44 26501 100008 15868 99949 251

Volt-18 40 14104 83679 11043 83621 213

All PHEVs 221 93902 378311 74903 377785 222

Logger Data: Vehicle Level Analysis of PEVs

Table 7. FCV Driving Data Overview

FCV Type

Number of

Vehicles

Trips- Raw

Data

Total

VMT- Raw

Data

Trips-

Filtered

Data

Total VMT-

Filtered

Data

Average

Driving

Days/Vehicle-

Filtered Data

Mirai 11 15301 104133 11327 91164 247

Table 8. FCV Refueling Data Overview

FCV Type

Number of

Vehicles

Refueling

Sessions

Total

Hydrogen

(kg)

Average

Refueling

Days/Vehicle

Mirai 11 408 823.40 20.67

*If two vehicle models’ annualized VMT means do not share a letter, they are significantly different.

Figure 8. Average Annualized VMT by PEV Model

Logger Data: Vehicle Level Analysis of PEVs 25

Table 9. Annualized VMT by Vehicle Types

Vehicle Type Average Standard Error Median Standard Deviation Max

ICE 9009.7 308.5 8106.3 5560.8 37786.1

PHEV 12808.0 363.3 12080.6 5400.2 36273.1

BEV 12876.3 600.0 11722.6 7022.7 50326.6

SRBEV 10311.8 594.0 10017.3 4677.2 26642.5

LRBEV 14996.2 913.2 12502.4 7908.3 50326.6

FCV 10737.8 879.9 10148.4 3048.3 16009.5

Figure 8 and Table 9 summarize descriptive statistics of annual VMT for all types of logged vehicles. On

average, the BEVs had a slightly higher annualized VMT than the PHEVs. LRBEVs (Long-range BEVs) had

the highest average and median annual VMT out of all vehicle technologies logged, even compared to

the ICE. SRBEVs refer to short-range BEVs.

Figure 9. BEVs: Percentage Share of Total VMT by Trip Speed (in mph)

Logger Data: Vehicle Level Analysis of PEVs 26

Figure 10. PHEVs: Percentage Share of Total VMT by Trip Speed (in mph)

Figure 9 and Figure 10 show the share of total VMT by trip speed bin (the vehicle’s driving speed) for

BEVs and PHEVs, respectively. Compared to all other PEVs (Leaf, RAV4, Bolt, and all PHEV types), the

Model S BEVs have a higher share of VMT at trip speeds 60 mph or faster. In fact, almost 50% of Model S

total VMT was from trips at speeds of 60 mph or faster, whereas only around 15% of its VMT was from

trips with speeds less than 30 mph. Furthermore, the high all-electric range (AER)/battery capacity could

have contributed to the Model S having the highest share of VMT at very high speeds (75 mph or more)

compared to all other PEVs. Among the PHEVs, there is a comparable share of VMT from trips driven at

speeds of 60 mph or faster (about 40%) and from trips driven at speeds less than 45 mph (about 40%).

At very high speeds (75 mph or more), Prius Plug-in-4.4 has the lowest share of total VMT, followed by

Volt 16, Prius Prime-8.8, Volt 18, Pacifica 16, and C-Max/Fusion-7.6.

4.8

4.9

5.5

5.2

4.3

5.4

14.0

15.0

14.9

16.3

13.0

15.0

20.3

20.2

21.8

21.0

19.2

20.9

19.4

18.7

19.2

18.6

24.1

20.5

35.7

32.6

31.6

30.7

33.3

30.8

5.8

8.7

6.9

8.2

6.1

7.5

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Prius Plug-in 4.4 C-Max-Fusion 7.6 Prius Prime 8.8 Pacifica 16 Volt 16 Volt 18

Percentage of Total VMT

0-15 mph 15-30 mph 30-45 mph 45-60 mph 60-75 mph 75-155+ mph

Battery Electric Vehicles Driving

As shown in Table 10, on average, the Leaf-24 and Leaf-30 drivers make more trips and drive shorter trip

distances than do Tesla Model S-60_80 and Tesla Model S-80_100 BEV drivers. The average trip distance

of the RAV4 EV-42 is comparable to that of the Leafs (Leaf-24 and Leaf-30). The average trip distance of

the Model S BEVs (60_80 and 80_100) was almost twice that of the Leafs and RAV4 EV-42. The average

trip distance of the Leafs did not vary much between weekdays and weekends. Except for the Leaf-30

and Tesla Model S-80_100 vehicles, the weekday maximum trip distance of Leaf-24, RAV4 EV-42, Bolt-

60, and Tesla Model S-60_80 vehicles was higher than their respective weekend maximum trip distance

(). More than half of the Leaf-24, Leaf-30, RAV4 EV-42 and Bolt-60 trips were less than 5 miles. About

15% of Tesla Model S-60_80 and Tesla Model S-80_100 trips were more than 30 miles, whereas around

95% of the Leaf-24, Leaf-30, RAV4 EV-42 and Bolt-60 trips were less than 30 miles (Figure 12). RAV4 EV-

42 had the highest kWh/mile consumption for average trip speeds ranging from 15 mph to 75 mph.

Logger Data: Vehicle Level Analysis of PEVs 27

Tesla Model S-60_80 and Tesla Model S-80_100 have significantly lower average kWh/mile consumption

at higher trip speeds than their respective average kWH/mile consumption at lower trip speeds (Figure

13).

Table 10. BEV Driving Trip Level Summaries (on days when the BEV was driven)

BEV Type

Average

Trips/Day

Average Trip

Distance (miles)

Average

kWh/Trip

Average

kWh/Mile

Leaf-24 4.5 7.7 1.8 0.2

Leaf-30 4.5 8.0 2.1 0.3

RAV4 EV-42 4.5 7.8 2.9 0.4

Bolt-60 4.8 9.7 2.4 0.3

Model S-60_80 3.3 17.8 6.0 0.4

Model S-80_100 3.3 14.2 4.9 0.4

All BEVs 4.2 10.5 3.0 0.3

Logger Data: Vehicle Level Analysis of PEVs

*If two vehicle models’ weekday daily mile means do not share a letter, they are significantly different.

Figure 11. Average and Maximum Trip Distance on Weekdays and Weekends by BEV Type

Figu

re 12. Percentage of Trips by Trip Distance Bins (miles) and by BEV Type

Logger Data: Vehicle Level Analysis of PEVs

Figure 13. Effect of Speed on Energy Consumption per Mile

Logger Data: Vehicle Level Analysis of PEVs 30

*If two vehicle models’ VMT means do not share a letter, they are significantly different.

Figure 14. Average Daily VMT of the Individual BEVs by BEV Model

Figure 14 shows the average daily VMT of the individual BEVs. The average daily VMT of the Leaf-24,

Leaf-30, and RAV4 EV-42 was less than the overall BEV fleet average daily VMT of 43.54 miles. The

average daily VMT of the Bolt-60, Tesla Model S-60_80 and Tesla Model S-80_100 was greater than

overall BEV fleet average daily VMT. Though the average daily VMT could be a useful and

straightforward metric to compare how different BEVs utilize their All-Electric Range (AER), it could be

quite misleading as we will illustrate in the following subsection.

Routine Destinations and Activity Spaces

In addition to their effect on daily and annual VMT and trip distance, battery capacity and AER have a

strong impact on the number of different destinations BEVs are used to access and over how wide an

area these destinations are spread. Short-range vehicles that are reliant on daily charging at one or

more locations cannot be used as flexibly as longer-range vehicles that can go multiple days between

charging events, and their owners are more likely to use an internal combustion engine vehicle as the

household’s primary means of transportation for long trips. Supplemental data to support this can be

found in Figure 76. In this section, analysis of the importance of routine destinations on annual travel

and of the total area covered by each vehicle’s annual travel show that shorter-range BEVs are used

largely for regular travel between home and a single other location whereas longer-range luxury BEVs

are used to access a wider range of destinations spread over a much larger area.s The Bolt-60 in our

sample, have usage patterns in between the shorter-range and longer-range BEVs, despite larger

batteries and range similar to the shorter-range Tesla Model S.

An analysis of the common destinations visited by vehicles in this study that were supplied with GPS

loggers indicated that shorter-range BEVs are mainly used for regular travel between home and a small

Logger Data: Vehicle Level Analysis of PEVs 31

number of routine destinations, whereas longer-range BEVs are used for less routine travel and are

much more likely to be used for overnight travel away home. Routine locations were identified by

clustering the spatial coordinates of the endpoints of all trips made by each vehicle using the Density-

Based Spatial Clustering of Applications and Noise (DBSCAN) algorithm with a distance threshold of 250

meters and a cluster density threshold of 5 neighboring points. DBSCAN is commonly used to identify

distinct locations from GPS points because this method clusters points based only on proximity rather

than to populate a predetermined number of clusters and because it distinguishes between points that

are part of a feature and “noise” points (Schubert et al., 2017). Across all vehicles in the study, 87.9% of

destinations were classified as being in a cluster and 24.4% of destinations were classified as being in the

most-visited cluster. Destinations within a single cluster were classified as being the same location, and

points not within any cluster were classified as having been visited only once. For all vehicles, the most-

visited location was also the last destination of almost every day, which indicates that it is a home

location. Locations that were visited more than 12 times (approximately once per month) over the study

period were classified as routine destinations for the purposes of this analysis. Monthly recurrence was

chosen as the cutoff for routineness because it is near the maximum suggested duration for studies of

routine activity space and routine travel behavior (Schönfelder & Axhausen, 2003; Zenk et al., 2018), but

the results were similar when the threshold was set to visits every two weeks.

To analyze the relative significance of routine travel for vehicles of different ranges, each vehicle’s

annual VMT was partitioned based on whether any routine destinations were visited on the travel day.

The four categories used for this analysis were Primary nonhome destination, which comprised days in

which the vehicle started or ended the day at home and made at least one trip to the most common

non-home destination; Other routine nonhome destinations, which comprised days in which the vehicle

started or ended the day at home, did not visit the most common non-home destination, but visited

another destination that was classified as routine; No routine nonhome destinations, which comprised

days in which the vehicle started or ended the day at home but did not travel to any other routine

destinations; and Away from home area, which included all travel on days in which the vehicle did not

start or end at home.

The results of this partitioning are shown in Figure 15. In general, the most-visited destination is most

important for short-range BEVs, and days with non-routine travel (both local and overnight) make up a

much larger share of travel for longer-range BEVs. Interestingly, the Bolt-60 in the study showed a

similar importance of routine travel as the Leafs, despite having as much range as some of the Teslas,

which suggests that the size and comfort of the vehicles together with cast charging availability may

influence the way these vehicles are used separate from the impact of battery capacity. For all vehicle

types, the largest fraction of annual VMT occurred on days that included the most-common non-home

destination (likely a commute location); this ranged from about 60% of annual VMT for the Leaf vehicles

in the study to less than 40% of annual VMT for the Tesla Model S and RAV4 EV-42 vehicles. The smallest

portion of travel for all vehicle types occurred on days in which none of the trips ended at home, but

these days accounted for a much larger proportion of annual VMT for Tesla Model S vehicles (about

15%) than for Leaf and Bolt-60 vehicles (under 6%). Travel on non-commute days to locations visited at

least once a month made up a similar share of VMT across all vehicle types, although it was slightly

higher for the small number of RAV4 EV-42 in the study.

Logger Data: Vehicle Level Analysis of PEVs 32

Figure 15. Proportion of annual VMT on days with and without routine destinations

Activity spaces are a geographic concept that are used to model the range of opportunities and

interactions a person has access to during their habitual movements. Standard deviational ellipses

(SDEs) are particularly useful for estimating activity spaces from GPS data both because they can be

processed quickly without relying on secondary information about the road network and because they

can be scaled to include an arbitrary proportion of destinations (Sherman et al., 2005). SDEs are

constructed from a set of coordinates by identifying an axis on which the points vary most significantly

and determining the standard deviation of the coordinates along that axis and a perpendicular axis. An

ellipse is constructed with its center at the midpoint of the coordinates and its axes scaled to the

standard deviations along the two rotated axes. For this analysis, we calculate SDEs with a radius of a

single standard deviation to capture approximately 68% of each vehicle’s destinations that are closest to

the center and then compute the area of this ellipse in square miles. GPS coordinates with destinations

spread over a larger geographic area will have larger SDEs, and those with fewer destinations will have

smaller SDEs, even if those destinations are far apart. An example of an SDE computed from destinations

visited by a Tesla driver is shown in Figure 16; note that the SDE covers their frequently visited

destinations, but not their less-frequently visited ones.

10%

20%

30%

40%

50%

60%

70%

Primary nonhome

destination

Other routine nonhome

destinations

No routine nonhome

destinations

Away from home area

Proportion of Annual VMT

Leaf 24 Leaf 30 RAV4 EV 41.8 Bolt 66 Model S 60-80 Model S 80-100

Logger Data: Vehicle Level Analysis of PEVs 33

Figure 16. Trip destinations of Tesla Model S-60_80

Analysis of vehicles SDE activity spaces show that longer-range BEVs (particularly Tesla Model S vehicles)

are also used to travel frequently over a much wider area and potentially a much larger range of

opportunities than shorter-range BEVs. The results of the SDE analysis are plotted against annual VMT in

Figure 17, with a logarithmic scale on the vertical axis. Vehicles with larger SDEs are used to visit

destinations spread over a much larger area. As the slight upward trend in the plot shows, vehicles that

travel more miles per year generally cover a larger area, but longer-range BEVs, particularly Teslas, have

larger SDEs than other BEVs with similar annual VMTs. The several Teslas and one Bolt-60 with SDEs

larger than 3,000 square miles were used to travel over an area more than an order of magnitude larger

than the most wide-ranging Leafs.

Logger Data: Vehicle Level Analysis of PEVs 34

Table 11. Summary statistics for area of Standard Deviation Ellipse area activity space by vehicle type

Vehicle Type

Minimum

(mi

)

percentile

(mi

)

Median

(mi

)

percentile

(mi

)

Maximum

(mi

)

Leaf-24 4 15 32 70 335

Leaf-30 9 27 46 144 357

RAV4 EV-42 17 33 42 71 88

Bolt-60 19 58 101 155 2,941

Model S-60_80 33 221 341 866 6,870

Model S-80_100 62 164 487 1,624 5,100

Figure 17. Area of standard deviation ellipse of vehicle destinations plotted against annual VMT

Table 11 identifies quartiles of SDE areas for the six vehicle types and shows that almost all the Tesla

Model S traveled over a wider area than the median Leaf. In addition, the median Tesla traveled over a

wider area than any Leaf. As is the case with the analysis of destinations, the activity spaces of Bolt-60

100

1,000

10,000

0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000

Standard Deviation Ellipse Area, square miles

Annual VMT

Leaf 24 Leaf 30 RAV4 EV 41.8 Bolt 66 Model S 60_80 Model S 80_100

Logger Data: Vehicle Level Analysis of PEVs 35

drivers are roughly halfway between those of Leaf and Tesla drivers, despite their vehicles having as

much range as many of the Tesla Model S, perhaps due to differences in access to chargers away from

home.

Figure 18. Percentage of Daily VMT by Distance Bins: Weekdays vs. Weekends

Figure 18 shows the share of weekdays and share of weekends when the BEV was driven, binned by

daily VMT. For all BEV models, the percentage of days with trips that were 10 miles or less was higher on

the weekends than on weekdays. The Leaf-24, Leaf-30, RAV4 EV-42 and Tesla Model S-60_80 had a

higher percentage of days with trips that were 100 to 200 miles on weekends than on weekdays.

Using the criteria for 50 miles or more to define Long Distance Travel (LDT)(BTS 2017) days, Figure 19

shows the share of VMT accomplished on these days as a percentage of the total VMT. A given point on

one of the six box and whisker plots denotes the average share of LDT VMT for a specific vehicle,

belonging to the vehicle model category represented by that box plot. The box outlines the range of

data between the first and third quartile. VMT on LDT days accounted for an average of 52% of the total

VMT for all BEVs, and 36.1%, 44.7%, 40.0% and 57.3% for the Leaf-24, Leaf-30, RAV4 EV-42, and Bolt-60,

respectively. Both the Tesla Model S-60_80 (65.9%) and Tesla Model S-80_100 (64.8%) have almost two-

thirds of their VMT accomplished on LDT days, perhaps indicating that long-range BEVs are more often

used for long-distance travel rather than regular weekday commuting.

Logger Data: Vehicle Level Analysis of PEVs 36

Figure 19. Share of VMT on LDT (50 miles or more) as Percentage of Total VMT by BEV Type

PEVs Used for Commuting

Figure 20 below shows share of PEVs by type that were used by HH members working fulltime for

commuting purposes and non-commuting purposes across all the HHs in which logger data was used.

Overall, about 70% of each vehicle type was used for commuting, except for the Chrysler Pacifica-16

where only 50% were used for commuting. This could be attributed to the fact that the Chrysler Pacifica-

16 was the only minivan in our sample and the rest of the vehicles were all sedans.

The whiskers, or lines connected to the box, illustrate the range between the minimum value and first quartile as

well as the range between the third quartile and the maximum value. The line intersecting the box denotes the

median value and the cross shows the average value.

Logger Data: Vehicle Level Analysis of PEVs 37

Figure 20. Number of PEVs Used for Commute Purposes by Type.

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Toyota Prius

Plug-in

Toyota Prius

Prime

Chrysler

Pacifica Hybrid

Ford Energi Chevrolet Volt Nissan Leaf Chevrolet Bolt

Tesla Model S

Used for Commuting Not Used for Commuting

Battery Electric Vehicle Charging

Descriptive summaries and analyses depicted in Figure 21- Figure 33 are based on the data collected

from the loggers. BEV charging summary statistics are presented in Table 12. Figure 21 shows the

probability that the BEV charges on a given day within the duration for which it was logged, called the

logging window. Figure 22 and Figure 23, respectively, depict the percent share of charging sessions and

charged kWh by charging level.

Table 12 summarizes the key charging related information of the BEVs. When we consider only the days

when the BEV charged, the Bolt-60 had a comparable number of charging sessions per day, as did the

Leaf-24 and Model S-60_80. However, when we include the days on which the BEV did not charge, as

may be expected, the Model S-60_80 and Model S-80_100 had fewer charging sessions per day than the

Leaf-24 and Leaf-30, respectively. The Bolt-60 had the longest average charging duration per day,

whereas the RAV4 EV-42 had the lowest.

Logger Data: Vehicle Level Analysis of PEVs

Table 12. Charging Summaries on by BEV Type

Within the

Logging

Window

Including

Days When

BEV Did not

Charge

BEV

Average

Sessions/Day

Average

DCFC

Sessions/Day

Average

L1/L2

Sessions/Day

Average

kWh/Day

Average

Duration/Day

(minutes)

Average

VMT/Day

(miles)

Leaf-24 0.9 0.0 0.8 5.8 196.0 27.2

Leaf-30 0.7 0.1 0.6 6.7 138.2 28.8

RAV4 EV-42 0.8 0.0 0.8 10.5 106.3 31.7

Bolt-60 0.8 0.0 0.8 10.1 246.2 38.6

Model S-

60_80

0.8 0.1 0.7 17.6 160.1 47.3

Model S-

80_100

0.6 0.1 0.5 13.3 108.6 35.1

Logger Data: Vehicle Level Analysis of PEVs

On Days

When the

BEV Charged

BEV

Average

Sessions/Day

Average

DCFC

Sessions/Day

Average

L1/L2

Sessions/Day

Average

kWh/Day

Average

Duration/Day

(minutes)

Average

VMT/Day

(miles)

Leaf-24 1.4 0.1 1.3 9.0 306.9 36.4

Leaf-30 1.3 0.3 1.0 12.1 247.9 43.3

RAV4 EV-42 1.2 0.0 1.2 15.8 159.2 39.2

Bolt-60 1.7 0.0 1.7 20.5 499.1 53.4

Model S-

60_80

1.4 0.2 1.2 28.0 254.6 62.9

Model S-

80_100

1.3 0.2 1.1 27.7 159.2 56.4

Figure 21. Probability of Charging Within the Logging Window of Individual BEVs by BEV Type

Logger Data: Vehicle Level Analysis of PEVs

Figure 22. Share of Charging Sessions by Charging Level and BEV Type

Figure

23. Share of Charging kWh by Charging Level and BEV Type

Out of the 36,921 charging sessions in total, 22% were at L1, 69% were at L2, and 9% were DCFC

sessions. L2 charging accounted for most charging sessions and charged kWh for all the BEV types. Leaf-

30 had the highest share of DCFC sessions and the highest share of charged kWh from DCFC charging.

DCFC charging sessions by Leaf-30 accounted for close to 40% of all the DCFC charging sessions,

Logger Data: Vehicle Level Analysis of PEVs 41

followed by Model S-60_80 and Model S-80_100, which respectively accounted for 23% and 20% of all

the DCFC charging sessions. Figure 24 shows the percent of charging sessions (all charging levels

combined) by start time (hourly intervals) on weekdays and weekends. There was a noticeable weekday

peak around 8 am, which can be attributed to charging at work, and the 11 pm-1am window on

weekdays, which is typical of home charging. On the weekends, the highest percentage of charging

sessions occur during the 11pm-1am window, followed by 9 pm and 7 pm. Since the RAV4 EV-42 had

only a few L1 sessions and is not DCFC compatible, it has been omitted from the charging session

starting time, charger utilization, charging session starting and charged SOC plots (Figure 25

, Figure 27,

Figure 29, Figure 31, Figure 32).

Figure 24. Charging Session Starting Time: Weekdays vs. Weekends (all BEVs and all charging levels)

Figure 25-Figure 27 show the results of a closer inspection of the charging session start time by charging

level (L1, L2, and DCFC). For all BEV types, most of the highest percentages of L1 Sessions were between

2pm and 11pm. The peak in L1 charging session start time was around 9pm for Leaf-30 and around

11pm for the Leaf-24, respectively. Across all BEV types, the highest percent of L2 sessions started

around or after 11pm. There was a noticeable spike in the share of L2 charging sessions for all BEVs

starting at around 8am, perhaps indicative of access to L2 charging away from home. Another noticeable

spike in the share of L2 charging sessions was for the RAV4 EV-42 at 3am. Not all BEV drivers use DCFC

on our study as 3 Bolt-60 and 5 NIssan Leaf havn’t had DCFC capabilties but 22 Bolts-60, 13 Leaf-24s, 5

Leaf-30s, and 1 Model S-80_100 did not use it even once on the logging period. An interesting

observation with respect to DCFC charging was the noticeable peak in DCFC charging session start times

of the Leaf-30 at 5am. This time window could potentially reflect the preference of Leaf-30 owners to

stop and use DCFC charging on their commutes. The peak DCFC sessions were 9am and noon for Bolt-60,

8am and 6pm for both the Model S-60_80 and Model S-80_100, and 11am and 2pm for the Leaf-24.

Logger Data: Vehicle Level Analysis of PEVs

Figure 25. Percentage of L1 Charging Start Times by Time of Day and BEV Type (RAV4 EV-42 is

excluded from this dataset, as it was very rarely charged on an L1 charger)

Figure 26. Percentage of L2 Charging Start Times by Time of Day and BEV Type

Logger Data: Vehicle Level Analysis of PEVs 43

Figure 27. Percentage of DCFC Charging Start Times by Time of Day and BEV Type (RAV4 EV-42 is

excluded from this dataset, as it cannot be charged on a DCFC)

Figure 28-Figure 30 show the average charging session duration and average kWh charged by charger

level. For L1 charging, the Leaf-30, on average, had higher charging energy per session and longer

charging session duration on weekends than on weekdays. The Bolt-60, on average, had similar charging

energy per session and similar charging session duration on weekends and on weekdays when using L1

charging. On average, the Leaf-24, Model S-60_80, and Model S-80_100 all had lower L1 charging energy

per session and shorter L1 charging session duration on weekends than on weekdays. When using L2

charging, all BEV models, except the Bolt-60 and RAV4 EV-42, on average, had lower charging energy per

session and shorter charging session duration on weekends than on weekdays. The Bolt-60 had higher

average charging energy per session and longer average charging session duration on weekends than on

weekdays, when using L2 charging. The RAV4 EV-42 had similar average L2 charging energy per session

and similar average L2 charging session duration on weekends and on weekdays.

The average charging session duration and average amount of charge per session on DCFCs were similar

between the weekdays and weekends for the Leaf-30, but greater on weekends than on weekdays for

the Bolt-60, Model S-60_80, and Model S-80_100 (Figure 30). On the other hand, a vehicle with a large

battery size, the Leaf-24, had shorter average charging session duration and less average charge per

session on weekends than on weekdays.

Logger Data: Vehicle Level Analysis of PEVs

*If two vehicle models’ weekday kWh charged energy means do not share a letter, they are significantly different.

Figure 28. Average L1 Charging kWh Charged and Charging Duration: Weekdays vs Weekends (RAV4

EV-42 is excluded from this dataset, as it was very rarely charged on an L1 charger)

Logger Data: Vehicle Level Analysis of PEVs

*If two models’ weekday kWh charged means do not share a letter, they are significantly different.

Figure 29. Average L2 Charging kWh Charged and Charging Duration: Weekdays vs Weekends

*If two vehicle models’ weekday kWh charged energy means do not share a letter, they are significantly different.

Figure 30. Average DCFC Charging kWh and Duration: Weekdays vs Weekends

Logger Data: Vehicle Level Analysis of PEVs 46

Figure 31-Figure 33 show the average charging session starting and ending battery SOC by charger level

on weekdays and on weekends. When using L2 charging, the Leaf-30 compared to the other BEV types

had the lowest average starting SOC on weekdays, but when using L1 charging, it had the second highest

average starting SOC on weekdays. The Bolt-60, compared to all other BEV types, had the lowest

average charged SOC when using L1 charging on weekdays, L1 charging on weekends, and when using

L2 or DCFC charging on weekdays. The Bolt-60 also had the highest average starting SOC when using L1

or L2 charging on weekdays and on weekends. When using DCFCs, the Leaf-30 average starting SOC on

weekdays was the lowest and its charged SOC on weekdays and on weekends was the highest. Overall,

for short-range BEVs (Leaf-24 and Leaf-30), the charging session ending SOC was around 90% or more on

weekdays and on weekends when using L1 or L2 charging. In addition, the Leaf-30 average charging

session ending SOC was highest (90% or more) when using DCFCs on weekdays and on weekends. The

Model S-60_80 and Model S-80_100 started their L2 charging sessions on weekdays and on weekends at

higher SOCs compared to the short-range BEVs (Leaf-24 and Leaf-30).

Figure 31. L1 Charging: Average Starting and Charged SOC

Logger Data: Vehicle Level Analysis of PEVs

Figure 32. L2 Charging: Average Starting and Charged SOC

Figure 33. DCFC: Average Starting and Charged SOC

Logger Data: Vehicle Level Analysis of PEVs 48

Plug-in Hybrid Electric Vehicles (PHEVs) Driving

Results presented in Table 13 and depicted in Figure 34-Figure 44 in this section are based on the data

collected from the loggers. In this section, we present the vehicle level analysis of the PHEVs. We used

the methods presented in Section 2.5 to estimate the trip level distribution of electric vehicle miles

travelled (eVMT), gasoline vehicle miles travelled (gVMT), and the total energy consumption per trip,

reported in gallons of gas and kWh of electricity used. We also compare the different PHEVs in terms of

their utility factor (UF), which is the ratio of the charge depleting range to the distance travelled (SAE

2010). Compared to BEVs, which have only one source of propulsive power, estimating the eVMT of

PHEVs is not as straightforward since the PHEVs have three driving modes: Charge Sustaining (CS),

Charge Depleting Blended (CDB), and All Electric (AE) or Zero Emission (ZE) modes. In the CS mode, a

PHEV is driven like a regular hybrid electric vehicle using only gasoline. When the PHEV is driven in ZE

mode, the engine is never turned on and only electricity is consumed, whereas in the CDB mode, both

gasoline and electricity are consumed.

Table 13. PHEV VMT, eVMT, gVMT, Fuel and Energy Consumption by PHEV Type

Table 13 provides an overview of the PHEV driving and charging data considered in the vehicle level

analysis. Table 13 shows the total eVMT, total miles travelled on gasoline (gVMT), and total PHEV VMT

for the individual PHEVs by type.

Figure 35 shows the average utility factor UF by PHEV type. On average, the Volt-18 had the highest UF,

followed by the Volt-16. The UF of the C-Max/Fusion PHEV was half that of the Volt-18. The UF

measured in our project is different than that used for current policies and regulations. Current

regulations are based on a UF standardized in the SAE J2841 (SAE 2010) that is based on daily miles from

travel surveys and the assumption that each vehicle starts the travel day fully charged. Our sample

suggests that not all PHEVs are charged every day and that different PHEVs charge differently.

Furthermore, we did not install loggers in vehicles that were used as hybrids or charged less than 4

times per month. Based on the project survey, 4% of the Chrysler Pacifica-16, 5.9% of Volt owners, 9% of

the Toyota Prius Prime-8.8, 16.5% of the Ford Energi owners, and 18.5% of the Prius Plug-in-4.4 owners

PHEV Type

Total eVMT

(miles)

Total gVMT

(miles)

Total VMT

(miles)

Total

Gasoline

Consumed

(Gallons)

Total

Charging

Energy(kWh)

Prius Plug-in-4.4 46106 268125 314231 5528 17774

C-Max/Fusion-7.6 249868 476501 726389 11800 77174

Prius Prime-8.8 142209 184477 336465 3593 29948

Pacifica-16 137049 123153 272970 4131 70301

Volt-16 361704 204650 566354 5947 99949

Volt-18 313926 145331 459257 3952 83621

All PHEVs 1250863 1402236 2675665 34951 378767

Logger Data: Vehicle Level Analysis of PEVs 49

drove mostly on gas and were not accumulating eVMT. We believe that these figures underestimate the

phenomena because of a selection bias, where users who do not plug in their cars are less likely to take

and finish a survey on the topic. Figure 36 shows the UF for each of the vehicles based on the SAE2841

standard, the actual eVMT and VMT measured, and the utility factor adjusted to the survey results,

including the vehicles with utility factor of zero. For all vehicles, except the Volt-16, we measured lower

UFs than the SAE standard. For instance, the logged Prius PHEVs achieve only 62% of the expected UF or

48% when accounting for users who are not plugging in. For the longer-range Volts, we measured UFs

that were closer to the values determined by the SAE2841.

Figure 34. PHEV eVMT, gVMT, and VMT of Individual PHEVs by PHEV Type

Logger Data: Vehicle Level Analysis of PEVs

Figure 35. Utility Factor (UF) for Each PHEV by Type

Figure 36. Average UF by PHEV Type

0.29

0.45

0.53

0.62

0.66

0.76

0.18

0.40

0.47

0.51

0.67

0.69

0.14

0.35

0.43

0.49

0.63

0.66

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

Prius Plug-in 4.4 C-Max-Fusion 7.6 Prius Prime 8.8 Pacifica 16 Volt 16 Volt 18

Comb Utility Factor UC Davis Study with Hybrid only Users

Logger Data: Vehicle Level Analysis of PEVs 51

Figure 37. Average Trips per Day by Driving Mode

As shown in Figure 37, At the day level, on average, the Prius Plug-in-4.4 was driven approximately 7

trips per day, the C-Max/Fusion-7.6 and Pacifica-16 were driven approximately 6 trips per day, and the

Volt-16, appoximately 5.5 trips per day. The Volt-16, Volt-18, and Prius-Prime-8.8 had fewer average

daily trips than the PHEV fleet (5.52 trips/day). On average, when compared to other PHEVs, the Volt-18

had the greatest share of trips accomplished on electricity alone (ZE only mode), also referred to as zero

emission trips. The Volt-18 also had the lowest share of trips that were accomplished on gasoline alone

in the charge sustaining mode (CS only mode).

Logger Data: Vehicle Level Analysis of PEVs 52

Figure 38. Percentage of Total PHEV Trips by PHEV Driving Mode

Referring to Figure 38, we can see that, compared to the other PHEVs, the Volt-18 by far had the lowest

percentage of CS only trips and the highest percentage of ZE only trips. In contrast, the Prius Plug-in-4.4

had the highest percentage of CS only trips and CDB/CS trips. At the PHEV fleet level, there was a

relatively even split between ZE only trips and CS only or CDB/CS trips. The Volt-18 had a higher share of

ZE only trips and lower share of CS only trips than did the Volt-16.

Referring to Figure 10, which showed the total share of VMT by trip speed bins and PHEV type, we see

that the Volt-18 had a slightly higher share of VMT accomplished at low trip speeds (30 mph or less) and

at high speeds (75 mph or more). The incremental battery capacity of Volt-18 compared to Volt-16 is

enabling the Volt-18 to do a higher share of blended trips. The share of CDB and CS trips for the Prius

Plug-in-4.4 is higher than for other PHEVs, simply due to its smaller battery.

Logger Data: Vehicle Level Analysis of PEVs 53

Figure 39. Share of Trips by Trip Distance Bins: Weekdays vs Weekends

Figure 39 shows the percent share of trip distance by trip distance bin on weekdays and weekends. At

least 90% of the trips were less than 30 miles for all the PHEV types on weekdays and weekends. During

weekends as compared to weekdays, PHEVs, except for the Pacifica-16, are driven on a higher share of

trips less than 10 miles and a lower share of trips of 10–20 miles. The Pacifica-16 has a slightly lower

share of trips less than 10 miles and around the same share of trips of 10-20 miles, during the weekends

as compared to weekdays. The Volt-18 has the same share of trips between 30–50 miles on weekends

(3%) as it does on the weekdays (3%); this contrasts with the Volt-16, which has a lower share of 30–50

mile trips on weekends(4%) than it does on weekends (7%).

Logger Data: Vehicle Level Analysis of PEVs 54

Figure 40. Daily Average VMT, eVMT, and gVMT Share by PHEV Type

Figure 40 shows the average daily VMT, eVMT and gVMT along with the percentage share of eVMT and

gVMT. The Prius Plug-in-4.4 had the highest daily average VMT, and the Volt-18 had the lowest daily

average VMT. Compared to the Volt-18, the Volt-16 had a higher daily average VMT, higher share of

gVMT, and lower share of eVMT. The average daily VMT of the C-max/Fusion-7.6 and the Volt-16 were

approximately equal but their split between eVMT and gVMT were opposite, with the Volt-16 eVMT

share being 64% and the C-max/Fusion-7.6’s gVMT share being 66%.

Logger Data: Vehicle Level Analysis of PEVs 55

Figure 41. Share of ZE Days, CS Days and CDB/CS Days

Figure 41 shows the share of days the travel was accomplished on electricity alone (ZE only days),

gasoline alone (CS only days), and gasoline and electricity (CDB/CS days). It also shows that even among

households that charged the vehicle regularly, for all PHEVs, 4.2% of days start with zero SOC, and this is

more common for the C-max/Fusion-7.6 than other vehicle types. The Volt-18 had an almost negligible

percentage of days when it was driven on gasoline only, with two-thirds of its driving days being ZE only

days. Even though the C-max/Fusion-7.6 has a bigger battery than the Prius Plug-in-4.4 has, it had a

higher percentage of CS only days.

Logger Data: Vehicle Level Analysis of PEVs 56

*If two vehicle models’ weekday LDT shares do not share a letter, they are significantly different.

Figure 42. Share of Long-Distance Travel (LDT; 50 miles or more) Days: Weekdays vs Weekends

*If two vehicle models’ weekday LDT shares do not share a letter, they are significantly different.

Figure 43. Share of Long-Distance Travel (LDT; 100 miles or more) Days: Weekdays vs Weekends

Figure 42 and Figure 43 show the share of days on weekdays and weekends, out of the total logged

days, that the PHEV was driven 50 miles or more and 100 miles or more, respectively. Pacifica-16 and

Logger Data: Vehicle Level Analysis of PEVs 57

Volt-18 had a higher percent of weekends than weekdays when the vehicle was driven 50 miles or more.

All the PHEVs had a higher percent of weekends than weekdays when they were driven 100 miles or

more.

Figure 44. Share of Daily VMT by Distance Bin: Weekdays vs Weekends

Figure 44 shows the percentage of weekdays and weekends by daily VMT bin. Approximately 70% of all

the PHEV distances (except for the Pacifica-16 and Volt-18) on weekdays were less than 50 miles. About

84% of the Volt-18 VMT on weekdays were less than 50 miles. During the weekends, for all the PHEVs,

60% of the distances were less than 35 miles. The Volt-18 had the highest percentage of weekdays when

it was driven 35–50 miles or 20–35 miles. The percentage of days when VMT was less than 10 miles was

almost double on weekends compared to weekdays, for all PHEV types. The percentage of days when

the VMT was 75–100 miles was lower on weekends than on weekdays for all PHEVs.

Plug-in Hybrid Electric Vehicle Charging

Results presented in Table 14 and depicted in Figure 45 – Figure 50 are based on the logger data. Table

14 summarizes the average number of PHEV charging sessions, kWh charged, and the duration of

charging per day by charging level.

9.2%

11.1%

12.3%

12.5%

11.4%

11.3%

23.1%

22.4%

17.7%

20.2%

23.4%

23.6%

22.0%

12.3%

17.2%

16.4%

18.2%

13.9%

18.5%

16.2%

18.3%

18.9%

20.0%

19.8%

19.5%

20.8%

19.6%

25.5%

21.9%

27.4%

23.6%

27.5%

24.3%

16.7%

19.8%

21.0%

21.6%

19.0%

19.1%

19.5%

19.4%

18.6%

17.9%

18.8%

17.7%

21.0%

18.9%

13.5%

11.3%

12.5%

11.2%

11.8%

12.9%

12.1%

18.5%

15.7%

19.3%

13.0%

16.1%

13.1%

15.8%

12.4%

11.0%

14.4%

11.3%

13.1%

10.4%

12.0%

6.7%

8.3%

5.9%

4.4%

9.7%

4.1%

6.9%

5.8%

6.3%

7.2%

5.1%

6.1%

5.7%

5.4%

4.9%

3.8%

5.0%

2.7%

4.6%

6.6%

5.8%

4.5%

4.4%

4.7%

5.2%

1.7%

1.1%

0.7%

0.6%

1.7%

1.1%

1.2%

1.7%

2.4%

1.5%

2.0%

2.2%

1.3%

1.9%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Prius Plug-in 4.4

C-Max-Fusion 7.6

Prius Prime 8.8

Pacifica 16

Volt 16

Volt 18

All PHEVs

Prius Plug-in 4.4

C-Max-Fusion 7.6

Prius Prime 8.8

Pacifica 16

Volt 16

Volt 18

All PHEVs

Weekday Weekend

Percentage of Weekdays and Weekends (%)

Percentage of Days by Daily VMT Bin(miles) by PHEV Type

0-10 10-20 20-35 35-50 50-75 75-100 100-150 150-200 200-400 400-600 600-800

Logger Data: Vehicle Level Analysis of PEVs

Table 14. PHEV Charging Summary Statistics

On Days

when the

PHEV Charged

PHEV Type

Average

Sessions/Day

Average L1

Sessions/Day

Average

Sessions/

Day

Average

kWh/Day

Average

Duration

/Day

(minutes)

Average

eVMT/Day

(miles)

Prius Plug-in 4.4 1.49 1.37 0.12 3.43 150.64 8.19

C-Max-Fusion 7.6 1.67 0.92 0.75 5.92 252.26 18.13

Prius Prime 8.8 1.39 0.98 0.41 4.83 228.47 21.12

Pacifica 16 1.57 0.59 0.98 11.72 344.68 22.77

Volt 16 1.43 0.66 0.77 9.03 375.84 30.16

Volt 18 1.29 0.53 0.76 9.80 388.75 30.27

Within the

Logging

Window

Including

Days When

PHEV Did not

Charge

PHEV Type

Average

Sessions/Day

Average L1

Sessions/Day

Average

Sessions/

Day

Average

kWh/Day

Average

Duration

/Day

(minutes)

Average

eVMT/Day

(miles)

Prius Plug-in 4.4 0.93 0.86 0.07 2.15 94.45 5.59

C-Max-Fusion 7.6 1.08 0.59 0.48 3.83 163.08 12.48

Prius Prime 8.8 0.93 0.65 0.27 3.21 151.84 15.35

Pacifica 16 1.11 0.42 0.69 8.27 243.26 17.35

Volt 16 1.01 0.46 0.54 6.36 264.50 23.00

Volt 18 0.76 0.31 0.45 5.76 228.52 21.63

Logger Data: Vehicle Level Analysis of PEVs 59

Figure 45. Share of Charging Sessions Charged Energy by Charging Level

Referring to Figure 45, L1 charging accounted for most of the Prius Plug-in-4.4, C-max/Fusion-7.6, and

Prius Prime-8.8 charging sessions and charging energy. The Volt-16 almost had an even split between L1

and L2 charging sessions and charged energy. For the Volt-18 and Pacifica-16 roughly 40% of its charging

sessions and 40% of its charged energy were using L1 charging.

Logger Data: Vehicle Level Analysis of PEVs 60

Figure 46. Share of Total Number of Sessions by Charging Level

Referring to Figure 46, C-max/Fusion-7.6, Pacifica-16, and Volt-16 had a comparable number of L1 and

L2 charging sessions on weekdays. Compared to the Volt-16, the Volt-18 had a slightly lower percentage

of L1 charging sessions on weekends and weekdays and a relatively higher percentage of L2 charging

sessions on weekends and weekdays.

Figure 47-Figure 48 show the average kWh charged per charging session and the average charging

session duration by charging level on weekdays and weekends. Except for the Pacifica-16, on average, all

PHEVs were plugged in for relatively longer times (irrespective of the charger level) on weekdays than

on weekends and subsequently the average charging energy per session was also higher on weekdays

than on weekends. Compared to other PHEVs, the Volt-18 and Pacifica-16 had relatively longer charging

sessions and higher charged energy per session (irrespective of the charger level) on weekdays and

weekends.

69.03%

42.50%

52.18%

32.95%

34.46%

31.82%

7.45%

36.10%

22.57%

41.26%

42.02%

45.20%

23.12%

11.99%

17.91%

10.70%

11.55%

9.13%

0.40%

9.41%

7.34%

15.09%

11.98%

13.85%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Prius Plug-in

4.4

C-Max-Fusion

7.6

Prius Prime

8.8

Pacifica 16 Volt 16 Volt 18

Percentage of Charging Sessions

Percentage of Sessions by Charging Level on

Weekdays and Weekends

Weekday L1 Weekday L2 Weekend L1 Weekend L2

Logger Data: Vehicle Level Analysis of PEVs

Figure 47. Average L1 and L2 Charging kWh/Session: Weekdays vs Weekends

2.3

3.8

3.6

6.5

7.8

2.5

3.5

3.2

7.8

6.5

7.8

2.2

3.3

3.5

6.6

5.8

6.5

2.2

3.3

3.0

6.7

5.5

6.9

Prius Plug-in 4.4 C-Max-Fusion 7.6 Prius Prime 8.8 Pacifica 16 Volt 16 Volt 18

Charging Energy, kWh

Average Charging kWh/Session

Weekday L1 Weekday L2 Weekend L1 Weekend L2

Logger Data: Vehicle Level Analysis of PEVs

Figure 48. Average L1 and L2 Charging Session Duration : Weekdays vs Weekends

Figure 49. Percentage of Charging Sessions Starting Time (L1 and L2): Weekdays

105

219

205

354

403

515

118

156

167

101

190

200

360

364

451

109

131

143

100

200

300

400

500

600

Prius Plug-in 4.4 C-Max-Fusion 7.6 Prius Prime 8.8 Pacifica 16 Volt 16 Volt 18

Duration, Minutes

Average Charging Session Duration

Weekday L1 Weekday L2 Weekend L1 Weekend L2

% of Weekday Sessions

Hour Beginning

% of Charging Sessions Start Time by Time of Day

(Weekdays)

Prius Plug-in 4.4 C-Max-Fusion 7.6 Prius Prime 8.8

Pacifica 16 Volt 16 Volt 18

Logger Data: Vehicle Level Analysis of PEVs 63

Figure 50. Percentage of Charging Sessions Starting Time (L1 and L2): Weekends

Figure 49 and Figure 50 show the percentage of charging sessions for each starting time on weekdays

and weekends. The percentage of charging sessions noticably spike on weekdays at around 8 am,

around noon-1pm, and between 5-7pm; and on weekends at around 1pm, 6pm-8pm and after 11 pm.

Charging Distance Based on GPS Location of PEVs

We used the survey data to analyze charging location based on self-reported information about home,

work, or public charging events. We used the logger GPS location to estimate charging location based on

a “crow’s flight” distance from the most common vehicle location at 3am while collecting data

(designated as “home” in this section), and from the over-night location before the charging. The total

number of charging events used in this section is 133,027; of those, 27,031 are out of home events

logged from 324 vehicles. Overall, 88% of the recorded level 1 charging events happened at the highest

frequency over-night location, meaning that other level 1 charging events may have happened in the

household’s other “home”, or in public locations. Similarly, 72% of the level 2 charging events occured at

the same location, and even 4% of the DC fast charging events happened within a one mile distance

from home.

Midnight

1:00 AM

8:00 AM

12:00 PM

11:00 PM

% of Weekend Sessions

Hour Beginning

% of Charging Sessions Start Time by Time of Day

(Weekends)

Prius Plug-in 4.4 C-Max-Fusion 7.6 Prius Prime 8.8

Pacifica 16 Volt 16 Volt 18

Logger Data: Vehicle Level Analysis of PEVs

Figure 51. Percentage of Charging Sessions More Than 1 Mile From Home (includes 13% of the L1

events, 29% of L2 events and 97% of DCFC events)

Share of Charges Away from Home

Distance from Home (miles)

BEV and PHEV Charges Away from Home

Level 1 Level 2 DC Fast

Logger Data: Vehicle Level Analysis of PEVs 65

Figure 52. Percentage of DCFC Charging Sessions by BEV Type and Distance from Home

As presented in Figure 51, around 25% of the level 2 and DCFC events are within 5 miles from home,

while level 1 peaked at 27% within 10-15 miles from home, most likely at the commute location. 63% of

the DCFC events are within 25 miles from home and only 8% are more than 100 miles from the main

home. Charges that are within 25 miles or more categorize 78% of level 2 events and 70% of level 1

events, respectively. 8% of level 1 events and 5% of level 2 events are more than 100 miles from home.

Figure 52 shows that more than 70% of the DCFC charging events happen within 35 miles from home for

the Leaf-24, Leaf-30, and Bolt-60 vehicles. Charging events within 35 miles from home categorize around

50% of the Model S-60_80 and 80_100 vehicles, respectively. When exploring the number of DCFC

charging events based on one way trips (using two thirds of the BEV travel range to reflect the

difference between straight lines and the road network) we conclude that 99% of the Leaf-24, 98% of

the Leaf-30, 80% of the Bolt-60 and more than 71% of the Tesla charging events are within range for a

round trip from home, if starting the day with a full battery.

Logger Data: Vehicle Level Analysis of PEVs

Figure 53. Percentage of DCFC Charging Sessions by BEV Type and Distance from Last Night Location

Using the “last night’s” location rather than the “home” location reduces the distance even more,

especially for the longest trips. Away from the start of day location, around 85% of DCFC charges were

within 25 miles from the start of day location for the Leaf-24 and Leaf-30, whereas the Bolt-60, Tesla

60_80 and 80_100 had a share of 69%, 33%, and 59% respectively. The Tesla 80-100 DCFC charging

sessions over 100 miles from home drops from 19% to 14%, most likely because of multi-day trips that

end and start on the road without an overnight charging opportunity. This method also accounts for

long vacations, summer homes, etc. that result in short trips every day but many charging events far

from home.

Share of BEV

DCFC Charges Away from Start of Day

Location

Distance from Start of Day Location (miles)

BEV DCFC Charges Away from Start of Day Location

Leaf-24 Leaf-30 Bolt-66 Model S-60_80 Model S-80_100

Household Level Analysis of PEVs

Figure 54. Percentage of Level 2 Charging Sessions by BEV Type and Distance from Last Night Location

As expected, most of the L2 events are within 1-25 miles from home, with additional smaller spikes for

Tesla 60_100 and 80_100 who travel longer trip distances. There is also a noticeable 25% share of L2

events within 25-30 miles for Bolt-60 BEVs. Overall, level 2 is being used at the destination, and

therefore most events are at work and within the vehicle range.

4 Household Level Analysis of PEVs

Self-reported trip diaries of travel behavior surveys (PSRC TCS 2006, Kunzmann and Masterman 2013,

TxDOT 2015, FHWA 2017) are often used as the starting point for generating the set of assumptions

about PEV driving and charging behavior. Instrumented ICE data has better spatio-temporal resolution

compared to trip diaries(Aviquzzaman 2014). This still cannot characterize PEV travel patterns because

of the implicit assumption that ICEs and PEVs are operated the same manner. It dilutes the risk

perception associated with modern technology adoption, especially in the case of range anxiety

associated with ZEVs. stated and revealed preferences of current PEV owners are increasingly being

used to obtain information about how current PEV owners drive and charge(Nicholas, Tal et al. 2017).

Instrumented PEVs by far are the best source of data compared to cross sectional or longitudinal survey

data of ICEs and stated or revealed preferences of existing PEV users(Nicholas, Tal et al. 2017, Raghavan

and Tal 2019). Prior research advocates the need to have realistic representation of PEV usage to

increase their usefulness to policymakers. Assuming homogenous usage of a specific PEV model across

diverse strata of demographics and travel needs, and subsequently their emission reduction potential

presents an inaccurate picture of the day-to-day substitution patterns between an ICE and PEV. Even if

high-resolution data from actual PEV usage is available, it is necessary to observe them over a longer

duration of time to capture rare and infrequent long-distance travel, which may have a bearing on the

purchase or lease and use of the vehicle.

Household Level Analysis of PEVs 68

A crucial aspect, which is often overlooked in majority of PEV usage studies in literature as well as in the

policy realm, is the household (HH) context. While evaluating travel behavior and emissions implications

of PEV adoption, household context is pivotal because day-to-day activities are allocated between PEVs

and the other vehicles in the household on a per-trip basis at disaggregated temporal levels.

Furthermore, in a survey of 15,000 PEV owners in California, roughly 45% of BEVs and 42% of PHEVs

belong to two-car households(Turrentine and Tal 2015, Nicholas, Tal et al. 2017). PEVs have unique

features that will alter how they are driven and charged compared to ICEs. Depending on travel needs,

individual driver preferences, fuel, and electricity costs, charging access and opportunities, VMT by the

PEV has cascading effects on VMT of other household vehicles. Apart from the quantity of miles, it is

also important to account for the derived impact of miles (GHG/mile) PEVs substituted at the household

level. Therefore, studying PEV usage in isolation may lead to inaccurate estimates of their net

environmental impacts, since it is based on partial information.

To ensure parity when comparing different PEV households, we excluded households that have more

than 1 PEV of the same type, irrespective of the number of ICEs in the household (for example a 3-car

household with 2-Leaf and one ICE or a 2-car household with 2 Volts were dropped). Furthermore, to

understand substitution and emission profile at the household, the sample size of the household was

limited to single PEV(BEV or PHEV), single ICE-PEV(ICE-BEV or ICE-PHEV), double ICE and single PEV (ICE-

ICE-BEV or ICE-ICE-PHEV), and household with both BEV and PHEV (BEV-PHEV or ICE-BEV-PHEV). The

above selection criteria was deemed fit because 65% of California households have 2 or less vehicles ;

16% of California households have 3 vehicles(McGuckin and Fucci 2017). Since only 9% of California

households have 4 or more vehicles (McGuckin and Fucci 2017), we excluded households and the

respective vehicles with 4 or more vehicles from our analysis. In addition, as outlined in Section 3, the

BMW i3 REX and its households were excluded because the logger could not acquire any data from

them. Out of the 364 Households that were logged, 61 households were dropped which accounted for

all the BMW i3, Soul EV, and Audi Etron households as well as any households with over three vehicles

or an extremely low share of PEV/ICE driving days.

The sample size of PEVs used in the household level analysis differs from the sample size referred in

Table 3-Table 6 simply because of excluding the households and their vehicle holdings due to household

car ownership patterns exceeding the 3 and/or the type of vehicles belonging to the household (double

BEV or PHEV of the same type). In our household (HH) level analysis, there are 117 BEVs (21 Leaf-24, 25

Leaf-30, 27 Bolt-60, 19 Tesla Model S-60_80, 22 Tesla Model S-80_100, and 3 RAV4 EV-42) and 197

PHEVs (20 Prius Plug-in-4.4, 50 C-max/Fusion-7.6, 26 Prius Prime-8.8, 26 Pacifica-16, 41 Volt-16, and 34

Volt-18). The total household level sample size is 303.

Table 15 summarizes the multi-PEV HHs with and without an ICEV. Approximately 60% of the HHs in our

study had two-vehicles, 30% had one vehicle, and 10% had three vehicles. Out of the 85 single-vehicle

HHs, 63 had a PHEV and 22 had a BEV. Referencing Figure 55-Figure 56, Of the 190 two-vehicle HHs, 111

have an ICEV and a PHEV, 72 have an ICEV and a BEV, and 7 had a BEV and PHEV. Among the 28 HHs

with three-vehicles, 12 had two ICEVs and a PHEV, 12 had two ICEVs and a BEV, and 4 had an ICE, a BEV,

and a PHEV. Overall, 96% (292 out of 303) of the HHs had only one PEV (BEV or PHEV). There were 85

single-vehicle HHs with only a BEV or a PHEV, 183 two-vehicle HHs with an ICEV and a PHEV or BEV, 24

three-vehicle HHs with a PEV and two ICEVs, and 11 multi-PEV HHs (with and without an ICEV).

Summary statistics and results presented in Table 15 – Table 24 and depicted in Figure 55 – Figure 72

are based on the logger data.

Compared to the vehicle level analysis, where we focused on the days when the PEV was driven or

charged, in the HH context, it was important to have parity in terms of the number of days each vehicle

was logged within each HH as well as across different HHs. When comparing two HHs with the same

Household Level Analysis of PEVs 69

number of vehicles and vehicle types—for example two-vehicle HHs with one ICEV and one Leaf-24—if

the first HH was logged for 350 days and the second household was logged for 400 days, at an aggregate

level, comparing the VMT and energy consumption (gasoline and electricity) between these two HHs

would be inaccurate and could potentially lead to false conclusions about PEV usage and the HH level

eVMT. It was crucial to classify the days on which we had no data (no trips or charging sessions) as

unobserved or unused to avoid over- or underestimating HH level eVMT, which depends on the VMT of

not just the PEVs but also the ICEVs.

Unobserved days typically denoted days when we knew the data logger had a problem, and the unused

days denoted days when we had no issues with the data logger and the vehicle was simply not used.

Reasons for the vehicle not being used could be that the study participant was out of

town/traveling/taking a vacation, the car was temporarily unavailable because of service/maintenance,

or there was no demand for travel on that day. Consider the same example of 2 HHs each having an ICEV

and a Leaf-24. If the ICEV in one HH had data logger issues for a few weeks, and we had data from the

BEVs during the period, if we incorrectly assume the ICEV was not used, then eVMT will be

overestimated.

We used the days the individual vehicles (ICEV, BEV, PHEV) were logged (used and unused days) to

annualize all the key metrics (trips, charging sessions, VMT, driving/charging energy, gasoline

consumed).

The HH level analysis section is organized as follows: we present first the results from BEV HHs (only a

BEV; an ICEV and BEV; and two ICEVs and a BEV) and then the results from PHEV HHs (only a PHEV; an

ICEV and PHEV; and two ICEV and a PHEV). Finally, since only 4% of the HHs (11 HHs in total) in our study

had both a BEV and PHEV (with or without an ICEV), and 9 of these 11 HHs did not have any of the same

types of BEVs and PEVs, their results are presented separately and are not analyzed as a group. Table 15

shows the breakdown of double PEV HHs.

Figure 55. Composition of Households Included in the Analysis

111

Households Logged by Vehicle Types

ICE-PHEV ICE-BEV PHEV BEV 2ICE-PHEV 2ICE-BEV BEV-PHEV ICE-BEV-PHEV

Household Level Analysis of PEVs

Figure 56. Number of Households with only One BEV or PHEV Logged in the Study (Upper) and

Number of Two Car Households with One ICEV and One BEV or PHEV Logged (Lower)

Table 15. Double-PEV (1 BEV and 1 PHEV) With or Without an ICEV (N=11)

Type of HH BEV, PEV in the HH Number of HHs

ICE-BEV-PHEV RAV4 EV-42, C-Max/Fusion-7.6 1

Household Level Analysis of PEVs 71

Type of HH BEV, PEV in the HH Number of HHs

ICE-BEV-PHEV Leaf-24, C-Max/Fusion-7.6 2

ICE-BEV-PHEV RAV4 EV-42, Volt-16 1

BEV-PHEV Model S-60_80, Volt-16 1

BEV-PHEV Leaf-30, Volt-18 1

BEV-PHEV Leaf-24, C-Max/Fusion-7.6 1

BEV-PHEV Model S-60_80, Pacifica-16 1

BEV-PHEV Bolt-60, Prius Prime-8.8 1

BEV-PHEV Leaf-24, Volt-16 1

BEV-PHEV RAV4 EV-42, Prius Plug-in-4.4 1

Households with a BEV Only or BEV and ICEV

Table 16 summarizes the (average) annualized estimates of key metrics such as eVMT, gVMT, HH VMT,

UF, and energy consumption (driving and charging). Figure 57 depicts the HH UF in BEV HHs by number

of vehicles in the HH and the type of BEV in the HH.

Table 16. (Average) Annualized Estimates of VMT and Energy Consumption in BEV HHs

HH Type

Num.

HHs

BEV

Trips

BEV

eVMT

BEV kWh

Driving

ICEV

gVMT

ICEV Fuel

(gallons)

HH VMT HH UF

2-ICEV-BEV 2 L24 1191 12287 3000 18610 722 30897 0.41

4 L30 1425 12982 3400 20682 891 33664 0.39

1 B60 1395 22917 5285 16316 888 39233 0.58

2 T60 1187 22276 6918 8900 395 31176 0.71

3 T80 405 10483 3462 22457 1024 32940 0.33

ICEV-BEV 13 L24 1483 10326 2395 10908 451 21233 0.49

15 L30 1456 12352 3218 8689 335 21040 0.57

22 B60 1438 13399 3338 9530 387 22929 0.58

Household Level Analysis of PEVs

HH Type

Num.

HHs

BEV

Trips

BEV

eVMT

BEV kWh

Driving

ICEV

gVMT

ICEV Fuel

(gallons)

HH VMT HH UF

12 T60 984 17716 6000 7885 393 25601 0.65

10 T80 916 14797 5118 7154 334 21951 0.67

BEV 2 L24 1235 7955 1806 0 0 7955 1

5 L30 1137 7169 1892 0 0 7169 1

3 B60 1336 13489 2500 0 0 13489 1

3 T60 1085 9797 3322 0 0 9797 1

9 T80 1054 13639 4889 0 0 13639 1

Figure 57. Household Utility Factor by BEV Type and Household Car Composition

Some of the key insights regarding the HH level UF of HHs with ICEVs and BEVs are as follows:

• The average UF of HHs with a Bolt-60 with either one or two ICEVs were relatively similar (0.578

vs. 0.584).

Household Level Analysis of PEVs 73

• On average, among two-vehicle HHs and three vehicle HHs, HHs with a Model S-60_80 and two

ICEVs had the highest UF, whereas HHs with a Model S-80_100 and two ICEVs, had the lowest

UF.

• The UF in ICEV-BEV HHs increased with the battery capacity.

• HHs with a Bolt-60 tended to have a higher average daily HH VMT relative to other BEV types in

HHs with either one ICE or two ICEs.

Figure 58. HH Average Daily VMT in HHs with BEVs, Showing the eVMT and gVMT Percentages

Figure 58 shows the average daily HH VMT and the share of eVMT and gVMT in BEV HHs. Three-vehicle

HHs with two ICEVs and one BEV, apart from having a Leaf-24, had higher average daily HH VMT than

did three-vehicle HHs with one Model S-60_80 BEV. HHs with two ICEVs and one Leaf had nearly two-

thirds of their VMT attributable to gasoline-power (gVMT). In two-vehicle HHs, there is a trend of higher

eVMT as battery size increses.

Household Level Analysis of PEVs 74

Figure 59. Percentage of BEV and ICEV trips in HHs with BEVs

Figure 59 summarizes the percentage of HH trips taken using the BEV and the ICEVs. In three-vehicle

HHs, the household with two ICEVs and one Bolt-60 had the highest percentage of BEV trips (55%)

whereas HHs with two ICEVs and one Model S-80_100 had the lowest percentage of BEV trips (21%). On

average, the BEV share of HH trips in two-vehicle HHs was approximately 60% for HHs with one ICEV and

either one Leaf or Bolt-60 and approximately 50% for HHs with one ICEV and one Tesla. Figure 58 and

Table 16 together show that, in two-vehicle BEV HHs, roughly 50% of HH trips were taken using the

ICEV, but the share of miles replaced by the BEV is noticably different between HHs with a smaller

battery capacity BEV and a larger battery capacity BEV. The percentage of total HH VMT driven using the

ICEV in two-car BEV HHs was 52%, 43%, 41%, 33%, and 33% in Leaf-24, Leaf-30, Bolt-60, Model S-60_80

and Model S-80_100 HHs.

Household Level Analysis of PEVs 75

Figure 60. Number of Days/Year BEV was Used for Long Distance Travel (LDT). LDTx: 50-300 or More

Miles/Day

Figure 60 shows the absolute number of days per year BEV was used for LDT. In the three-vehicle

households, the Bolt-60 was the most used BEV for LDT50, while for the two-vehicle households it was

the Model S-80_100, and for single-BEV households it was the Bolt-60. However, the Model S-60_80

was the most used BEV among three-vehicle and two-vehicle households for trips greater than 100 and

200 miles. LDT50 days for Leaf-30 were greater than those for Leaf-24 in three-vehicle and two-vehicle

HHs, but the contrary was observed in HHs with just one BEV.

Household Level Analysis of PEVs 76

Figure 61. (Average) Annualized Number of L1 or L2 charger (L1/L2), and DCFC Sessions in BEV HHs

Figure 61 shows the average annualized number of L1 or L2 charger and DCFC sessions by BEV type and

number of cars in the HH. The Model S-60_80 in three-vehicle HHs used DCFCs the most (103 times per

year) followed by the Leaf-30 in two-vehicle HHs (87 times per year). In two-vehicle ICE-BEV(Tesla)

households, the BEV was used for commuting in 10 out of the 14 Model S-60_80 cases, and 7 out of the

18 Model S-80_100s cases.

Households with a PHEV Only or PHEV and ICEV

Analyses and results summarized in Table 17–Table 21 and shown in Figure 62 – Figure 66 are based on

the logged data. Households with a PHEV only or a PHEV and ICEV have no range limitation on their

trips, but have lower potential for eVMT. Table 17– Table 19 summarizes the average annualized

estimates of VMT and energy consumption by number of vehicles in the HH and the PHEV type. As a

reminder, we recruited only PHEV households that plugged-in their vehicle, which complicates the

interpretation of our results in this section.

Household Level Analysis of PEVs

Table 17. (Average) Annualized Estimates of VMT and Energy Consumption in PHEV HHs by Number of

Cars in HH and PHEV Type

Type

PHEV

Model

PHEV

Trip

Total

PHEV

Only

Trips

PHEV

CDB/

Trips

PHEV

Only

Trips

PHEV

VMT

Total

PHEV

eVMT

PHEV

gVMT

PHEV

Fuel

(gal)

PHEV

kWh

ICEV

Trips

ICEV

VMT

ICEV

Fuel

(Gal.)

2ICEV

-PHEV

Prius Plug-

in-4.4

1327 38 642 647 13685 1671 12015 248 -482 2159 28104 958

C-max/

Fusion-7.6

1269 426 303 541 13353 3463 9890 239 -1058 1930 18283 765

Volt-16 976 846 70 60 8610 7228 1382 39 -2077 1941 18762 678

ICE-

PHEV

Prius Plug-

in-4.4

1383 171 745 467 14779 2040 12739 266 -571 906 6422 288

C-max/

Fusion-7.6

1385 470 464 450 14475 4769 9705 245 -1423 1182 9955 424

Prius

Prime-8.8

1463 704 356 391 14077 5589 8488 162 -1259 1049 8096 368

Pacifica-16 1485 811 339 314 11599 6295 5304 179 -2464 976 7071 279

Volt-16 1206 855 210 141 15073 9705 5369 157 -2641 889 7961 350

Volt-18 1238 980 191 68 11264 7731 3533 95 -2221 1021 8632 356

PHEV Prius Plug-

in-4.4

1298 360 647 291 10851 2354 8497 168 -623 0 0 0

C-max/

Fusion-7.6

1303 548 400 355 12029 4061 7968 195 -1281 0 0 0

Prius

Prime-8.8

1399 1048 162 189 9754 5706 4049 78 -1268 0 0 0

Pacifica-16 1707 969 456 262 13734 6004 7730 250 -2412 0 0 0

Volt -16 1249 977 104 169 10744 6322 4422 129 -1815 0 0 0

Volt -18 1330 941 284 105 12103 7285 4818 128 -2006 0 0 0

Household Level Analysis of PEVs 78

Table 17 shows that in three-car HHs with two ICEVs and one PHEV, the total HH VMT decreased and

the average HH UF increased with an increase in the AER of the PHEV. The percentage of HH miles

driven in ICEVs was the highest in Volt-16 HHs (69%), followed by Prius Plug-in-4.4 HHs (67%), and C-

max/Fusion-7.6 HHs (58%). Annual mileage of ICEVs in C-max/Fusion-7.6 and Volt-16 HHs was almost

the same (18,283 and 18,762 miles), but the number of trips and the VMT of the C-max/Fusion-7.6 was

higher than those of the Volt-16. Table 18 collapses the data presented in Table 17 by household fleet

type and utility factor.

In two-car HHs, the percentage of total HH VMT driven using the PHEV was roughly the same in Prius

Prime-8.8 and Pacifica HHs (63% and 62%). The average HH UF was higher in Pacifica HHs than in Prius

Prime-8.8 HHs (0.35 vs. 0.28) compared to the vehicle’s measured utility factor (0.47 vs. 0.51), but the

Prius Prime-8.8 had about 2500 miles more than the Pacifica-16 (14,077 miles vs. 11,599 miles), had a

lower share of ZE only trips (28% vs. 33%), and charged less often (316 sessions vs. 390 sessions). The

Volt-18 has a slightly bigger battery than the Volt 16 and its eVMT was greater than that of the Volt-16

(9,705 miles vs. 7,731 miles), within two-car households. The average household UF did not improve by

upgrading from a Volt-16 to a Volt-18 within ICE-PHEV households (0.44 vs. 0.41) or within 1-car single

PHEV households (0.62 vs. 0.60). In 2 car HHs, as compared to 1-car single PHEV HHs, the Volt-18 had a

lower UF, as did all other PHEV types. To better understand these aspects, we looked at certain key HH

level attributes reported by the respondent in the survey; our observations are summarized below for

the 2 car (ICE-PHEV) and 1 car (single PHEV) HHs separately.

Volt-16 and Volt-18 in 2-car HHs:

Out of the 22 Volt-16 HHs (ICE-Volt-16 HHs), only 1 was leased, whereas out of the 19 Volt-18 HHs (ICE-

Volt-18 HHs) 13 of them were leased. 21 of the 22 Volt-16 HHs reported that they either charged at

home only, or home and away in the past 30 days. Out of the 19 Volt-18 HHs, 15 of them reported that

they either charged at home only, or home and away in the past 30 days. Only 1 of the Volt-16 HHs

reported that they charged away only, whereas this number was slightly higher in Volt-18 HHs, where 4

of them reported that they charged away only. The average number of drivers in both the Volt-16 and

Volt-18 HHs was comparable (2.1 vs. 2). The average HH size of Volt-18 HHs was slightly higher (3)

compared to the average HH size of Volt-16 HHs (2.36). 70% of the Volt-16 (16 out of the 22) and 90% of

the Volt-18 (17 out of the 19) were used by HH members working full-time for commuting purposes.

Despite the longer range of the Volt-18 compared to the Volt-16, the ICE was probably used more often

due to the HH size. Since the Volt-16 was the first ever mass-produced series type PHEV, higher annual

VMT of Volt-16 in two car HHs could also be because these were driven by early adopter technology

enthusiasts who were also innovators. Furthermore, the smaller HH size and lower share of Volt-16

being leased and charging exclusively away as compared to Volt-18 are the other reasons for the

difference in usage between Volt-16 and Volt-18 in 2 car HHs (ICE-PHEV HHs).

Volt-16 and Volt-18 in 1-car HHs:

9 out of 12 Volt-16 were purchased, whereas only 5 of out of 14 Volt-18 were purchased. The higher

annual VMT of Volt-18 compared to Volt-16 can be primarily attributed to the higher share of drivers in

Volt-18 HHs who used it for commute purposes.

Household Level Analysis of PEVs 79

Table 18. (Average) Annualized Estimates of PHEV VMT, HH VMT, and HH UF

HH Type PHEV Type

Number of

HHs

PHEV eVMT HH VMT UF

2ICEV-

PHEV

Prius Plug-in-4.4 2 1671 41789 0.04

C-max/Fusion-7.6 6 3463 31635 0.12

Volt-16 4 7228 27373 0.27

ICE-PHEV Prius Plug-in-4.4 12 2040 21201 0.11

C-max/Fusion-7.6 24 4769 24430 0.22

Prius Prime-8.8 17 5589 22173 0.28

Pacifica-16 18 6295 18670 0.35

Volt-16 22 9705 23034 0.44

Volt-18 18 7731 19897 0.41

PHEV Prius Plug-in-4.4 5 2354 10851 0.25

C-max/Fusion-7.6 16 4061 12029 0.36

Prius Prime-8.8 8 5706 9754 0.62

Pacifica-16 7 6004 13734 0.47

Volt-16 12 6322 10744 0.62

Volt-18 15 7285 12103 0.60

Table 19 summarizes the average annualized estimates of PHEV charging needs by number of cars in the

HH and the PHEV type and Table 20 summarizes the average daily estimates of PHEV charging by PHEV

type.

Household Level Analysis of PEVs

Table 19. (Average) Annualized Estimates of Number of Charging Sessions and kWh Charged in PHEV

HHs by Number of Vehicles and PHEV Type

HH Type PHEV Type

Annualized

Charging

Sessions

Annualized

Charging kWh

Average Charging

Duration/Session

(minutes)

Average kWh/

Session

2ICEV-PHEV Prius Plug-in-4.4 272 658 112 3

C-max /Fusion-7.6 218 956 270 5

Volt-16 379 2085 223 7

ICE-PHEV Prius Plug-in-4.4 350 812 118 3

C-max /Fusion-7.6 401 1458 204 4

Prius Prime-8.8 316 1168 217 4

Pacifica-16 390 3023 323 10

Volt-16 393 2612 331 8

Volt-18 287 2131 281 8

PHEV Prius Plug-in-4.4 410 874 105 2

C-max /Fusion-7.6 380 1309 178 4

Prius Prime-8.8 410 1213 159 3

Pacifica-16 541 2997 109 8

Volt-16 328 1784 326 6

Volt-18 249 1962 344 9

Household Level Analysis of PEVs 81

Table 20. (Average) Annualized Estimates of Charging Sessions by PHEV Type in PHEV HHs

Average Annual PHEV

Charging

Sessions/Year

Charging Energy

kWh/Year

Prius Plug-in-4.4 357 812

C-max/Fusion-7.6 370 1341

Prius Prime-8.8 346 1183

Pacifica-16 433 3016

Volt-16 371 2295

Volt-18 270 2054

Average Daily PHEV

Charging

Sessions/Day

kWh/Day

Prius Plug-in-4.4 0.984 2.23

C-max/Fusion-7.6 1.040 3.73

Prius Prime-8.8 0.950 3.24

Pacifica-16 1.108 8.28

Volt-16 1.018 6.32

Volt-18 0.740 5.63

Figure 62–Figure 64 depict the HH UF from four different perspectives calculated using the logged data:

individual HH level UF by number of vehicles in the HH and PHEV type, average HH UF by PHEV type, and

average HH UF by number of vehicles in the HH and PHEV type.

Household Level Analysis of PEVs

Figure 62. Individual HH UF by PHEV Type in PHEV HH

Figure 63. Average HH UF by P

HEV Type (Left: all HHs); and Average HH UF by Number of Cars in the

HH (Right: all PHEVs)

Household Level Analysis of PEVs 83

Figure 64. Average HH UF by Number of Cars per HH and PHEV Type

Table 21. Average Utility Factor (UF) of PHEVs by Model Year (MY) According to the EPA Dataset

PHEV MY

City EPA Fuel

Economy UF

Highway

EPA Fuel

Economy UF

Combined

EPA Fuel

Economy UF

CARB

Midterm

Report

Prius Plug-in-4.4 2012-2014 0.320 0.250 0.290 0.15

C-max/Fusion-7.6 2013-2017 0.481 0.421 0.455 0.32

Prius Prime-8.8 2017 0.553 0.498 0.529 N/A

Pacifica-16 2017-2018 0.640 0.586 0.617 N/A

Volt-16 2011-2015 0.664 0.642 0.652 0.6

Volt-18 2016-2017 0.778 0.737 0.761 0.6

Table 21 shows the average UF of PHEVs by different model years that are in the logged vehicle dataset

from the EPA. The UF based on the CARB Midterm Review (CARB 2017b, 2017a) is also added to Table

21 for comparison purposes. Overall, the EPA UFs are higher than the CARB Midterm Review (MTR) UFs

0.04

0.11

0.25

0.12

0.22

0.36

0.28

0.62

0.35

0.47

0.27

0.44

0.62

0.41

0.60

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

2ICEV-PHEV

ICE-PHEV

PHEV

2ICEV-PHEV

ICE-PHEV

PHEV

ICE-PHEV

PHEV

ICE-PHEV

PHEV

2ICEV-PHEV

ICE-PHEV

PHEV

ICE-PHEV

PHEV

Prius Plug-in C-max Fusion Prius Prime Pacifica 16 Volt 16 Volt 18

Average HH Utility Factor (UF)

Household Level Analysis of PEVs 84

and the UC Davis values calculated from the logger data. UFs of logged PHEVs from single PHEV HHs are

closer to the CARB MTR UFs except in the case of the Prius Plug-in-4.4 UF.

The interpretation of UFs varies noticeably by level of aggregation (vehicle or household level) and the

number of vehicles in the household. In addition, the marginal improvements in upgrading from Volt-16

to Volt-18 were negligible in one-car HHs and two-car HHs. If we ignore the context of the HH (Figure 63,

left), we find that the fleet average UFs of PHEVs in our study were lower than the CARB MTR UFs by

0.01-0.12, depending on the PHEV type.

Figure 65. Percentage of Household Trips Powered by Different PHEV Driving Modes or ICEVs

13%

29%

18%

28%

33%

41%

43%

28%

42%

75%

57%

78%

71%

18%

33%

18%

14%

10%

50%

31%

12%

27%

21%

19%

17%

20%

18%

16%

13%

22%

27%

14%

16%

14%

62%

60%

67%

40%

46%

42%

40%

42%

45%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Prius Plug-in

Cmax

Volt16

Prius Plug-in

Cmax

Prius Prime

Pacifica

Volt 16

Volt 18

Prius Plug-in

Cmax

Prius Prime

Pacifica

Volt 16

Volt 18

2ICEV-PHEV ICE-PHEV PHEV

% of Household Trips

ZE Only CDB/CS CS Only ICE Trips

Household Level Analysis of PEVs 85

Figure 66. Daily Average HH VMT and Percentage Share of PHEV eVMT, PHEV gVMT, and ICE gVMT

Figure 66 shows the average daily HH VMT and percentage share of eVMT and gVMT. This figure

demonstrates that the average daily HH VMT of C-max/Fusion-7.6 HHs did not change much between

three car HHs and two car HHs. Daily eVMT of Pacifica-16 was similar in two car and single car HHs

(within 10 miles). On average, the daily HH VMT of Volt-16 HHs was lower than that of C-max/Fusion-7.6

HHs in one car, two car, and three car HHs.

Two-PEV Households: BEV and PHEV Mix

In the following section we present data from households with two PEVs. The sample size is only 11

households and, even though the total number of days and miles is high, the analysis cannot be

generalized to the population of PEV users. Analyses and results presented in Table 22– Table 24 and

depicted in Figure 67– Figure 69 are based on the logger data.

Table 22. Double-PEV (1 BEV and 1 PHEV) HHs With or Without ICEV(s) (N=9)

Type of HH BEV, PHEV in the HH Number of HHs

ICE-BEV-PHEV Leaf-24-C-Max/Fusion-7.6 2

ICE-BEV-PHEV RAV4 EV-42-Volt-16 1

ICE-BEV-PHEV RAV4 EV-42-C-Max/Fusion-7.6 1

100

120

140

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Prius Plug-in

C-max Fusion

Volt 16

Prius Plug-in

C-max Fusion

Prius Prime

Pacifica 16

Volt 16

Volt 18

Prius Plug-in

C-max Fusion

Prius Prime

Pacifica 16

Volt 16

Volt 18

2ICEV-PHEV ICE-PHEV PHEV

Daily Average HH VMT, Miles

Percentage Share by Vehicle

Avg. PHEV eVMT Avg. PHEV gVMT Avg. ICE gVMT Avg. HH VMT

Household Level Analysis of PEVs 86

Type of HH BEV, PHEV in the HH Number of HHs

BEV-PHEV Leaf-24-C-Max/Fusion-7.6 1

BEV-PHEV Model S-60_80-Pacifica-16 1

BEV-PHEV Bolt-60-Prius Prime-8.8 1

BEV-PHEV RAV4 EV-42-Prius Plug-in-4.4 1

BEV-PHEV Leaf-24-Volt-16 1

BEV-PHEV Model S-60_80-Volt-16 1

BEV-PHEV Leaf-30-Volt-18 1

Households with a BEV and a PHEV

As shown in Figure 67 and Figure 68, the average household VMT of Leaf-30-Volt-18 HHs was lowest but

had the highest UF compared to all other BEV-PHEV HHs.

The average annualized estimates of other metrics in BEV-PHEV HHs are summarized below in Table 23

and Table 24.

Household Level Analysis of PEVs

Figure 67. Daily Average HH VMT, and Percentage of eVMT and gVMT BEV-PHEV Households

59.1

45.0

41.7

50.7

42.1

57.7

53.3

21.2

22.3

31.3

6.7

27.3

25.3

44.6

19.8

32.7

27.1

42.6

30.6

17.0

2.2

59.58

56.91

58.76

72.28

87.66

97.39

28.22

100

120

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Household VMT (Miles)

eVMT Share of Household VMT (%)

BEV eVMT % PHEV eVMT % PHEV gVMT % HH VMT (Miles)

Household Level Analysis of PEVs

Figure 68. Average Annual HH VMT and Share of BEV eVMT and PHEV eVMT in BEV-PHEV HHs

Table 23. Annualized Driving Metrics in BEV/PHEV HHs

BEV-PHEV

PHEV

eVMT

BEV

eVMT

PHEV

gVMT

HH VMT HH UF

PHEV

Fuel

(gal)

PHEV

Driving

Energy

(kWh)

BEV

Driving

Energy

(kWh)

Leaf-24-CMax/Fusion-7.6 4601 12851 4296 21748 0.802 113 -1497 -3317

Model S-60_80-Pacifica-16 4632 9337 6801 20771 0.673 86 -704 -1128

Bolt-60-Prius Prime-8.8 6706 8937 5806 21449 0.729 78 -1269 -2089

RAV4 EV-42-Prius Plug in-4.4 1772 13368 11242 26382 0.574 258 -585 -5379

Leaf-24-Volt-16 8741 13454 9801 31996 0.694 261 -2608 -3350

Model S-60_80-Volt-16 8985 20507 6054 35546 0.830 189 -2806 -6342

Leaf-30-Volt-18 4592 5485 224 10300 0.978 7 -1496 -1674

0.59

0.45

0.42

0.51

0.42

0.58

0.53

0.21

0.22

0.31

0.07

0.27

0.25

0.45

5000

10000

15000

20000

25000

30000

35000

40000

0.00

0.20

0.40

0.60

0.80

1.00

1.20

Average Annual HH VMT

BEV/PHEV eVMT Share

BEV eVMT share PHEV eVMT share HH VMT

Household Level Analysis of PEVs

Table 24. Annualized Charging Metrics in BEV/PHEV HHs

BEV-PHEV

PHEV Total

Sessions

PHEV Total

Charging

kWh

BEV Total

Sessions

BEV Total

Charging

kWh

BEV DCFC

Sessions

Leaf-24-CMaxFusion 470 1620 484 3217 0.0

Model S 60_80-Pacifica-16 257 1525 226 3288 2.6

Bolt-60-Prime 321 1381 133 2537 0.0

RAV4 EV-42-Prius Plug-in-4.4 278 749 343 4823 0.0

Leaf-24-Volt-16 393 2647 381 3346 4.0

Model S 60_80-Volt-16 361 2858 275 6832 33.0

Leaf-30-Volt-18 230 1372 183 1276 0.0

UF and GHG Profile

In this section, we focus on the utility factors and GHG emissions of the PEVs and the household fleets

using the logger data. Of the total 303 HHs, we use a subset of HHs with one PEV and one ICE, exclude

households with one vehicle and households with more than two vehicles, and we analyze the

disparities in vehicle level and household level UF and GHG emissions. The UF and GHG profile of the

183 two car households (72 HHs with ICE-BEV and 111 HHs with ICE-PHEV) is analyzed using the average

annualized estimates of the relevant PEV usage metrics by PEV type covered in Sections 4.1-4.2. For

parity purposes, we restrict this analysis to only two car HHs with single ICE and single PHEV or BEV.

Since the logged HHs are PEV early adopter HHs, the ICEs in these HHs are not representative of general

population of ICE owners. The well to wheel emissions factors for electricity and gasoline are 378.54

gCO2e/kWh and 11405.85 gCO2e/Gallon of gasoline.(CARB 2017c)

Household Level Analysis of PEVs 90

Figure 69. Two car HHs VMT by Vehicle Type, PEV UF and HH UF

Figure 69 presents the VMT of ICE-PEV HHs by PEV type, fuel source, PEV UF, and HH UF. The total

annual miles of these households range between 18,700 for the Pacifica-16 and 25,600 for the Tesla

Model S-60_80, but, apart from Volt-18 HHs, the HH utility factor is always increasing with the PEV

range. For short range PHEVs, the household utility factor is just over half of the PEV utility factor. For

the Volts, the PHEVs electrify about 40% of the household miles. The longer range BEVs electrify around

60% to 70% of the household VMT partly because of lower miles for the ICEV in long-range Tesla HHs.

Figure 70 presents the average GHG per mile for the PEVs in the studied fleet. As expected, the short-

range BEVs have the best performance followed by the larger battery capacity BEVs and the PHEVs. We

see that the relatively gas-efficient engine on PHEVs results in GHG emissions not much higher than

larger battery vehicles. The results are based on average electricity derived GHG described above and on

the logged travel behavior.

Household Level Analysis of PEVs 91

Figure 70. Average GHG per Mile and Utility Factor

Figure 71 has the household level (PEV+ICEV households) GHG sources in comparison to the GHG per

mile from energy (gas and electricity) consumed by the PEVs and from gasoline consumed by the ICEV in

two vehicle households. We also include the household utility factor (HH eVMT/VMT).

Household Level Analysis of PEVs 92

Figure 71. Household Level GHG and Utility Factor per Mile

The actual performance of each household depends on the metric considered. At the household level,

the total VMT and ICE VMT substituted with PEV eVMT are the major determinants of the HH UF. From

an emissions perspective, in addition to the aforementioned factors, it is also important to account for

not just the quantity of ICE VMT substituted but also the quality. The disparities in HH GHG/mile

between HHs with different PEVs is therefore influenced by (1) energy and carbon intensity of ICE; (2)

usage intensity of the ICE (absolute VMT); (3) energy intensity (kWh/mile) of the PEV and its charging

related emissions; (4) battery capacity of the PEV which directly impacts the eVMT; (5) CDB or CS mode

miles and gasoline consumption in PHEV HHs. Figure 69, Figure 71, and Figure 72, when analyzed

together, present a complete picture of HH level emission impacts of PEVs for a two-vehicle household.

The mean ICE gVMT in Tesla HHs is between 7000-8000 miles, whereas in the Leaf HHs, it is between

8000-11000 miles. Leaf-30 HHs have lower HH GHG/mile compared to Leaf-24 HHs because of its bigger

battery. The incremental eVMT enabled is also due to the bigger battery of the Leaf-30, which

overcompensates for the fact that ICEs in Leaf-30 HHs are less efficient compared to the ones in Leaf-24

HHs (445 gCO2e/mile compared to 415 gCO2e/mile).

Household Level Analysis of PEVs

Figure 72. Ratio of PEV and ICE GHG/Mile to Total HH GHG/Mile

The ICEs in Model S-60_80 HHs are the most inefficient (599 gCO2e/mile) but have the fourth lowest ICE

usage intensity on an absolute VMT basis, and thereby the second highest HH UF. However, on a per

mile HH GHG/mile it performs best among the rest of PEV types simply because of its lower ICE usage

intensity. In contrast, the ICE in Model S-60_80 HHs has a higher usage intensity on an absolute VMT

basis when compared to the ICE in Tesla Model S-80_100 HHs. The ICE GHG/mile in Tesla Model S-

80_100 HHs was only 6% lower (561 gCO2e/mile vs. 599 gCO2e/mile) when compared to ICE in Tesla

Model S-60_80 HHs, but the usage intensity of ICE in Model S-60_80 HHs was 10% higher than those in

Model S-80_100 HHs (7885 miles vs 7154 miles).

The Leaf HHs on the other hand have a higher ICE usage intensity compared to the Model S HHs on an

absolute VMT basis, which is the reason for Leaf HHs having lower UF compared to the UF of Model S

HHs. On average, the ICEs in Bolt-60 HHs are about 15% more efficient than the ICEs in the Model S HHs

(both 60-80 and 80-100 kWh) on a gCO2e/mile basis and this causes the overall HH GHG/mile in Bolt-60

HHs to be lower than that of the Model S HHs.

Three factors cumulatively work in favor of the Bolt-60 HH to have the lowest HH GHG/mile compared

to other BEV HHs: lower ICE usage intensity compared to Leaf-24 HHs, lower energy intensity of the PEV

and carbon intensity of the ICE compared to Model S HHs. When we look at the PHEV HHs, Volt-16 HHs

have the lowest HH GHG/mile. Though the UF of Volt-16 HH was similar to that of the Volt-18 HH, the

Volt-18 HH GHG/mile is higher. This is because the ICEs in Volt-18 HHs have higher usage intensity.

Referring to Fig. 83, we can clearly see that on a GHG/mile basis, the only distinguishing aspect between

Household Level Analysis of PEVs 94

Volt-16 and Volt-18 HHs is the ICE usage intensity. Prius Prime-8.8 HHs have the highest ICE usage

intensity among PHEV HHs and C-Max/Fusion-7.6 HHs have the highest HH GHG/mile across all PEV HHs.

The HH GHG/mile (blackline in the middle of Figure 71) shifts upwards if the ICE usage intensity and ICE

carbon intensity increases. As far as determining how the curve would shift, we must consider the

carbon, energy, and usage intensity of the ICE and PEVs. If we ignore the specific ICE class/segment

(compact, SUV, sedan etc.), ICE carbon intensity increases with the AER in BEV HHs. The reverse of this

trend can be observed as we move (left to right) from C-max/Fusion-7.6 HHs up to Volt-18 HHs. Long-

range BEV HHs (Model S HHs) on average have higher emissions from the ICE on a per-mile basis

compared to all other PEVs.

Additional ICE Usage Metrics

We briefly summarize usage metrics of the ICE in PEV HHs using the average annualized estimates

summarized in Table 16 - Table 24 based on the logger data. For clarity, the ICE usage summaries of 2

car HHs and 3 car HHs are presented separately. In the case of 3 car HHs (2 ICEs and 1 BEV or PHEV), the

total ICE VMT is considered. Due to a low sample size of HHs with single ICE and more than 1 PEV (4

HHs) and HHs with two PEVs (7 BEV-PHEV HHs), we have excluded them, therefore the sub-sample of

HHs considered is 207. Overall, among the logged households, the average fuel economy of the ICE was

23.89 mpg and the average model year was 2010.

Average Annual ICE VMT

Figure 69 depicts the average annualized ICE VMT in 2 car HHs with one PEV and one ICE. In most cases,

the ICE vehicle in a mixed PEV-ICEV household is used less than the average vehicle in California and less

than the plug-in vehicle in the household.

Referring to Figure 73, we notice a steady decline in the annual ICE VMT in BEV HHs with increase in

range/battery capacity in 2 car HHs, except for 2 car Leaf-30 HHs. In the case of PHEV HHs, the annual

ICE VMT exhibited a variation. ICEs in Prius Plug-in-4.4 HHs drove the least among the 2 car PHEV HHs

and was even lower than the ICE VMT in 2 car Model S-80_100 HHs.

Household Level Analysis of PEVs

Figure 73. Average Annualized ICE VMT in 2 Car HHs (Single ICE and Single PHEV or BEV) by PEV Type.

N=183 HHs

Household Level Analysis of PEVs 96

Figure 74. Average Annualized ICE VMT in 3 Car HHs (Two ICEs and Single PHEV or BEV) by PEV Type.

N=24 HHs.

Referring to Figure 74, there was considerable variation in annual ICE VMT of 3 car HHs across all PEV

types. The ICEs in 3 car Prius Plug-in-4.4 HHs had the highest annual ICE VMT followed by Model S-

80_100 HHs and the ICE VMT in Leaf-30 HHs.

ICE Usage for Long Distance Travel (LDT)

We characterized long distance travel (LDT) using two daily VMT thresholds, 50 miles and 100 miles

(LDT50 and LDT100). Figure 75 and Figure 76 depict the average annualized number of days per year the

PEV and ICE was used for LDT50 and LDT100 in 2 car PHEV and 2 car BEV HHs, respectively. The ICE

share (%) of total HH LDT50/100 days is shown using the secondary Y axis in Figure 75 and Figure 76.

Referring to Figure 75, in 2 car ICE-PHEV HHs, the ICE share of total HH LDT50 days per year was lowest

for Volt-16 HHs (22%), whereas the ICE share of total HH LDT100 days per year was lowest for Prius Plug

in 4.4 HHs (27%). The ICE usage for LDT50 and LDT100 in Prius Plug-in-4.4 HH were comparable on a

percentage share of the HH LDT50(100) days/year and a similar trend was observed in Volt-18 HHs and

Pacifica-16 HHs. The ICE in 2 car Volt-16 HHs was used roughly on 15% more days for LDT100(38%)

compared to LDT50(22%).

Household Level Analysis of PEVs

Figure 75. PHEV and ICE Use (Days/Year) for Long Distance Travel 50(100) Miles or More in 2 Car HHs

(Single ICE and Single PHEV); ICE Share (%) Total HH LDT50(100) days/year shown on the secondary Y

axis. N=77 HHs.

Household Level Analysis of PEVs 98

Figure 76. BEV and ICE Use (Days/Year) for Long Distance Travel 50(100) in 2 Car HHs (Single ICE and

Single BEV); ICE Share (%) of Total HH LDT (50/100) days/year shown on the secondary Y axis. N=72

HHs.

Referring to Figure 76, in 2 car ICE-BEV HHs, we notice a clear trend in decreasing ICE usage for LDT with

an increase in the range of the BEV, and this effect is more pronounced in the case of LDT100. There was

only a 6% reduction in ICE usage for LDT50 in Bolt-60 HHs (33%) compared to Leaf-30 HHs (39%).

However, the reduction in ICE usage for LDT100 was greater in Tesla Model S HHs compared to Leaf

HHs. Overall, on an absolute days/year basis, Leaf-24 HHs had the least number of LDT50 and LDT100

days compared to all other BEVs.

Household Level Analysis of PEVs 99

Figure 77. PHEV and ICE Use (Days/Year) for Long Distance Travel 50(100) Miles or More in 3 Car HHs

(Two ICEs and Single PHEV); ICE Share (%) of Total HH LDT50(100) days/year shown on the secondary

Y axis. N=12 HHs.

Figure 77 and Figure 78 depict the average annualized number of days/year the PEV and ICE was used

for LDT50 and LDT100 in 3 car PHEV and BEV HHs, respectively. The ICE share (%) of total HH LDT50/100

days is shown using the secondary Y axis in Figure 77 and Figure 78.

Referring to Figure 77, in 3 car HHs, we can observe that the ICEs in Volt-16 HHs were used the most for

LDT50 and LDT100 followed by the ICEs in Prius Plug-in-4.4 HHs and the ICEs in C-Max/Fusion-7.6 HHs.

PHEV Engine Starts Analysis 100

Figure 78. BEV and ICE Use (Days/Year) for Long Distance Travel 50(100) Miles or More in 3 Car HHs

(Two ICEs and Single BEV); ICE Share (%) of Total HH LDT50(100) days/year shown on the secondary Y

axis. N=12 HHs.

Referring to Figure 78, in 3 car HHs, we can observe that the ICEs in Leaf-24 HHs were used the most for

LDT50, whereas the ICEs in Leaf-30 HHs were used the most for LDT100. The ICE usage in Bolt-60 and

Tesla Model S-60_80 HHs for LDT50 was almost similar whereas the LDT 100 data showed no similarities

in ICE shares across BEV types.

5 PHEV Engine Starts Analysis

Cold Starts

According to CARB’s vehicle emission inventory model (EMFAC), for typical ICE vehicles, a cold start is

defined as an engine ignition event after the engine has been off and the vehicle is stationary for 12

hours(CARB 2018). PHEVs have both a battery and an ICE engine and under certain circumstances, the

ICE engine may go through an ignition event while the vehicle is already on the road after it was initially

started by the battery. Under this circumstance the ICE engine in a PHEV may be going through both a

cold start under the usual ICE vehicle definition while also being high power because it is already on the

road and operating at an elevated speed or at high torque. In some PHEVs, the first time an engine

starts may be when higher power is required at some point during a trip, negating some of the

environmental benefits of reducing total number of cold engines starts results from completing trips and

travel days on electric mode only and the benefit of the low total gas consumption. High-power engine

starts have been associated with high local emissions of NOx and organic gases. Estimates based on

dynamometer measurements demonstrate that during such events, blended PHEVs emit at rates higher

PHEV Engine Starts Analysis 101

than they do during the lower power start events that occur during emission certification tests(CARB

2017, Pham and Jeftic 2018).

The objectives of this section are to characterize the engine start activity profiles of PHEVs, including: 1)

to define characteristics associated with all PHEV engine start events; 2) to identify conditions including

driving behavior, battery level, and other factors that trigger high SOC start engine events; and 3) to

determine the frequency of various types of starts. Further, more information is needed on total

number of engine-starts and how these compare with conventional vehicles. The analysis of this activity

data will be combined by CARB with previous emissions test results to better characterize real-world

emissions levels and to improve a future version of CARB’s EMFAC vehicle emission inventory model.

Based on results of this project, regulators may want to work with car manufacturers to devise emission

control strategies that mitigate high emission events during high power cold starts.

This study logged combined PHEV models (i.e., C Max/Fusion Energi) and non-blended PHEV models

(i.e., Volt). The second-by-second logger activity data from the logged PHEV models were analyzed to

better understand ICE-engine high power cold starts in the PHEVs described in this report. Because the

data on some parameters was collected at high frequency (approximately once every 1 to 10 seconds),

we can monitor the existing conditions in the few seconds before the engine starts in a PHEV. Our

analysis was able to classify all engine starts by state of charge (SOC), soak time, travel distance, and

speed. However, due to technical limitations inherit in the loggers in the second-by-second activity

logging, we were unable to pinpoint the reason for engine starts, such as high-power requirement

resulting from acceleration or a change in road grade.

The data collection was not synchronized for all parameters despite some parameters updating every 1

to 10 seconds. Furthermore, any parameter update generates a new timestamp and update of all the

old values of the other parameters that were not updated. We cannot distinguish between parameters

that have been updated but remained constant over several seconds versus those that have not been

updated and are simply duplicated from the previous measurement. A quick split-second change in

pedal position from 0% to 100% and back to 0%, for example, can be missed all together or alternatively

“stuck” for a few seconds on 100%. To overcome this limitation, we used the maximum value recorded

five and ten seconds before the engine start (RPM>500) to explore reasons for engine starts. For vehicle

models older than 2019, the SOC On-Board Diagnostic (OBD) Parameter Identification (PID) value is not

reported in a standardized way. Note that results for SOC reported here are shown as reported by the

CAN bus, but may not reflect absolute battery SOC. Our logger reported modeled catalyst temperature

only for the Volt and Energi. The data shows that cold starts happened mostly for the first engine start

of a trip and even for the longer-range Volt we did not record even one cold start that is not the first in

the trip. Our analysis, therefore, is focused on the first engine start in each trip.

Proportion of Days with Engine Starts

For PHEVs, engine starts are a function of many parameters, including SOC and power requirement,

among others. Figure 79 suggests a high correlation between battery size and days with no engine starts

that is similar to the zero emission trips and zero emission miles described in Section 3.5. For example,

the percentage of travel days that end without engine starts is around 4% for the short-range Prius Plug-

in-4.4 compared to 17% for the Energi. The Volts have a relatively high percentage of zero-emission

driving days (35%, 39%) because they are a non-blended PHEV. These percentages may be lower when

including PHEV users who drive their vehicle primarily as a conventional non-plug-in hybrid (charge less

than 4 times per month).

PHEV Engine Starts Analysis 102

Figure 79. Share of Drive Days with No Engine Starts

Engine Start Event Description

The data collected per trip was chronologically ordered in a time series database to extract valid engine

start events. An engine start event captures key metrics such as travel time and SOC within or around a

timeframe in a trip wherein the RPM is greater than zero for more than 10 seconds. Figure 80 provides a

snapshot of the raw, time trace of a valid engine-on event. The total number of engine start events

shared with CARB and used for this anlysis is 2,252,785 events, generated using data collected from 166

PHEVs, for up to one year per vehicle.

Figure 80. Engine-on Time Trace

3.86

17.32

26.86

25.24

35.42

38.63

27.23

0.00

5.00

10.00

15.00

20.00

25.00

30.00

35.00

40.00

45.00

Percentage of Driving Days (%)

PHEV Engine Starts Analysis 103

It is critical to note that the sample frequencies of the collected data attributes are not always

consistent. For instance, some attributes are collected every few seconds while other parameters are

recorded only when a change in value is detected; in such cases, a distinction cannot be made between

parameters that have been updated but remained constant over several seconds versus those that have

not been updated and are simply duplicated from the previous measurement. This lack of synchronicity

makes is extremely challenging to analyze the relationship between certain attributes. In Figure 80, for

example, we have a consistent speed trace for 10 seconds with one change in pedal position 3 seconds

in. We do not know if the speed change and pedal position change happened within 3 seconds as both

events could have happened within 5-10 seconds from reporting.

Travel Conditions at Engine Start

We first isolated and analyzed the following metrics, recorded at or prior to engine start events: SOC,

maximum power requirement (calculated based on battery current and voltage), and catalytic converter

temperature when available. We then analyzed the engine soak time (i.e., time elapsed between two

consecutive engine start events). Although we aimed to explore the relationship of vehicle power

requirements with road grade, we could not do so due to the differing data sample rates and imprecise

data values. The relationship with accelerator pedal position is based on max pedal position recorded 10

seconds before the engine start to cover for the data limitations.

SOC at Engine Start

One of the major causes for engine starts is the inability of the electric motor to propel the vehicle due

to a low battery SOC (state of charge). We, therefore, explored the distribution of battery SOC when the

engine is first turned-on within trips for all three PHEV models in the study. Figure 81 illustrates this SOC

distribution and highlights the fact that, for all vehicle models, most engine starts are invoked at a near-

zero usable SOC (reported by the vehicle) as expected. Over 70% of Energi, Prius Prime-8.8 and Volt

engine starts occur at SOCs below 5% while over 60% of Prius Plug-in-4.4 and Pacifica-16 engine starts

occur at SOCs under 5%. As presented in previous sections, the Prius Plug-in-4.4 engine is more likely

than the other models to start at high SOCs due to its relatively low battery capacity while the Volt

engine is least likely to start at high SOCs due to being a non-blended PHEV and having a significantly

higher battery capacity. The Pacifica-16’s engine is more likely to start than other vehicles with similar

battery capacities because it is a minivan, and therefore a heavier vehicle, leading to potentially higher

overall power demand.

PHEV engine starts can be invoked by either low battery energy or high power demand. A PHEV can start

its engine even with relatively high SOC if the power demand imposed on it exceeds the power that can

be supplied by just its electric motor. Therefore, to decouple the effect of vehicle SOC and power

demand, we developed three SOC classifications for engine starts. For all vehicle models, low or Empty

(E) SOC is between 0% to 1%, medium (M) SOC is between 1% to 10%, and High (H) SOC is over 10%.

PHEV Engine Starts Analysis 104

Figure 81. SOC at First Engine Start

Maximum Estimated Power Requirement before Engine Start

As discussed earlier, in certain driving situations such as traveling at high speeds or climbing a steep

incline, a PHEV’s power requirement may exceed the power that can effectively be provided by its

electric motor, regardless of the vehicle’s battery SOC; these situations can force the internal

combustion engine to start up to provide the additional power required to propel the vehicle at an

appropriate speed. We explored the distribution of the maximum power requirement 5 seconds before

the first engine start within trips, acknowledging the potential error due to time reporting gaps between

the parameters, broken down by the SOC classifications determined in section 5.4.1, for each vehicle

model (Figure 82). For the PHEVs with relatively lower battery capacities such as the Prius Plug-in-4.4

and the Energi vehicles, most low SOC engine starts correlate with lower power requirements (0-12 kW)

while majority of high SOC engine starts correlate with relatively higher power requirements (25-42 kW).

On the other hand, the engine starts of PHEVs with relatively high battery capacities, such as the Volts,

do not seem to correlate strongly with power requirements; these vehicles are least likely of the models

to invoke an engine start in the incidence of high-power requirements, regardless of their SOC.

However, the Pacificas, despite having battery capacities close to that of the Volts, seem to invoke the

engine at medium and high SOCs under a wide range of power requirements. This is probably because

the Pacificas, being classified as mini-vans, are much larger and heavier than the Volts, potentially

leading to a higher incidence of greater power requirements from larger road loads. Overall, the Prius

Plug-in-4.4 and Energi vehicles, having relatively smaller battery capacities, are more likely to turn on

their engine to meet high power requirements while the Volts, being non-blended PHEVs and having a

larger battery capacity, are least likely to start their engine in the presence of high-power requirements.

100

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Share of Engine Starts (%)

Battery SOC at Engine Start

SOC At Engine Start

Plug-in Prius

C-Max-Fusion

Prius Prime

Pacifica

Volt 16

Volt 18

PHEV Engine Starts Analysis 105

Figure 82. Maximum Power Requirement 5 Seconds before Engine Start

Catalyst Temperature before Engine Start

Our loggers captured modeled catalyst temperature data for only the Energi and Volt vehicles. For all

engine start trips of these two PHEV models, we analyzed the distribution of catalyst temperature for

the first engine starts and all subsequent engine starts separately, assuming that the first starts would

include a mixture of cold and hot starts and that subsequent starts would include hot starts. Figure 83

depicts the distribution of catalyst temperature of first engine starts in blue and all subsequent engine

starts in red. For both vehicle models, around half of the first engine starts occurring at temperatures

above ambient temperatures. We did not observe any cold starts after the first start for all trips even

though 0.4% of the starts may not be fully warmed up to 425ºC. The lack of cold restarts could be

because the vehicles are keeping the engine on for enough time to ensure that the first engine start

warms the catalyst for any potential subsequent starts within the same trip. In addition, the time

elapsed between consecutive engine starts is fairly small; among all the PHEV trips, the longest time

PHEV Engine Starts Analysis 106

elapsed between the first engine start and its successive start was 245 seconds (about 4 minutes) which

is not enough time for the catalyst to completely cool off.

Figure 83. Catalyst Temperature at Engine Start

Engine Soak Time

For all engine start trips, we analyzed the time elapsed between two consecutive engine starts (soak

time). This analysis includes any engine start regardless of travel distance and is based only on time and

RPM. Engine starts that were not the first engine start of trips were filtered out; we solely studied the

soak time of the first engine start of every trip. Cold starts were defined, as starts after 720 minutes (i.e.,

12 hours) , which is consistent with EMFAC, with variation of warm starts depending on the minutes the

engine is at idle. The soak time of each engine start was calculated by measuring the duration between

it and the engine start preceding it. The SOC classification criteria derived in section 5.4.1 was again used

to categorize the engine starts. Figure 84 to Figure 89 present the soak time distribution of Prius, Energi,

Pacifica, and Volt engine starts, respectively.

For all vehicles, there seems to be an inverse relationship between soak times and engine start shares;

the proportion of engine start events decay as soak time increases. For all PHEV starts, high SOC starts

seems to be more prevalent with greater soak times. Engine starts with higher soak times may be more

likely to have higher SOCs than engine starts with lower soak times because vehicle charging sessions

are more likely to have occurred between trips that result in a high soak times given there is a relatively

larger potential charging window.

For comparison to the PHEVs, Figure 90 presents the soak time distribution of ICE vehicles starts from

the conventional gasoline vehicles from the households participating in this study. The soak distribution

from these conventional vehicles seems to be similar to that of the PHEVs.

PHEV Engine Starts Analysis

107

Figure 84. Prius Plug-in-4.4 Soak Time by SOC at Engine Start

Figure 85. C-max Energi Soak Time by SOC at Engine Start

PHEV Engine Starts Analysis

108

Figure 86. Prius Prime-8.8 Soak Time by SOC at Engine Start

Figure 87. Pacifica-16 Soak Time by SOC at Engine Start

PHEV Engine Starts Analysis

109

Figure 88. Volt-16 Soak Time by SOC at Engine Start

Figure 89. Volt-18 Soak Time by SOC at Engine Start

PHEV Engine Starts Analysis 110

Figure 90. ICE Soak Time for the Conventional Gasoline Vehicles in Households

Distance Between Engine Starts

For each engine start trip, we analyzed two key distance metrics: the distance traveled from the

beginning of a day to the first engine start of the day and the distance traveled from the beginning of a

trip to the first engine start of the trip. To derive the first distance metric, we first grouped trips into

days with a 3AM cutoff rather than the standard 12AM cutoff and then aggregated the distance of all

trips that took place between the start of a day and the first engine start of the day for all days with an

engine start. We chose a 3AM cutoff as it is the hour with the lowest trip frequency for all vehicle trips in

our dataset. For the second distance metric, we simply calculated the distance from the start of a trip to

the point at which the engine is first initiated for all engine start trips. For the first metric, we are only

considering the first engine start of each day with an engine start while for the second metric, we are

considering the first engine start of every trip. Figure 91 and Figure 92 depict the distribution of these

two-distance metrics for all PHEV vehicles.

Over 80% of the Prius Plug-in-4.4’s first engine starts occurring after less than 5 miles of travel from the

beginning of the day; most of these starts happening at medium to high SOCs. On the other hand, less

than around 40% of the Prius Prime-8.8, Pacifica-16 and Volt first engine starts occur after less than 5

miles of travel from the beginning of the day; while in this case, most of the Volts’ starts occur at low

SOCs, the Pacifica-16 starts occur at mostly medium to high SOCs. The Volts are also more likely to have

engine starts after longer distances of travel from the start of the day than other PHEVs. The Energi

vehicles have a lower proportion of engine starts than the Prius-Plug-in-4.4 and a greater proportion of

engine starts than the Volts after less than 5 miles of travel from the beginning of the day. These

observations are in line with section 5.4.2 which found that PHEVs with relatively small battery

capacities such as the Prius Plug-in-4.4’s and the Energi vehicles are more susceptible to engine starts at

medium and high SOCs than PHEVs with larger battery capacities such as the Volts, to meet high power

demands. Overall, the occurrence of engine starts is more correlated to power demand than with SOC

(vehicle range) for small battery PHEVs than it is for large battery PHEVs. For all PHEVs, over about 70%

% of All Trips

Soak Time (minutes)

ICE Soak Time

PHEV Engine Starts Analysis

111

of engine starts occurring after less than 5 miles of travel from the start of the trip; most of these starts

happening at low to medium SOCs, suggesting that most engine start trips start with low SOCs.

Figure 91. Distance from Start of Day to First Engine Start of Day for all PHEVs

PHEV Engine Starts Analysis 112

Figure 92. Distance from Start of Trip to First Engine Start of Trip for all PHEVs

Potential Emission Impacts of Engine Starts

Figure 93 illustrates the probability of various engine starts to occur on a given day for each vehicle

model. The probabilities were derived from the annualized engine start days for each vehicle in the

study. For this analysis, a cold start is an engine start with a soak time greater than 12 hours and a high-

power cold start is a cold start with a maximum power requirement 5 seconds before the start of over

25 kW. There is an inverse correlation between vehicle battery capacity and the probability of an engine

start with the Prius Plug-in vehicles having far more engine start days than the Volt vehicles. However,

this is not the case for cold starts and high-power cold starts. The Prius Plug-in vehicles and the Volts

PHEV Engine Starts Analysis 113

logged fewer instances of cold starts than the mid battery capacity vehicles; this is probably because the

Prius Plug-in starts it engine more frequently, resulting in shorter engine cool down periods than the

other vehicles, while the Volt vehicles are least likely to start the engine in the first place given their

larger battery capacity.

The mid-sized battery vehicles show large variation in engine start probabilities since, as observed in the

previous section, there are several factors, ranging for low peak electric motor power to frequent high-

power demands due to high curb weight, that determine when the engine is invoked for these cars. The

C-Max-Fusion cars and the Pacificas, on average, have much higher proportions of cold start days than

the other vehicles. The C-Max-Fusion cars have low electric motor power capabilities than the other

vehicles, owing to its considerable proportion of high-power cold starts. The Pacificas, being minivans,

are far heavier than the other models and so are more likely to have high power demand instances,

resulting in a high frequency of cold starts despite their high battery capacity.

Figure 93. Probability of Engine Start per Vehicle Model (Derived from Annualized Engine Start Days)

Figure 93 depicts the average number of daily cold starts that occur for each vehicle model juxtaposed

with the average number of daily cold starts if the models behaved like conventional ICE vehicles with

the engine being invoked at the beginning of every trip. As expected, the Volts had fewer real cold starts

than their hypothetical ICE cold starts given that their larger battery capacity and drivetrain design seek

to reduce engine starts to optimize fuel displacement. The Plug-in Prius also had few real cold starts

compared to its hypothetical ICE vehicle performance since it is more likely to have shorter engine cool

PHEV Engine Starts Analysis

114

down periods between starts than the other vehicles. The C-Max-Fusion vehicles and Pacificas do not

seem to have a stark difference between actual cold starts and Hypothetical ICE cold starts because they

are more likely to have high-power requirement instances during trips, owing to their low electric power

capabilities or high curb weight. The mid capacity battery vehicles are less likely than the Prius-Plug in

cars to have short engine cool down periods between starts and more likely than the Volts to have

engine starts, making then more prone to cold starts than PHEVs with considerably lower or higher

battery capacities. Overall, the Volts do a much better job than the other vehicles at both curbing start

emissions via logging very few engine starts and maximizing fuel displacement.

Figure 94. Average Daily Cold Starts

Engine Starts Discussion

This section includes only the initial analysis of the data collected. The main task of this project was to

provide to CARB the full dataset of engine starts including the events before and after the engine starts

for further analysis. The data preparation included quality control and cleaning missing and bad data

results from problems in logger configurations. We also tested the GPS elevation data using GIS models

and conclude that the accuracy level was not sufficient for energy and power analysis. Overall, the data

collected, and the sample times are not sufficient for calculating their power requirement and other

factors for engine starts. Nevertheless, data analysis shows that long-range plug-in hybrids can finish

many days and trips without any engine starts. We also conclude that long-range plug-in hybrids engine

starts are mostly correlated with battery state of charge, while short range PHEVs’ engine starts may be

correlated with other factors. For all vehicles, there seems to be an inverse relationship between soak

times and engine start shares; the proportion of engine start events decay as soak time increases. For all

PHEV Starts, high SOC starts to seem to be more prevalent with greater soak times; engine starts with

Fuel Cell Vehicle Analysis 115

higher soak times may be more likely to have higher SOCs than engine starts with lower soak times

because the vehicles had more time to potentially recharge their batteries.

The probability of an engine start to occur in a day, for each model, can be explained by the models’

individual specifications. In general, there is an inverse correlation between vehicle battery capacity and

the probability of an engine start with the Prius Plug-in vehicles having logged far more engine start days

than the Volt vehicles. However, this trend does not hold for cold starts and high-power cold starts. The

Prius Plug-in vehicles and the Volts logged fewer instances of cold starts than the mid battery capacity

vehicles; this is probably because the Prius Plug-in starts it engine more frequently, resulting in shorter

engine cool down periods than the other vehicles, while the Volt vehicles are least likely to start the

engine in the first place given their larger battery capacity. The mid-sized battery vehicles show large

variation in engine start probabilities since there are several factors, ranging for low peak electric motor

power to frequent high-power demands, that determine when the engine is invoked for these cars.

Comparing the average daily cold starts for each PHEV model to the average daily cold starts if the

model vehicles behaved like conventional ICE vehicles showed that all models would have logged more

cold starts if they behaved as ICE vehicles, suggesting that they all incur start emission savings

functioning as PHEVs. The vehicles with mid electric range are less likely than those with low electric

range (Prius Plug-in) to have short engine cool down periods between starts and more likely than those

with high electric range (Volt) to have engine starts, making them more prone to cold starts than PHEVs

with lower or higher battery capacities. There was a 136% increase in cold starts for the Volt 18 vehicles

if they behaved as ICE vehicles compared to just 11% increase in cold starts for the C-Max-Fusion cars.

The Volts, the PHEVs with the highest electric range, do a much better job than the other vehicles at

both curbing start emissions via logging very few engine starts and maximizing fuel displacement as

suggested by their relatively high electric miles to total miles ratio.

6 Fuel Cell Vehicle Analysis

In this section, we present the data collected from Fuel Cell Vehicle (FCV) usage at the vehicle and

household level using data collected from the loggers. In total, we have 12 Toyota Mirais in our FCV

dataset, but one Mirai was dropped from our analysis as it had trouble acquiring key data points (speed,

GPS, etc.) for lengthy periods of time within its logging window. The remaining 11 vehicles that have

reliable data for calculating key distance, speed, and energy metrics for at least 120 days are included in

the analysis. Problems with logger reliability allow us to use only 60% of the days when loggers were

installed, loggers that were connected for more than one year.

FCVs have driving ranges of more than 300 miles and can be refueled in less than 10 min at a hydrogen

fueling station. Three original equipment manufacturers (OEMs) currently offer FCVs in California, with

the Toyota Mirai being the most common (Figure 95). Sales of these vehicles began in 2014, with most

vehicles leased for a period of three years. In most cases, hydrogen fuel cost is subsidized by the OEM.

Fuel Cell Vehicle Analysis 116

Figure 95. Fuel Cell Vehicle Sales by Model, Country, and Model for California

California currently has 41 active hydrogen-fueling stations, built through a combination of industry

funds and capital and operating cost support from the California Energy Commission (CEC). These are in

the Los Angeles, San Francisco Bay, and Sacramento areas as shown in Figure 106. In our study, we focus

on the Toyota Mirai which can hold approximately five kilograms of compressed hydrogen gas, enabling

it to travel up to 312 miles on a full tank. The descriptive summaries and analyses of the FCVs are

presented in Table 25-Table 26 and visualized in Figure 96-Figure 97. Table 25-Table 26 summarize the

data collected on FCV driving and refueling. As we did for filtering BEV and PHEV trips, we used a

distance threshold of 1 km to filter out GPS noise and noticeably short trips recorded by the loggers with

zero to minimal energy usage. Overall, around 88% of FCV VMT still remained after filtering. The

refueling sessions were not directly reported by the loggers but were determined by looking at the

difference between the fuel levels of consecutive trips; if there is a significant difference the ending fuel

level of a trip and the starting fuel level of the trip immediately following it, we gauged that a refueling

event occurred between those trips. Unfortunately, fuel level was one of the raw data points that was

reported very sparsely by the loggers for most vehicles, so we probably failed to capture some refueling

events that occurred within the logging window of each vehicle.

Table 25. FCV Driving Data Overview

FCV Type

Number of

Vehicles

Raw Data

Trips

Raw Data

Total VMT

Filtered

Data Trips

Filtered

Data Total

VMT

Filtered Data

Average

Driving

Days/Vehicle

Mirai 11 15301 104133 11327 91164 247

Fuel Cell Vehicle Analysis 117

Table 26. FCV Refueling Data Overview

FCV Type

Number of

Vehicles

Refueling

Sessions

Total

Hydrogen

(kg)

Total

Refueling

Days

Mirai 11 408 823.4 228

Figure 96. Annualized VMT of Mirais

Figure 96 shows the annualized VMT of the Mirais based on data collected from the loggers. The fleet

average annualized VMT for the Mirais is 10,738 miles. There is a difference of over 10,000 miles

between the vehicle with the highest annualized VMT and the vehicle with the lowest annualized VMT,

showing a wide range in vehicle mileage.

5844

7617

8149

9303

9575

9710

10587

10719

12810

13841

16009

10738

2000

4000

6000

8000

10000

12000

14000

16000

V1 V4 V2 V5 V3 V8 V6 V7 V9 V10 V11

Annualized VMT (miles)

Vehicles

Annualized VMT Mean VMT

Fuel Cell Vehicle Analysis 118

Figure 97. Mirai: Percentage Share of Total VMT by Trip Speed (in mph)

Figure 97 depicts the share of total VMT by trip speed bin for each Mirai. Most vehicles cover a high

proportion of miles at speed of 30-75 mph and cover very few miles at speeds over 75 mph. V8,

however, covers a considerable proportion of miles at speeds over 75 mph.

Fuel Cell Vehicle Driving

Table 27 and Figure 98-Figure 102 provide descriptive summaries of Mirai driving patterns captured by

data collected from the loggers. As shown in Table 27, on average, Mirai drivers, make around 4 trips

per day with each trip ranging from 6 to 11 miles. There is extremely low variation in hydrogen usage

amongst the vehicles; all vehicles consistently use about 0.02 kg of Hydrogen, on average, for covering

one mile. The average trip distance of Mirai weekday trips is higher than the average weekend trip

distance by about 2 miles while there is around a 23-mile difference between the maximum distance of

weekday and weekend trips. 70% of the trips of half the vehicles (namely V1-V5) were less than 35 miles

long whereas roughly 60% of the trips of most of the remaining vehicles (V7-V11) were over 35 miles.

Table 27. Mirai Driving Trip Level Summaries (on days when the FCV was driven)

Mirai Vehicles

Average

Trips/Day

Average Trip

Distance

(miles)

Average

kg(H2)/Trip

Average

kg(H2)/Mile

Average

VMT/Day

(miles)

V4 4.11 7.29 0.12 0.0163 21.71

V10 5.86 8.14 0.16 0.0204 42.18

V2 3.76 6.69 0.12 0.0168 10.28

V6 4.75 6.93 0.14 0.0206 32.13

7.3

3.7

6.7

9.2

3.6

4.2

3.5

5.5

6.4

6.6

3.8

23.2

10.2

15.7

26.0

9.9

11.8

11.2

18.5

17.9

19.8

11.0

23.7

14.8

27.1

20.8

16.1

18.1

17.5

34.4

17.5

27.2

18.5

16.6

12.9

12.7

16.3

23.9

15.4

11.2

21.3

21.0

21.2

17.1

27.4

52.3

25.9

24.9

35.2

48.4

33.2

18.9

37.1

21.1

48.7

1.8

6.0

11.9

2.8

11.3

2.0

23.3

1.4

0.1

4.1

0.9

0.0

10.0

20.0

30.0

40.0

50.0

60.0

70.0

80.0

90.0

100.0

V4 V10 V2 V6 V7 V5 V8 V3 V1 V9 V11

Percentage of Total VMT

0-15 mph 15-30 mph 30-45 mph 45-60 mph 60-75 mph 75+ mph

Fuel Cell Vehicle Analysis

119

Mirai Vehicles

Average

Trips/Day

Average Trip

Distance

(miles)

Average

kg(H2)/Trip

Average

kg(H2)/Mile

Average

VMT/Day

(miles)

V7 3.85 8.57 0.19 0.0266 38.53

V5 4.09 7.52 0.15 0.0208 26.48

V8 3.70 10.25 0.20 0.0191 36.2

V3 4.14 6.81 0.12 0.018 25.36

V1 4.18 5.59 0.09 0.0157 17.7

V9 4.00 10.54 0.20 0.0208 43.58

V11 4.96 11.00 0.18 0.0182 52.05

All Vehicles 4.13 8.05 0.15 0.019 30.8

Figure 98. Average and Maximum Trip Distance on Weekdays and Weekends by Mirai

8.93

213.71

6.92

191.21

100

150

200

250

mean distance max distance

Daily Miles

weekdays

weekends

Fuel Cell Vehicle Analysis

120

Figure 99. Percentage of Trips per Vehicle by Trip Distance Bins (miles)

27.9

39.0

11.8

21.7

10.7

14.7

17.8

2.7

7.1

9.3

18.3

25.7

38.8

10.7

21.7

6.1

19.2

8.6

5.0

4.0

3.7

6.3

26.8

9.6

58.4

27.7

62.1

21.6

6.8

20.1

16.4

15.7

10.3

10.6

1.6

10.3

9.4

9.8

26.0

57.5

59.4

50.0

31.5

11.6

4.5

6.4

5.7

11.1

6.5

15.9

8.3

10.0

16.8

25.0

23.2

4.5

1.1

2.3

6.0

2.8

1.2

0.6

1.8

3.1

11.1

18.8

100

V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11

Percentage of Trips (%)

Vehicles

200+ mi

100-200 mi

75-100 mi

50-75 mi

35-50 mi

20-35 mi

10-20 mi

0-10 mi

Fuel Cell Vehicle Analysis 121

Figure 100. Average Monthly Vehicle VMT Across Deployments

Figure 100 depicts the monthly average VMT of the Mirais in the dataset broken down by individual

deployments.

Fuel Cell Vehicle Analysis 122

Figure 101. Average Daily FCV VMT per Vehicle

Figure 102. Average FCV Energy Efficiency (miles/kg(H2)

Figure 101 presents the average daily VMT of individual Mirais while Figure 102 shows the average

energy efficiency of the vehicles. Only 4 out the 11 Mirais had an average daily VMT higher than the

fleet average of 35 miles. The fleet average efficiency of the vehicles was around 60 miles per 1 kg of

hydrogen. While most vehicle noted average efficiencies that were fairly close to the fleet average,

some vehicles recorded efficiencies far lower than the fleet average. V7 showed a particularly low

34.80

V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11

Average Daily Distance (miles)

Average Daily FCV VMT per Vehicle

Average Daily VMT Overall Average VMT

59.98

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

V7 V5 V10 V6 V9 V8 V3 V11 V2 V4 V1

Energy Efficiency (miles/kg(H2))

Vehicles

Average FCV Energy Efficiency (miles/kg(H2))

Average miles/kg(H2) Overall Average miles/kg(H2)

Fuel Cell Vehicle Analysis 123

efficiency of just around 40 miles per kg of hydrogen; this observation could be attributed to the fact

that over 70% of V7’s VMT is covered at speeds over 45 mph, according to Figure 97. Vehicles that use

an electric motor to power themselves tend to be more efficient at low to medium speeds, so a greater

frequency of high-speed driving can lower the average efficiency of these cars.

Fuel Cell Vehicle Refueling

Table 28 and Figure 103-Figure 109 provide descriptive summaries of Mirai refueling patterns captured

by data collected from the loggers. Refueling events were identified from significant increases in

hydrogen level between trips. Refueling events were matched to fueling stations using the GPS

coordinates at the end of the trip preceding the refueling event. Over 90% of refueling events had GPS

coordinates within 100m of a known hydrogen fueling station, and over 96% could be matched to a

fueling station within 200m. No refueling events matched multiple stations. Mirai refueling summary

statistics are presented in Table 28. Figure 103 illustrates the distribution of the time elapsed between

the refueling events of all vehicles while Figure 104 displays the distribution of the distance traveled

between the events. About 70% of refueling events occur within a week after their immediately

preceding refueling event and around 80% of the events occur after covering distances of over 100 miles

(32% of the vehicle’s range of 312 miles) from their preceding refuel.

Table 28. Refueling Summary

Average Sessions/Day Average kg(H)/Day

Days within Logging Window 0.08 0.17

Days when the FCV Refueled 1.02 2.59

Figure 103. Time (in minutes) Elapsed between Refuels

Share of Refuel Events (%)

Time (minutes)

Fuel Cell Vehicle Analysis 124

Figure 104. Distance Traveled Between Refuels

Figure 105. Average Fuel in during Refuel Days per Vehicle

Figure 105 displays the average fuel in during refuel days for each vehicle. While on average Mirai

drivers seem to fill their tank with around 1.97 kg of hydrogen on refuel days, there seems to be a wide

Share of Refuel Events (%)

Distance (miles)

1.97

0.5

1.5

2.5

3.5

V4 V10 V6 V7 V2 V1 V11 V5 V3 V9 V8

Average Fuel in (kg(H2))

Vehicles

Average Fuel in during Refuel days per Vehicle

Avg. Fuel in (kg(H2)) Overall Avg. Fuel in (kg(H2))

Fuel Cell Vehicle Analysis 125

variation in refuel behavior. Some drivers seem to top off their tanks while others seem to wait to fill up

their tanks after covering a significant amount of distance. V8 and V9, on average, wait until over 60% of

their fuel is consumed before refueling. This variation could be linked to the distance between the

typical trip origin/destination sets of these vehicles and the hydrogen stations that they frequent;

further analysis is required to assess this connection.

Figure 106. Number of Accessible Hydrogen Fueling Stations within 140 Miles of Each Block Group

Figure 106 captures how many hydrogen stations that are open to the public are accessible from each

block group within the range of half a tank of a Mirai FCV. There seems to be a high to moderate

concentration of accessible public hydrogen stations around areas (colored in red to beige), surrounding

major metropolitan areas i.e. San Francisco and Los Angeles. On the other hand, the regions to the far

north and far east of the state (colored in white) have no public stations that can be reached on half of a

fuel tank when driving a Mirai, constraining the travel radius of FCV drivers that enter those regions.

Fuel Cell Vehicle Analysis 126

Figure 107. Areas within 5 and 10 Miles from a Refueling Station

Figure 107 visualizes the areas that are less than 5 and 10 miles away from a public hydrogen refueling

station. There seem to be a limited number of areas that have public hydrogen refueling stations within

a 10-mile radius. The areas are mostly clustered around the state’s major metropolitan regions (San

Francisco, Los Angeles, Sacramento, etc.). However, since most of the state does not have accessible

hydrogen stations within a 10-mile driving distance, it potentially plays a role in dissuading drivers from

investing in FCVs.

Fuel Cell Vehicle Analysis

127

Figure 108. Average Fuel pumped in during Refuel, per Station

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

4.50

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14 S15 S16 S17 S18 S19 S20 S21 S22 S23 S24 S25

Average Fuel in (kg(H2))

Stations

Fuel Cell Vehicle Analysis 128

Figure 109. Refueling Share per Station

Figure 108 shows the average fuel during a refuel for individuals at hydrogen stations while Figure 109

presents the share of refuels per hydrogen station. Majority of the refuels seem to take place at 10 out

of the 25 stations visited by the vehicles in the dataset. This could be attributed to the fact that there

are a lot of missing refuel events due to a high proportion of missing fuel level data for some cars as

mentioned earlier. Of these 25 stations, 11 are in Los Angeles County (stations 2, 3, 4, 5, 6, 10, 12, 18,

19, 24, and 25), 7 are in Santa Clara County (1, 8, 9, 13, 14, 16, and 17), 4 are in Orange County (7, 11,

15, 21), and one each are in Fresno (20), San Diego (22), and Ventura (23) counties.

Fuel Cell Household Analysis

Our dataset contains 11 FCV households with 4 Mirai-only households, 6 single ICEV Mirai household

and one double ICEV Mirai household. The double ICEV household was dropped from this analysis due

to low sample size. Seven of these households are in the San Francisco Bay Area and four are in the Los

Angeles Area. Table 29 summarizes the (average) annualized estimates of key FCV metrics at the

household level. The metrics include eVMT, gVMT, HH VMT, UF and energy consumption. Figure 110

presents the HH UF of individual FCV households. The UF of individual Mirai households ranged from

0.47 to 0.83 with an average of 0.65. Figure 111 shows the average daily HH VMT and the share of

eVMT, gVMT in FCV HHs.

Based on Figure 111 that plots the average VMT of individual households, households with higher daily

average VMT (>40 miles) tend to have high UF; this can be observed in HH6, HH8 and HH10.

S17 S9 S12 S8 S5 S13 S6 S10 S1 S15 S11 S18 S19 S2 S21 S3 S4 S16

Share of Fuel consumed from station (%)

Stations

Fuel Cell Vehicle Analysis

129

Table 29. (Average) Annualized Estimates of VMT and Energy Consumption on FCV HHs

HH Type Num HHs FCV Trips

FCV eVMT

FCV Hydrogen

Consumed (kg)

ICEV gVMT

ICEV Fuel

(gallons)

HH VMT HH UF

ICE-FCV 6 1282 9892 139 5337 250 15229 0.65

FCV 4 1171 10620 190 0 0 10620 1

Figure 110. Household Utility Factor per Vehicle by Household Car Composition

1 1 1 1

0.47

0.72

0.62

0.72

0.52

0.83

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

HH1 HH2 HH3 HH4 HH5 HH6 HH7 HH8 HH9 HH10

FCV ICE-FCV

Household Utility Factor

Households

Fuel Cell Vehicle Analysis

130

Figure 111. HH Average Daily VMT in HHs with FCVs, Showing the eVMT and gVMT Percentages

Figure 112. Percentage of FCV and ICEV trips in HHs with FCV

100 100 100 100

57.0

71.5

47.6

60.8

70.3

57.1

0 0 0 0

43.0

28.5

52.4

39.2

29.7

42.9

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

HH1 HH2 HH3 HH4 HH5 HH6 HH7 HH8 HH9 HH10

FCV ICE-FCV

% of Household Trips

Households

ICEV Trips

FCV Trips

Fuel Cell Vehicle Analysis 131

Figure 112 depicts the percentage of HH trips taken using the FCV and the ICEVs. On average, FCV share

of HH trips was approximately 61% for the two car HHs. HHs that reported high UFs did not necessarily

have high ratio of FCV trips to ICEV trips. For instance, HH7 and HH10 showed relatively high UF but they

have relatively low ratio of FCV to ICEV trips (less than 57%), suggesting that some HHs are using their

FCVs for trips covering longer distances.

Figure 113. Number of Days/Year FCV was Used for Long Distance Travel (LDT).

198

0 0

0 0 0 0

0 0 0 0 0

100

150

200

250

HH1 HH2 HH3 HH4 HH5 HH6 HH7 HH8 HH9 HH10

FCV ICE-FCV

Number of Days/Year

Households

LDT50

LDT100

LDT200

LDT300

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews 132

Figure 114. Percentage of FCV and ICEV LDT (>100 mi) Days over LDT Days in HHs with FCV

Figure 113 shows the absolute number of days the FCVs were used for LDT while Figure 114 shows the

ratio of FCV and ICE LDT (over 100 miles) days. As expected HH7 and HH10 which showed relatively high

UFs and low FCV to ICE trip ratios, have a substantial proportion of FCV LDT days. On the other hand,

HH9 which had a relatively low HH UF and high FCV to ICE trip ratio, covered all its trips exceeding 100

miles in the ICE.

7 Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews

The data for this section is from forty semi-structured qualitative interviews that were conducted with

PEV drivers across California. The interviews explored driver commute information, charging behavior,

and experiences with charging etiquette (see Appendix 1 for the topics explored in the interviews).

Interviews lasted on average 45 minutes and were conducted at the interviewees’ homes. Interviewees

were selected based interviewer’s availability and the driver’s willingness to share their experiences.

Only forty interviews were conducted because that was when saturation occurred, and no innovative

ideas were being put forth. This section aims to uncover how workplaces manage charging to increase

the utilization of charging stations, and to understand drivers’ experiences with them. We can then

recommend improved techniques to maximize the utilization of charging infrastructure at work. By

managing how people charge at work, more people can use the workplace chargers per day, meaning

fewer chargers are needed to support a larger number of PEVs, and more vehicles can be driven on

electric miles. Workplace charging stations can be seen as a scare resource when improperly managed

due to their limited availability. Appropriately managed workplace charging can alleviate congestion and

increase the availability of the scarce resource. Some workplaces may experience charging station

stagnation if there is no motivation to move the vehicle when it has done charging. Conversely, if there

was higher turnover of vehicles per day, fewer chargers would be needed to meet demand. Workplaces

who are considering installing chargers can use this research to plan charging stations locations with

maximum benefits for charging vehicles and minimizing charging congestion. Compared to home-based

100 100 100 100

40.5

37.5

16.7

100

0 0 0 0

59.5

62.5

83.3

100

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

HH1 HH2 HH3 HH4 HH5 HH6 HH7 HH9 HH10

FCV ICE-FCV

% of Household LDT Days

Households

FCV LDT Share ICEV LDT Share

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews 133

charging, workplace charging stations can support a higher number of PEVs with a lower amount of

stations which can reduce infrastructure investment costs.

Introduction to Charging Management Strategies

Here we define three terms used throughout the section, these are charging congestion, charging

station stagnation, and charge management strategies. Charging congestion is defined as a greater

number PEVs wanting to charge than available electric vehicle chargers for those vehicles. This generally

means PEV owners are unable to charge or must wait to access a charger. Charging station stagnation is

when a PEV plugged into the charger has completed charging and the owner has not made the charger

available to other PEV owners, either due to not moving their vehicle or unplugging the charge outlet.

This can exacerbate the issue of charging congestion and can lead to chargers being underutilized.

Finally, charging management strategies are any rules aimed at reducing stagnation and congestion,

which results in higher utilization of chargers. These rules could be formal or informal, and may include

time restrictions, pricing, formalized queuing, or anything else.

Methods

The data for this section was collected from drivers across California, which were selected from the

households that participated in logged study and survey, in the summer of 2018. The interviews

explored driver commute information, charging behavior, and experiences with charging etiquette.

Interviews lasted on average 45 minutes and were conducted at the interviewees’ homes. Interviewees

were selected based interviewer’s availability and the driver’s willingness to share their experiences.

Forty interviews were conducted because that was when saturation occurred, and no innovative ideas

were being put forth. All interviews were then transcribed and coded in NVIVO software package.

This section focuses on a subset of forty drivers from the vehicle logger project who were part of the

study in 2018 and who agreed to be interviewed.

There is potential for selection bias in the sample as interviewees had previously agreed to and

participated in this project. We reached topical saturation with a diverse sample (geographically, by PEV

type, by house type, etc.), shown in Table 30 below, leads us to believe the interviews detect the most

common charging management strategies currently employed.

Results

Interviewee Descriptions

The PEVs discussed are as follows: four Nissan Leaf (9%), fifteen Tesla Model S (35%), six Ford C-Max

(12%), three Ford Fusion (7%), sixteen Chevrolet Volt (35%), and three other vehicles (5%) (one

Chevrolet Bolt, one Fiat 500e, and one Toyota RAV4 EV). There were 22 (47%) BEVs and 25 (53%) PHEVs

in the study. There were 11 (28%) MUD drivers and 29 (73%) single-family household drivers. Table 30

below has a full description of the households.

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews

134

Table 30. Summary of Interviewee Information

Interview

Total

# Cars

# of

PEVs

PEV

Vehicle

Style

# ICEs MUD Gender Age Region

1 1 1 Chevrolet Volt PHEV 0 1 Male 19-29 Bay Area

2 2 1 Ford C-Max PHEV 1 0 Male 30-39 Bay Area

3 2 1 Nissan Leaf BEV 1 0 Male 40-49 Bay Area

4 1 1 Tesla Model S BEV 0 0 Male 40-49 Bay Area

5 3 1 Tesla Model S BEV 2 0 Male 50-59 Bay Area

6 1 1 Chevrolet Volt PHEV 0 1 Male 19-29 Bay Area

7 3 1 Tesla Model S BEV 2 0 Male 60-69 Bay Area

8 1 1 Chevrolet Volt PHEV 0 0 Male 30-39 Northern

California

9 2 1 Nissan Leaf BEV 1 0 Male 30-39 Bay Area

10 2 1 Chevrolet Volt PHEV 1 1 Male 30-39 Bay Area

11 2 2

Tesla Model S /

Chevrolet Volt

BEV /

PHEV

0 0

Male &

Female

30-39;

30-39

Bay Area

12 2 1 Ford Fusion PHEV 1 0 Male 40-49 Sacramento Area

13 1 1

Chevrolet Volt /

Fiat 500e

PHEV /

BEV

0 0 Male 50-59 Bay Area

14 1 1 Chevrolet Volt PHEV 0 1 Male 30-39 Bay Area

15 1 1 Chevrolet Volt PHEV 0 1 Male 30-39 Los Angeles Area

16 3 1 Tesla Model S BEV 2 0 Male 40-49 Los Angeles Area

17 2 1 Tesla Model S BEV 1 0 Male 50-59 Los Angeles Area

18 2 2 Ford C-Max /

Chevrolet Volt

PHEV /

PHEV

0 0 Male 30-39 Los Angeles Area

19 2 1 Chevrolet Volt PHEV 1 1 Male 30-39 Los Angeles Area

20 2 2 Chevrolet Volt /

Nissan Leaf

PHEV /

BEV

0 0 Male &

Male

40-49;

50-59

Los Angeles Area

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews

135

Interview

Total

# Cars

# of

PEVs

PEV

Vehicle

Style

# ICEs MUD Gender Age Region

21 1 1 Tesla Model S /

Toyota RAV4 EV

BEV /

BEV

0 0 Male 40-49 San Diego Area

22 2 1 Chevrolet Volt PHEV 1 0 Male 50-59 Bay Area

23 2 1 Tesla Model S BEV 1 0 Female 50-59 Bay Area

24 1 1 Chevrolet Volt PHEV 0 1 Male 40-49 Los Angeles Area

25 3 1 Tesla Model S BEV 2 0 Male 50-59 San Diego Area

26 1 1 Chevrolet Volt PHEV 0 1 Female 40-49 Los Angeles Area

27 2 1 Tesla Model S BEV 1 0 Male 40-49 Los Angeles Area

28 2 1 Ford Fusion PHEV 1 0 Male 30-39 San Diego Area

29 2 2

Nissan Leaf /

Chevrolet Bolt

BEV /

BEV

0 1

Male &

Female

50-59;

40-49

San Diego Area

30 2 2

Chevrolet Volt /

Ford Fusion

PHEV /

PHEV

0 0 Male 50-59 San Diego Area

31 2 1 Tesla Model S BEV 1 0 Male 40-49 Los Angeles Area

32 1 1 Tesla Model S BEV 0 0 Male 70-79 Sacramento Area

33 2 1 Tesla Model S BEV 1 0 Female 60-69

Northern

California

34 1 1 Tesla Model S BEV 0 1 Male 30-39 Bay Area

35 4 1 Chevrolet Volt PHEV 3 0 Female 19-29 Sacramento Area

36 1 1 Ford C-Max PHEV 0 0 Male 19-29 Sacramento Area

37 2 1 Ford C-Max PHEV 1 0 Male &

Female

50-59;

40-49

Bay Area

38 1 1 Ford C-Max PHEV 0 1 Male 30-39 Bay Area

39 2 1 Ford C-Max PHEV 1 0 Male 40-49 Bay Area

40 4 1 Tesla Model S BEV 3 0 Male 50-59 Bay Area

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews 136

Charging Behavior

Interviewee charging behavior is discussed below because of its relevance to workplace charging

management. First, we explore home charging, followed by public, and then workplace charging. Then

we discuss workplace charging management in detail.

Home Charging

Thirty-five people charge their vehicles at home, ten are unable to, and two chose to not charge at

home (Figure 115). The interviewees who charge at home cite convenience and low refueling cost as

their main reasons behind this choice. Of those that live in MUDs, eight (73%) are unable to charge at

home. The drivers who live in a MUD noted the nuisances of not having a garage or access to electricity

at their parking spot. Of those who choose to not charge at home (a Tesla Model S and Nissan Leaf), the

Tesla Model S charges exclusively at work, and the Nissan Leaf has no issues relying solely on public

charging. The quotes below outline reasons why interviewees do not charge at home.

“I don’t have a garage, so I have to do street parking and that ultimately means I don’t charge my car

when I’m at home” (Interview 01, Chevrolet Volt)

“With the free options there didn’t really seem to me to be much need for that, um, and it’s not

inconveniencing me at the moment to do it the way I’m doing it. Umm so, yeah, I haven’t seriously

considered the charging at home” (Interview 03, Nissan Leaf)

Figure 115. Interviewees reported home based charging (N=47)

Public Charging

Public charging is classified as charging that is not at home or while at work, as defined by (Lee, J.H.

2020) [5]. The frequency of charging in public is broken down by always, frequently, occasionally, rarely,

and never (Figure 116).

Seven drivers reported always charging their vehicles when available in public. The vehicles include two

Nissan Leafs, three Chevrolet Volts, one Chevrolet Bolt, one Ford C-Max, and two Tesla Model S. Of

those seven, three are unable to charge at home. A few participants also mentioned that charging was a

determining factor of where they travel.

“I also charge when I go shopping at the mall or the park, wherever there’s a charging station…

[charging] actually kind of skews on where I watch a movie, so … I just look at which one has more

availability on charging stations and then that’s where I go.” (Interview 24, Chevrolet Volt)

Yes No, can't No, don't want to

Count

Reported Home Charging

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews 137

Seven interviewees reported charging in public frequently. This was defined as those who would charge

in public more likely than not when they go out. They may check the local EVSE they know for

availability and may not mind adjusting their parking patterns to charge their vehicles. The vehicles in

this category are two Chevrolet Volts, one Ford Fusion, one Ford C-Max, and three Tesla Model S. A few

mention the secondary benefits of charging in public such as parking priority or saving money.

“Whenever we go out if there’s a charger available well try to, to get it and just park there and you

know, even if we have to walk a little extra, we’ll do that” (Interview 08, Chevrolet Volt)

Sixteen drivers charge in public occasionally; from this sample, this is the largest segment. Occasional

public charging interviewees were defined as those who know their routine charging stations, but do not

seek out additional stations. As one interviewee put it, “I don’t really search for them. If they’re there,

they’re there” (Interview 02, Ford C-Max).

If it is convenient for them, they will charge their vehicle. The vehicles in this grouping are seven

Chevrolet Volts, two Ford C-Maxes, five Tesla Model S, and two Nissan Leafs.

“I generally just try to charge it when I can when it’s convenient for me. I don't want to go out of my

way” (Interview 14, Chevrolet Volt)

Ten interviewees rarely charge in public. Five Tesla Model S, one Ford Fusion, three Chevrolet Volts, and

one Ford C-max are part of this group. These drivers only use public charging as a last resort or very

infrequently.

“Public chargers are literally for emergency use cause especially since I have it at home, like I just don’t

need to use public chargers” (Interview 27, Tesla Model S)

One interviewee with a Ford C-Max does not charge in public. He cited the hassle of charging compared

to home charging and time to charge as preventing him from seriously considering using it.

“I don’t go that many places where I would have time to charge my car” (Interview 37, Ford C-Max)

Irrespective of public charging frequency, seven drivers only charge in public when it is free (Figure 116).

This is an additional grouping and is not a category of charging frequency. It contains two Ford C-Maxes,

three Chevrolet Volts, and two Tesla Model S. Paying for charging is not always a disincentive for drivers

to charge.

“I will only use a public charger if it’s available and they’re free” (Interview 23, Tesla Model S)

“There’s a lot of them that are like paid um I don’t usually bother using those um because we can do it

at home and get the electricity from the solar panels” (Interview 34, Tesla Model S)

Some drivers had no problem paying to charge their vehicle or had a maximum amount they would pay

to charge in public. Some find it useful to have a payment system to increase the chance of finding a

vacant charging station. A few PHEVs mention the emissions offset from driving on electric while others

want to support the EVSE companies by charging when they can.

“If you have to pay, then there’s better turnover. And I can usually find a place to charge.” (Interview 12,

Ford Fusion)

“We always have gas but I hate oil. So, we try to avoid it as much as we can.” (Interview 38, Ford C-Max)

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews 138

Figure 116. How often interviewees reported charging their vehicles in public (N=40).

Workplace Charging

Workplace charging is defined as where people charge while they are working. Some workplace

charging locations are in public lots; interviewees perceive this as workplace charging as it is where they

park their car while at work. This definition is consistent with Lee et al. (Lee, J.H. 2020) [5]. Seventeen

interviewees reported always charging their vehicle at work, six charged sometimes, five had the option

to charge at work and chose not to, and seven had no access to charging at work. An additional nine

interviewees were either retired, did not commute, or had an irregular driving schedule with no regular

commute location so they cannot charge at work (Figure 117).

Seventeen interviewees reported charging their vehicles at work whenever there is an available charging

station. The vehicles are ten Chevrolet Volts, two Ford C-Maxes, two Tesla Model S, one Ford Fusion,

one Nissan Leaf, and one Chevrolet Bolt. The main motivator to charge at work was cost-savings and

convenience. Free workplace charging meant that some interviewees prioritized charging at work over

charging at home. Ten MUD drivers (of 11 interviewed) always charge at work. Some drivers reported

wanting to charge every day but were unable to because of charging station congestion. Some reported

driving in earlier or later (after lunch when others would move their vehicle) to ensure they would be

able to access a charging station.

“It was kind of in the middle of a big parking garage that was a little bit further away, um but that was

totally fine” (Interview 10, Chevrolet Volt)

“I charge at work because it’s free. It’s convenience.” (Interview 11, Tesla Model S)

Six interviewees in the sample charged at work sometimes. These interviewees include two Tesla Model

S, one Nissan Leaf, one Ford Fusion, one Ford C-Max, and one Toyota RAV4 EV owning households.

These drivers used workplace charging as secondary to home charging. They would sometimes charge at

work because they needed additional charging because they forgot to charge at home or drove

additional miles beyond their normal routine. Some interviewees reported charging at work because the

parking space was in a better location than the non-charging spaces.

“The reason we’d charge at work is that someone forgets to plug it [the household’s BEV] in it at night.

That's the primary reason like, “oh!” Or if she's running errands during the day and just needs a quick

charge to get home.” (Interview 09, Nissan Leaf)

Five interviewees do not charge at work; the vehicles owned by these interviewees are three Tesla

Model S, one Nissan Leaf, and one Ford C-Max. Some drivers find workplace charging inconvenient

7 7

Always Frequently Occasionally Rarely Never

Count

Public Charging Frequency

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews 139

because the EVSE was too far from their office, due to cost, or because the charging management

strategies did not align with their schedule. One PHEV driver chose not to charge at work to keep the

spaces open for BEVs owners who may need to charge.

“[Work] has charging stations, but it’s a dollar an hour, and you have to move your car every 2 hours so

that's very inconvenient.” (Interview 33, Tesla Model S)

Seven interviewees reported being unable to charge at work. These include those who own the

following vehicles: one Fiat 500e, three Chevrolet Volts, two Ford C-Maxes, and one Tesla Model S.

Reasons for this include having no charging infrastructure installed or parking being too much of a luxury

to dedicate to charging at their workplace.

“We don’t have any outdoor outlets on our building. And even if we did, you would have to drape the

cord across sidewalks and planters, across parking lots, so, um, it’s just not practical and not worth it.”

(Interview 22, Chevrolet Volt)

Nine interviewees do not commute, have an irregular commute, or are retired. These people are

excluded from the workplace sample because their travel patterns do not allow them to charge at a

workplace.

Figure 117. Break down of interviewee's work charging patterns and availability. Nine interviewees

did not have a regular commute or were retired. (N=44)

Workplace Charging Management Strategies

Twenty-three (58%) interviewees reported charging at work either every day or sometimes while five

people said they have the option to charge at work, but do not (Figure 118). Twenty-six of these 28

interviewees provided information on charging management strategies at their workplace charging

location. From these interviewees, three categories of approaches were detected. We refer to these as

‘authoritative charging management strategies’, ‘collective charging management strategies’, and

‘unmanaged charging’.

Doesn't

Drive/No

Regular

Commute

No -

Doesn't

want to

No - Can't Yes -

Sometimes

Yes-

Always

Count

Workplace Charging Frequency

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews

140

Authoritative Management Strategies

Twelve interviewees reported having a set of rules enforced and created by their workplace, which we

call ‘authoritative charging management strategies’. These are rules that have been applied to

workplace charging by the company, without input from the charging station users.

Reported authoritative management strategies include digital queuing, rotation, a time limit with

pricing, and valet charging. Digital queuing occurs when one vehicle is charging and a second parks next

to it waiting to charge. This required having more parking sports surrounding the EVSE than there were

ports. The driver of the second vehicle taps a card (in this case a ChargePoint card) against the reader to

get in the virtual queue to use the charging station. When the first vehicle has finished charging, the

second driver is notified via email that they can unplug the first vehicle and plug in their vehicle.

Rotation is defined as the company forcing the driver to move their vehicle after a time limit or the

vehicle has completed its charge; this information is usually conveyed in the form of signs or digital

messaging. Non-compliance can lead to a penalty. These are not independent strategies: one workplace

could employ multiple strategies.

“[There is] a rotation system where you, since there are more cars than there are charging spots, park

next to charging station and you badge in and then whenever the car next to you is fully charged you get

notified and then you go down and you switch the charger.” (Interview 01, Chevrolet Volt)

Time limit with pricing is when the system is set up to be free or exceptionally low cost, and after a set

number of hours, the price ramps up significantly.

“If you keep your car charged for longer than 3 hours, then they uh start charging you like 5 bucks an

hour. So that’s ah, it’s a real- it’s a strong incentive to move your car and let other people use it.”

(Interview 19, Chevrolet Volt)

Valet charging was observed in workplaces that already had valet parking for their employees with EVSE

in the lot. The valets are responsible for charging the vehicles and rotating them, as necessary. The two

interviewees who had workplace valet charging were less aware of how their vehicle’s charging was

being managed because they were not physically involved in the charging process. These interviewees

were indifferent about how their vehicle was charged so long as it was ready for them upon leaving

work. For workplaces in this sample who had valet parking for their employees, the charging was

managed by the valet, and was paid.

“Valets are good about rotating the cars around, so everybody gets a charge and like, I’m usually

charged up full by the time I go home” (Interview 18, Ford C-Max)

Eighteen (64%) workplaces required joining a charging network company before being able to charge.

Requiring employees to have an EVSE membership (e.g. ChargePoint) does not qualify as authoritative

management because no mechanisms to increase charger utilization were in place.

Collective Management Strategies

Four interviewees stated they and their fellow co-workers created rules for charging at work, known

here as ‘collective management strategies’. These systems are organized by the employees who drive

and charge their PEVs at work. These interviewees’ workplaces did not have any formal rules in place

(i.e. Unmanaged Charging). Examples of collective charging management strategies include day

restrictions (e.g. only being able to charge on Mondays and Wednesdays), time of day restrictions (e.g. a

4-hour limit on charging), messaging an email group when your vehicle is done charging, and well-

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews

141

maintained spreadsheets of vehicle owner contact information which could be used to request someone

to move their vehicle. The vehicles in this category are two Chevrolet Volts and two Nissan Leafs.

Collective charging is only functional if there is a strong intent of commitment and leadership to get

everyone’s vehicle charged with a shared resource. One interviewee reported that their office’s “leader”

left the company, and it was proving difficult to pick up the pieces of their charging patterns. This

technique only arises when there is unmanaged charging at work, and the collective wants to do

something about it.

“I’m part of that google doc, so you know you have to coordinate with another person, go down and

meet them, swap cars, um you know and then you have to check the doc ‘cause someone may say I

want it when you’re done.” (Interview 20, Nissan Leaf)

One interviewee (quoted below) stood out because they were collectively charging from an 110V outlet.

Due to the inefficient time required to charge at this power level, it was it was unexpected that co-

workers would justify time to create a schedule for them all to charge. This interviewee does live in an

MUD without home-based charging, so this is one way he is able to charge his Chevrolet Volt.

“We just kind of had a meeting and said okay how can we best handle this without fighting for it?”

(Interview 24, Chevrolet Volt)

Unmanaged Charging

Ten (38%) drivers reported no organization or rules for charging at work; we have classified these as

‘unmanaged charging.’ Drivers report there is often competition to charge and sometimes cannot

charge if they arrive at work too late in the day. Two interviewees recently changed jobs and went from

an authoritative management strategy to unmanaged workplace charging. They both reported

frustrations with the lack of management for the EVSE. One even asked HR for some sort of managed

charging but had not heard back.

“It’s just like, you grab it, and it’s yours all day. If you’re going to be really nice you could move the car

out and like, move to a different spot, but I usually don't have time during the day to do that.”

(Interview 10, Chevrolet Volt)

“Over the course of the past few years, as more electric vehicles have come to [work], it’s become a

little bit more challenging to make sure you get charge” (Interview 40, Tesla Model S)

Five interviewees report that they charge by bringing their own charging cord to plug into an 110V or

240V outlet at the workplace. Five people have charging at work with installed charging infrastructure

that has no rules. Vehicles here include two Chevrolet Volts, three Tesla Model S, one Ford Fusion, one

Nissan Leaf, one Chevrolet Bolt, and two Ford C-Maxes.

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews

142

Figure 118. Three interviewees did not go into enough detail about their workplace charging

management strategies to categorize (n=26)

sInterviews Discussion

Charging management strategies appear to be effective in increasing vehicle turnover at workplace

EVSE. With higher turnover, more people can use the stations thus increasing their utilization and

availability. This allows more PEVs to charge and reduce the number of EVSE a workplace may need to

install. Most interviewees liked their workplace’s authoritative charging management strategies. Four

BEV owners chose to not charge at work. One has a collective charge management (messaging and

moving vehicle when done charging), one has authoritative charge management (priced charging), and

two did not give enough detail to classify (their lack of knowledge of the system indicative of the fact

they never charge at work). Drivers can complete their travel without using the EVSE which leaves the

stations available for others to use. The rules for those charging stations have the intended effect of

increasing availability of EVSE and decreasing congestion. Decreasing congestion leads to a more reliable

charging experience because drivers are more likely to be able to charge their vehicles when needed.

Had they charged when they did not need to, they would have created unnecessary congestion to the

workplace EVSE; charging management strategies are mitigating potential congestion. A few drivers

stated that if the management strategies aligned better with their work habits, they would charge at

work instead of or in addition to home.

Interviewees who were part of collective charging management systems also reported positive

experiences. They liked designing their own rules to fit the needs to the drivers. These systems also

increased the number of vehicles that could be serviced by the limited number of EVSE. Collective

strategies were only found at workplaces with a small number of stations and PEV drivers; this could be

because in a smaller office there can be more of a community mindset where the drivers interact more

regularly. Collective strategies, however, may not be intuitive to newcomers and could be impeded by

unaware drivers. Even a strategy designed by the drivers, for the drivers might not necessarily work for

everyone.

For those without any charging management rules in place, drivers may not always be able to charge

when they want or need. These workplaces may have underutilized charging stations, not necessarily

due to high charging congestion, but a high charging station stagnation as the drivers have no incentive

to move their vehicle when charging is complete. By increasing the availability of the EVSE, they can be

used more frequently throughout the workday.

Authoritative Collective Unmanaged

Count

Charging Management Strategy

Workplace Charging and Out of Home Charging: Lessons from In-depth Interviews

143

Interviews Policy Discussion

The charging management strategies presented in this section can be a good starting place for how

companies can offer charging to their workforce. Without enforced charging management, stagnation

and congestion can occur at the charging stations reducing the number of PEVs that charging stations

can support. This in turn increases the number of EVSE that a workplace would need to install, which can

increase costs.

From the interviewees’ perceptions, it appears that digital queuing and time limit with pricing are the

best received. With digital queuing, if the EVSE cord can reach multiple parking spots, then drivers do

not have to move vehicles during the workday and have to re-park the vehicle. Time limit with pricing

still allows drivers to obtain reasonable amount of SOC within the free time limit and motivates drivers

to move their vehicle.

When designing the infrastructure layout, digital queuing should be kept in mind. We recommend

considering more parking spaces around the EVSE and charge cords that can reach multiple parking

spaces. This coupled with an electronic queue means drivers do not have to move the PEV when it has

done charging, and others can plug their vehicle in when one is finished. A rotational system like this can

reduce congestion and allow more vehicles throughout the day to charge. Locating chargers away from

the front of the building can solve the issue of PEV owners charging just to get a good parking location.

Keeping the stations visible, however, can help create awareness of the available charging

infrastructure.

Another EVSE layout is to install slower charging stations in the front of the building and faster charging

stations farther away; this can create an additional soft barrier from charging to save the charging

stations for people who make need to charge for a longer time or have a larger battery to refuel. An

alternative is to install costly faster chargers in the front of the building, and many free slower charging

stations toward the back of the lot. People who need to refuel quickly can spend the money to charge

swiftly and maintain a nice parking spot, but those who are just taking advantage of free electricity can

choose to park further back.

For workplaces with existing unmanaged charging we recommend adding enforceability, cost, and/or

idle fee components to the EVSE to spur vehicle turnover. Adding a reasonable cost to charging vehicles

may help alleviate congestion from vehicles who otherwise could charge at home. A reasonable rate,

such as a few cents over the cost of electricity in the area (at home), would in theory not dissuade MUD

drivers who rely on workplace charging, since the electricity price would be cost comparable to what

they would pay if they had access at home. The threat of banning people from charging if the rules are

not followed may also be a strong motivator to honor the guidelines. Idle fees or a steep fee increase

when the car is done are particularly useful turnover techniques because the driver would be monetarily

charged when not receiving any electrical charge.

These charging strategies are intended for workplaces whose employees park locally. A driver who parks

at a public transit station, for example, would struggle to move their vehicle during the workday. The

ideal charging management strategy for those parking lots would be on a per kWh or on a per charging

session paid scheme, this may mean drivers will only charge when they need to do so. Making sure any

charging strategy design is simple and easy for the drivers to use is important.

Conclusions

144

8 Conclusions

This report encompasses the findings of two projects, presents the collection of data from alternative

fuel vehicles in California, and incorporates the largest independent objective source reporting on

vehicles’ performance. The project includes five years of data collection from different vehicles that

have no standardized protocol for on-board data reporting. Over these five years, loggers were installed

on about 800 vehicles, including ZEVs and ICEVs. The result is the collection of 7 million miles of data,

including 4.3 million miles that are collected from alternative fuel vehicles. The data includes second by

second observations of energy consumption, driving conditions, charging engine operations, and more,

which can serve as the basis for many more scientific publications and reports in the future. Results

from this study provide insights on first and second generation PEVs and FCVs as well as the

environmental impacts of their battery size, range, driving behavior, and charging/refueling behavior.

The data collected shows that longer-range BEVs and PHEVs with larger batteries had a greater

substitution of gasoline miles with electric miles. BEVs, PHEVs, and FCVs are used for commuting at

higher rates than those of the rest of the fleet. In addition, these vehicles tend to account for a higher

share of miles than that of other household vehicles, however, FCVs and small, short-range BEVs are

rarely used for long road trips. Among longer-range BEVs, the Model S was used for a higher proportion

of freeway driving and utilized DC fast charging far from home much more frequently than the Bolt-60

was used in a comparable manner. Other than battery size variations, the difference in behavior of these

two long-range BEVs can partly be attributed to the fact that the Model S has a wider available DC fast

charging network in California. There are three types of DC Fast charging available in the state to date:

Combined Charging System (CCS), CHAdeMO and Tesla Supercharger. While the Bolt-60 is only capable

of charging via CCS, the Model S can charge using a Tesla supercharger or CHAdeMO with an adapter.

According to the Alternative Fuels Data Center, there are approximately 1000 Tesla supercharger and

CHAdeMO public charging stations currently available in California compared to just around 780 public

CCS stations, making DC fast charging far more accessible to the Tesla than to the Bolt-60. Furthermore,

the Model S is larger than the Bolt-60 and therefore, is more likely to be used for long distance trips.

Most modeling and early assumptions hypothesized that electric vehicle drivers would plug-in every

night and start each day with a full battery, but our results show that charging every other night is more

common for longer range BEVs or when driving less, while charging more than once a day is common for

PHEVs. On average, the BEVs in our study charge less than once a day, including days when the vehicle

was not used as expected. PHEVs in the study had lower utility factor values than found in EPA results,

because of driving more and faster than expected. PHEV drivers with larger batteries charge their

vehicles more when needed, achieve higher utility factors, and have many days with no engine starts at

all.

Home charging and level 2 charging are the main sources of energy. DC fast charging is mostly near

home, and only Tesla vehicles use DC fast charging often for longer trips. Level 1 charging was also

significant for long, overnight trips. A sizable portion of users start charging events at midnight,

regardless of utility rates. The interviews highlight the importance of charging policy and charging

management in the workplace, where improving charger congestion, convenience, and dependability

could increase usage.

The fuel cell vehicles in our study are mostly used in metropolitan areas where fueling stations exist. In

interviews, FCV users report adjusting their driving needs based on long term experience with

infrastructure reliability. FCV users do not drastically change their driving behavior in situations where

infrastructure is unreliable, such as a station being out of order or there being an energy shortage.

Conclusions

145

Overall, longer-range PEVs have more electrified miles than their shorter-range counterparts, resulting

in a reduced greenhouse gas (GHG) footprint. However, to maximize the benefits of PEVs, a full set of

policies is needed to address charging behaviors and vehicle purchases. This study focuses on ZEV

performance and household performance but did not collect data to compare those households to the

general population and ICEV-only households. The results of this study address possible factors that

affect the environmental impact of ZEVs. As those factors continue to change over time, ongoing

research is necessary to better shape the policies that lead to more sustainable transportation and

efficient ZEV usage.

Glossary

146

9 Glossary

AE all electric (a mode of PHEVs)

AER all-electric range

BEV battery electric vehicle

CDB charge depleting blend

CS charge sustaining

DCFC DC fast charger

eVMT

FCV

electric vehicle miles traveled

fuel cell electric vehicle

GHG greenhouse gas

gVMT gasoline vehicle miles traveled

HDD habitual driving distance

HH household

HOV high occupancy vehicle

ICEV internal combustion engine vehicle

L1 Level 1 (refers to type of charger)

L2 Level 2 (refers to type of charger)

LDT long distance travel

MPG miles per gallon

MPGe miles per gallon equivalent

MY model year

PEV plug-in electric vehicle

PHEV plug-in hybrid electric vehicle

SOC state of charge

UF utility factor

Glossary

147

VMT vehicle miles travelled

ZE zero emission

zVMT zero tailpipe emission trip

References

148

10 References

Aviquzzaman, M. (2014). Analysis of Plug-In hybrid Electric Vehicles' utility factors using GPS-based

longitudinal travel data, Lamar University-Beaumont.

Bailey J, Axsen J. (2015). Anticipating PEV buyers’ acceptance of utility controlled charging. Transp Res

Part A Policy Pract, 82, 29–46.

Björnsson, L.-H. and S. Karlsson. (2017). "Electrification of the two-car household: PHEV or BEV?"

Transportation Research Part C: Emerging Technologies, 85, 363-376.

Brecht, P. and J. Orenberg. (2019). 2019-2020 Investment Plan Update for the Alternative and

Renewable Fuel and Vehicle Technology Program. California Energy Commission. Technical Report CEC-

600-2018-005-SD.

BTS. (2017). “National Household Travel Survey Long Distance Travel Quick Facts.”

https://www.bts.gov/statistical-products/surveys/national-household-travel-survey-long-distance-

travel-quick-facts.

BTS. (2018). “Long Distance Transportation Patterns: Mode Choice.” https://www.bts.gov/statistical-

products/surveys/national-household-travel-survey-summer-travel-quick-facts.

Caltrans. (2013). “2010-2012 California Household Travel Survey.”

https://www.nrel.gov/transportation/secure-transportation-data/tsdc-california-travel-survey.html.

Caperello N, Kurani KS, Tyree-Hageman J. (2013) Do You Mind if I Plug-in My Car? How etiquette shapes

PEV drivers’ vehicle charging behavior. Transp Res Part A Policy Pract. Epub ahead of print. DOI:

10.1016/j.tra.2013.07.016.

CARB. (2017). California's Advanced Clean Cars Midterm Review, Appendix H: Plug-in Hybrid Electric

Vehicle Emissions Testing

CARB. (2017). California's Advanced Clean Cars Midterm Review, Appendix I: Alternative Credits for Zero

Emission Vehicles and Plug-in Hybrid Electric Vehicles.

CARB. (2017). “California Climate Investments Quantification Methodology Emission Factor

Database.”https://www.arb.ca.gov/cc/capandtrade/auctionproceeds/cci_emissionfactordatabase.xlsx?_

ga=2.192111850.1503633160.1549935127-1695908195.1547853262.

CARB. (2018). EMFAC2017 Volume III - Technical Documentation, V1.0.2.

Chakraborty D, Bunch DS, Lee JH, et al. (2019). Demand drivers for charging infrastructure-charging

behavior of plug-in electric vehicle commuters. Transp Res Part D Transp Environ, 76, 255–272.

Contestabile, M., M. Alajaji and B. Almubarak. (2017). "Will current electric vehicle policy lead to cost-

effective electrification of passenger car transport?" Energy Policy, 110, 20-30.

Crain, T., T. Gorgia and R. J. Alley. (2016). Application of PHEV Fractional Utility Factor Weighting to

EcoCAR On-Road Emissions and Energy Consumption Testing, SAE Technical Paper.

Davies, J. and K. S. Kurani. (2013). "Moving from assumption to observation: Implications for energy and

emissions impacts of plug-in hybrid electric vehicles." Energy Policy, 62, 550-560.

Davies, J., and K. Kurani. (2010). Recharging Behavior of Households' Plug-In Hybrid Electric Vehicles:

Observed Variation in Use of Conversions of 5-kW-h Blended Plug-In Hybrid Electric Vehicle.

Transportation Research Record: Journal of the Transportation Research Board, 2191, 75-83.

References

149

Davis, L. W. (2019). "How much are electric vehicles driven?" Applied Economics Letters, 26(18),1-6,

1497-1502.

Debapriya, C., Scott, H., & Gil, T. (2020). Why do some consumers not charge their plug-in hybrid

vehicles? Evidence from Californian plug-in hybrid owners. Environmental Research Letters.

Delmonte E, Kinnear N, Jenkins B, et al. (2020). What do consumers think of smart charging? Perceptions

among actual and potential plug-in electric vehicle adopters in the United Kingdom. Energy Res Soc Sci,

60, 101318.

Dimitropoulos, A., P. Rietveld, and J. N. Van Ommeren. (2013). "Consumer valuation of changes in

driving range: A meta-analysis." Transportation Research Part A: Policy and Practice, 55, 27-45.

DOE/AFDC. (2017). "Electric Vehicle Benefits and Considerations." Retrieved July 17, 2018, from

https://www.afdc.energy.gov/fuels/electricity_benefits.html.

Dong, J., Liu, C., Lin, Z. (2014). Charging infrastructure planning for promoting battery electric vehicles:

An activity-based approach using multiday travel data. Transp. Res. Part C Emerg. Technol, 38, 44–55.

https://doi.org/10.1016/j.trc.2013.11.001

Federal Highway Administration. (2017). 2017 National Household Travel Survey, U.S. Department of

Transportation, Washington, DC. Available online: https://nhts.ornl.gov.

FHWA. (2017). Summary of Travel Trends: 2017 National Household Travel Survey.

Glat

ting, Gerhard, Peter Kletting, Sven N Reske, Kathrin Hohl, and Christina Ring. (2007). “Choosing the

Optimal Fit Function: Comparison of the Akaike Information Criterion and the F

-t

est.” Medical Physics,

34(11), 4285–92.

Grah

am-Rowe E, Gardner B, Abraham C, et al. (2012). Mainstream consumers driving plug-in battery-

electric and plug-in hybrid electric cars: A qualitative analysis of responses and evaluations. Transp Res

Part A Policy Pract, 46, 140–153.

Hardman, S., Shiu, E., Steinberger-Wilckens, R., & Turrentine, T. (2017). Barriers to the adoption of fuel

cell vehicles: A qualitative investigation into early adopters attitudes. Transportation Research Part A:

Policy and Practice, 95, 166–182. https://doi.org/10.1016/j.tra.2016.11.012

Hardman, S., Jenn, A., Tal, G., Axsen, J., et al. (2018). A review of consumer preferences of and

interactions with electric vehicle charging infrastructure. Transportation Research Part D: Transport and

Environment, 62, 508–523. https://doi.org/10.1016/j.trd.2018.04.002

Hinds, Brett. (2017). “Ford Pushing for Common Industry Metrics to Understand Customer Daily Electric

Driving Requirements.” https://www.greencarcongress.com/2017/02/20170213-ford.html.

Idaho National Laboratory. (2015). How Americans Charge Their Electric Vehicles Findings from the

largest plug-in electric vehicle infrastructure demonstration in the world.

International Energy Agency. (2018). "Global EV Outlook 2018-Towards Cross-Modal Electrification."

Retrieved July 17, 2018, from https://www.iea.org/gevo2018/.

Ji, W., Nicholas, M., Tal, G. (2015). Electric vehicle fast charger planning for metropolitan planning

organizations: Adapting to changing markets and vehicle technology. Transp. Res. Rec., 2502, 134–143.

https://doi.org/10.3141/2502-16

Kunzmann, M. and V. Masterman. (2013). California Household Travel Survey (CHTS)

References

150

Lee JH, Chakraborty D, Hardman SJ, et al. (2020). Exploring electric vehicle charging patterns: Mixed

usage of charging infrastructure. Transp Res Part D Transp Environ, 79, 102249.

Liao, F., E. Molin, and B. van Wee. (2017). "Consumer preferences for electric vehicles: a literature

review." Transport Reviews, 37(3), 252-275.

Lin, Zhenhong, Dong, Changzheng Liu, and David Greene. (2012). “Estimation of Energy Use by Plug-in

Hybrid Electric Vehicles - Validating Gamma Distribution for Representing Random Daily Driving

Distance.” Transportation Research Record. https://doi.org/10.3141/2287-05.

Lloro, Alicia, and David Brownstone. (2018). “Vehicle Choice and Utilization: Improving Estimation with

Partially Observed Choices and Hybrid Pairs.” Journal of Choice Modelling.

https://doi.org/10.1016/j.jocm.2018.05.005.

Loveday, S. (2018). "May 2018 Plug-In Electric Vehicle Sales Report Card." Retrieved July 17, 2018, from

https://insideevs.com/may-2018-plug-in-electric-vehicle-sales-report-card/.

Lutsey, N., D. Meszler, A. Isenstadt, J. German, and J. Miller. (2017). "Efficiency Technology and Cost

Assessment for US 2025–2030 Light-Duty Vehicles." International Council on Clean Transportation

(ICCT), Washington, USA, White Paper, 22nd March.

McGuckin, N. and A. Fucci. (2017). Summary of Travel Trends: 2017 National Household Travel Survey,

Federal Highway Administration.

McLaren, J., J. Miller, E. O’Shaughnessy, E. Wood and E. Shapiro. (2016). "Emissions associated with

electric vehicle charging: Impact of electricity generation mix, charging infrastructure availability, and

vehicle type." NREL (National Renewable Energy Laboratory (NREL), Golden, CO (United States)), Tech.

Rep.

Muratori, M. (2018). "Impact of uncoordinated plug-in electric vehicle charging on residential power

demand." Nature Energy, 3(3), 193.

Nicholas, M., G. Tal, and T. Turrentine. (2017). Advanced Plug-in Electric Vehicle Travel and Charging

Behavior Interim Report, Research Report–UCD-ITS-RR-16.

Nicholas M, Tal G. (2015). Charging for Charging at Work: Increasing the Availability of Charging Through

Pricing. Inst Transp Stud.

Nicholas, Michael, Gil, Tal, and Thomas S. Turrentine. (2017). “Advanced Plug-in Electric Vehicle Travel

and Charging Behavior Interim Report Advanced Plug in Electric Vehicle Travel and Charging Behavior

Interim Report.” International Energy Agency (IEA).

Pham, A., and M. Jeftic. (2018). Characterization of Gaseous Emissions from Blended Plug-In Hybrid

Electric Vehicles during High-Power Cold-Starts, SAE Technical Paper.

Plötz, Patrick, Niklas Jakobsson, and Frances Sprei. (2017). “On the Distribution of Individual Daily

Driving Distances.” Transportation Research Part B: Methodological, 101, 213–27.

Plötz, Patrick, Simon Árpád Funke, and Patrick Jochem. (2018). “The Impact of Daily and Annual Driving

on Fuel Economy and CO2 Emissions of Plug-in Hybrid Electric Vehicles.” Transportation Research Part A:

Policy and Practice, 118, 331–40.

PSRC TCS. (2006). 2004–2006 Puget Sound Traffic Choices Study.

Raghavan, S. S. and G. Tal. (2019). Characterizing Plug-In Electric Vehicle Driving and Charging Behavior:

Observations from a Year Long Data Collection Study.

References

151

SAE. (2010). “J2481_201009: Utility Factor Definitions for Plug-In Hybrid Electric Vehicles Using Travel

Survey Data.” https://www.sae.org/standards/content/j2841_201009/.

Santini, D.J., Zhou, Y.F., Elango, V., Xu, Y., Guensler, R. (2014). Daytime Charging: What Is the Hierarchy

of Opportunities and Customer Needs? Case Study Based on Atlanta Commute Data.

Schönfelder, S., & Axhausen, K. W. (2003). Activity spaces: Measures of social exclusion? Transport

Policy, 10(4), 273–286. https://doi.org/10.1016/j.tranpol.2003.07.002

Schubert, E., Sander, J., Ester, M., Kriegel, H.-P., & Xu, X. (2017). DBSCAN Revisited, Revisited: Why and

How You Should (Still) Use DBSCAN. ACM Transactions on Database Systems (TODS), 42(3).

http://delivery.acm.org/10.1145/3070000/3068335/a19-schubert.pdf

Sherman, J. E., Spencer, J., Preisser, J. S., Gesler, W. M., & Arcury, T. A. (2005). A suite of methods for

representing activity space in a healthcare accessibility study. International Journal of Health

Geographics, 4(1), 24. https://doi.org/10.1186/1476-072X-4-24

Smart, J. and S. Schey. (2012). "Battery electric vehicle driving and charging behavior observed early in

the EV project." SAE International Journal of Alternative Powertrains 1(2012-01-0199), 27-33.

Smart, J., T. Bradley, and S. Salisbury. (2014). "Actual versus estimated utility factor of a large set of

privately owned Chevrolet Volts." SAE International Journal of Alternative Powertrains 3(1), 30-35.

Smart, J., T. Bradley, and S. Salisbury. (2014). Actual versus estimated utility factor of a large set of

privately owned Chevrolet Volts. SAE International Journal of Alternative Powertrains, 3(1), 30-35.

Smart, J., W. Powell, and S. Schey. (2013). Extended range electric vehicle driving and charging behavior

observed early in the EV project, SAE Technical Paper.

Smith, Mary S, and Thomas A Butcher. (2008). “How Far Should Parkers Have to Walk?” Parking 47(4).

State and Federal Electric Vehicle Incentives | Clean Vehicle Rebate Project,

https://cleanvehiclerebate.org/eng/ev/incentives/state-and-federal (accessed 16 March 2020).

State of California. (2018). “Governor Brown Takes Action to Increase Zero-Emission Vehicles, Fund New

Climate Investments.” Governor Edmund G Brown Jr, www.ca.gov/archive/gov39/2018/01/26/governor-

brown-takes-action-to-increase-zero-emission-vehicles-fund-new-climate-investments/index.html.

Tal, Gil., Dunckley, J., (2016). Plug-In Electric Vehicle Multi-State Market and Charging Survey, in: EVS29.

Tal, Gil, Michael A Nicholas, Jamie Davies, and Justin Woodjack. (2014). “Charging Behavior Impacts on

Electric Vehicle Miles Traveled: Who Is Not Plugging In?” Transportation Research Record, 2454(1), 53–

60.

Tamor, Michael A, Chris Gearhart, and Ciro Soto. (2013). “A Statistical Approach to Estimating

Acceptance of Electric Vehicles and Electrification of Personal Transportation.” Transportation Research

Part C: Emerging Technologies, 26, 125–34.

Tamor, Michael A, Paul E Moraal, Briana Reprogle, and Miloš Milačić. (2015). “Rapid Estimation of

Electric Vehicle Acceptance Using a General Description of Driving Patterns.” Transportation Research

Part C: Emerging Technologies 51, 136–48. internal-pdf://210.174.249.38/Rapid estimation of electric

vehicle acceptanc.pdf.

Tukey, J. W. (1977). Exploratory data analysis, 2, 131-160.

Turrentine, T. and G. Tal. (2015). Advanced Plug-In Electric Vehicle Usage and Charging Behavior.

Contract #12-319. California, CARB and CEC.

References

152

Turrentine, Thomas, Gil Tal, and David Rapson. (2018). “The Dynamics of Plug-in Electric Vehicles in the

Secondary Market and Their Implications for Vehicle Demand, Durability, and Emissions.” Institute of

Transportation Studies. https://rosap.ntl.bts.gov/view/dot/36609

TxDOT. (2015). 2002-2015 Texas Regional Travel Surveys.

Weiller, C. (2011). Plug-in hybrid electric vehicle impacts on hourly electricity demand in the United

States. Energy Policy, 39, 3766–3778. https://doi.org/10.1016/j.enpol.2011.04.005.

Wietschel, M., P. Plötz, A. Kühn and T. Gnann. (2013). "Market evolution scenarios for electric vehicles."

Summary report. Fraunhofer Institute for Systems and Innovation Research ISI: Karlsruhe. Retrieved

April 11, 2014.

Wood, E. W., C. L. Rames, A. Bedir, N. Crisostomo, and J. Allen. (2018). California Plug-In Electric Vehicle

Infrastructure Projections: 2017-2025-Future Infrastructure Needs for Reaching the State's Zero

Emission-Vehicle Deployment Goals, National Renewable Energy Lab. (NREL), Golden, CO (United

States).

Yang, Yong, and Ana V. Diez-Roux. (2012). “Walking Distance by Trip Purpose and Population

Subgroups.” American Journal of Preventive Medicine. https://doi.org/10.1016/j.amepre.2012.03.015.

Zenk, S. N., Matthews, S. A., Kraft, A. N., & Jones, K. K. (2018). How many days of global positioning

system (GPS) monitoring do you need to measure activity space environments in health research?

Health & Place, 51, 52–60. https://doi.org/10.1016/j.healthplace.2018.02.004

Appendix A

153

11 Appendix A

Appendix 1 is a list of academic papers that uses data collected, by survey or loggers, as part of the

Emerging Technology Zero Emission Vehicle Household Travel and Refueling Behavior (CARB Contract

16RD009) and Advanced Plug-in Electric Vehicle Travel and Charging Behavior project (CARB Contract

12-319).

1. Plug-in hybrid electric vehicle observed utility factor: Why the observed electrification performance

differ from expectations

Authors: Seshadri Srinivasa Raghavan, Gil Tal

Publication date: 2020/11/13

Journal: International Journal of Sustainable Transportation

Publisher: Taylor & Francis

Description:

Plug-in hybrid electric vehicles (PHEVs) are an effective vehicle technology to reduce light duty vehicle

greenhouse gas emissions and gasoline consumption. They combine all-electric driving capabilities of a

battery electric vehicle with the engine downsizing and fuel economy improvements of a hybrid electric

vehicle. Their environmental performance is predicated upon the metric utility factor (UF). It is formally

defined in the Society of Automotive Engineers J2841 standard and denotes the fraction of vehicle miles

traveled (VMT) on electricity (eVMT). Using year-long driving and charging data collected from 153

PHEVs in California with 11–53 miles range, this article systematically evaluates what aspects of driving

and charging behavior causes observed UF to deviate from J2841 expectations. Our analyses indicated

that charging behavior, distribution of daily VMT, efficiency of electrical energy…

2. Why do some consumers not charge their plug-in hybrid vehicles? Evidence from Californian plug-in

hybrid owners

Authors: Debapriya Chakraborty, Scott Hardman, Gil Tal

Publication date: 2020/4/23

Journal: Environmental Research Letters

Publisher: IOP Publishing

Description:

The environmental benefits of plug-in hybrid electric vehicles (PHEVs) are closely related to the driving

and charging behavior of vehicle owners. It is often wrongly assumed that PHEV drivers plug-in once per

day. Using data from drivers of the vehicles we show this is not the case and that some drivers rarely

charge their PHEV. If the vehicle is not plugged-in regularly, the vehicle will drive fewer electric miles

and more gasoline miles, thereby losing out on potential emission savings. Analyzing 30-day charging

behavior of 5,418 PHEV owners using a logistic regression model, we explore the factors that influence

driver's decisions to not charge their vehicle. Several factors play a role in drivers' decision to plug-in

their PHEV or not, including vehicle characteristics and the availability and cost of charging at various

locations. Higher home electricity prices, lower electric driving range, lower electric motor power to …

Appendix A

154

3. Influence of User Preferences on the Revealed Utility Factor of Plug-In Hybrid Electric Vehicles

Authors: Seshadri Srinivasa Raghavan, Gil Tal

Publication date: 2020/3

Journal: World Electric Vehicle Journal

Volume:11

Issue: 1

Pages: 6

Publisher: Multidisciplinary Digital Publishing Institute

Description:

Plug-in hybrid electric vehicles (PHEVs) are an effective intermediate vehicle technology option in the

long-term transition pathway towards light-duty vehicle electrification. Their net environmental impact

is evaluated using the performance metric Utility Factor (UF), which quantifies the fraction of vehicle

miles traveled (VMT) on electricity. There are concerns about the gap between Environmental

Protection Agency (EPA) sticker label and real-world UF due to the inability of test cycles to represent

actual driving conditions and assumptions about their driving and charging differing from their actual

usage patterns. Using multi-year longitudinal data from 153 PHEVs (11–53 miles all-electric range) in

California, this paper systematically evaluates how observed driving and charging, energy consumption,

and UF differs from sticker label expectations. Principal Components Analysis and regression model

results indicated that UF of short-range PHEVs (less than 20-mile range) was lower than label

expectations mainly due to higher annual VMT and high-speed driving. Long-distance travel and high-

speed driving were the major reasons for the lower UF of longer-range PHEVs (at least 35-mile range)

compared to label values. Enhancing charging infrastructure access at both home and away locations,

and increasing the frequency of home charging, improves the UF of short-range and longer-range

PHEVs, respectively. View Full-Text

4. Exploring electric vehicle charging patterns: Mixed usage of charging infrastructure

Authors: Jae Hyun Lee, Debapriya Chakraborty, Scott J Hardman, Gil Tal

Publication date: 2020/2/1

Journal: Transportation Research Part D: Transport and Environment

Volume: 79

Pages: 102249

Publisher: Pergamon

Description:

This paper examines the charging behavior of 7,979 plug-in electric vehicle (PEV) owners in California.

The study investigates where people charge be it at home, at work, or at public location, and the level of

charging they use including level 1, level 2, or DC fast charging. While plug-in behavior can differ among

PEV owners based on their travel patterns, preferences, and access to infrastructure studies often

generalize about charging behavior. In this study, we explore differences in charging behavior among

different types of PEV owners based on their use of charging locations and levels, we then identify

Appendix A

155

factors associated with PEV owner’s choice of charging location and charging level. We identified socio-

demographic (gender and age), vehicle characteristics, commute behavior, and workplace charging

availability as significant factors related to the choice of charging location.

5. An in-depth examination of electric vehicle incentives: Consumer heterogeneity and changing

response over time

Authors: Alan Jenn, Jae Hyun Lee, Scott Hardman, Gil Tal

Publication date: 2020/2/1

Journal: Transportation Research Part A: Policy and Practice

Volume: 132

Pages: 97-109

Publisher: Pergamon

Description:

We investigate the impacts of a combination of incentives on the purchase decision of electric vehicle

buyers in California from 2010 through 2017. We employ a comprehensive survey on over 14,000

purchasers of electric vehicles in the state of California. The survey covers a swath of purchase

intentions, general demographics, and importance of various incentives. Our results indicate that the

most important incentives for plug-in electric vehicle (PEV) owners are the federal tax credit, the

California state rebate, and high occupancy vehicle (HOV) lane access. In addition, the importance of the

incentives and their associated effect on purchase behavior has been changing over time: respondents

are less likely to not change their decision and more likely to not buy a vehicle at all as time passes and

the technology moves away from early adopters. Incentives are becoming more important for vehicle

adopters as PEV …

6. Factors Affecting Demand for Plug-in Charging Infrastructure: An Analysis of Plug-in Electric Vehicle

Commuters

Authors: Gil Tal, Debapriya Chakraborty, Alan Jenn, Jae Hyun Lee

Publication date: 2020

Description:

The public sector and the private sector, which includes automakers and charging network companies,

are increasingly investing in building charging infrastructure to encourage the adoption and use of plug-

in electric vehicles (PEVs) and to ensure that current facilities are not congested. However, building

infrastructure is costly and, as with road congestion, when there is significant uptake of PEVs, we may

not be able to “build out of congestion.” We modelled the choice of charging location that more than

3000 PEV drivers make when given the options of home, work, and public locations. Our study focused

on understanding the importance of factors driving demand such as: the cost of charging, driver

characteristics, access to charging infrastructure, and vehicle characteristics. We found that differences

in the cost of charging play a key role in the demand for charging location. PEV drivers tend to substitute

workplace charging for home charging when they pay a higher electricity rate at home, more so when

the former is free. Additionally, socio-demographic factors like dwelling type and gender, as well as

vehicle technology factors like electric range, influence the choice of charging location.

Appendix A

156

7. Demand drivers for charging infrastructure-charging behavior of plug-in electric vehicle commuters

Authors: Debapriya Chakraborty, David S Bunch, Jae Hyun Lee, Gil Tal

Publication date: 2019/11/1

Journal: Transportation Research Part D: Transport and Environment

Volume: 76

Pages: 255-272

Publisher: Pergamon

Description:

The public as well as the private sector that includes automakers and charging network companies are

increasingly investing in building charging infrastructure to encourage the adoption and use of plug-in

electric vehicles (PEVs) as well as to ensure that current facilities are not congested. However, building

infrastructure is costly and, like road congestion, when there is significant uptake of PEVs we may not be

able to “build out of congestion.” Modelling the choice of charging infrastructure of more than 3000 PEV

drivers who had the opportunity to select among home, work, and public locations, we focus on

understanding the importance of factors driving demand such as: the cost of charging, driver

characteristics, access to charging infrastructure, and vehicle characteristics. We find that differences in

the cost of charging play a significant role in the demand for charging location. PEV drivers tend to

substitute …

8. Incentives for Plug-in Electric Vehicles Are Becoming More Important Over Time for Consumers

Authors: Alan Jenn, Scott Hardman, Jae Hyun Lee, Gil Tal

Publication date: 2019/10/1

Description:

Federal and state governments are offering incentives to those who purchase or lease plug-in electric

vehicles (PEVs), which include both battery electric vehicles (BEVs) and plug-in hybrid electric vehicles

(PHEVs). Policymakers are beginning to consider the phase-out of these incentives and how a phase-out

might impact PEV market growth. This brief highlights new research on the importance of incentives to

consumers over time, based on survey data from 14,000 PEV-owning households in California between

2010 and 2017, collected and analyzed by the Plug-in Hybrid & Electric Vehicle Research Center at UC

Davis. The PEV incentives included as part of this research are those currently available to Californian

consumers: high occupancy vehicle (HOV) lane access; the US federal tax credit, which offers up to

$7,500 to PEV buyers; and the California Clean Vehicle Rebate Project (CVRP), which offers $1,500 for a

PHEV, $2,500 for a BEV, and an additional $2,000 for low-income consumers.

Appendix A

157

9. Who is buying electric vehicles in California? Characterizing early adopter heterogeneity and

forecasting market diffusion

Authors: Jae Hyun Lee, Scott J Hardman, Gil Tal

Publication date: 2019/9/1

Journal: Energy Research & Social Science

Volume: 55

Pages: 218-226

Publisher: Elsevier

Description:

The successful market entry of plug-in electric vehicles (PEVs) is contingent on them being adopted by

consumers, the first of which will be early adopters. The current understanding of these early adopters

is based on small samples of PEV buyers gathered at one point in time. Here we present multi-year

(2012–2017) questionnaire survey data on the socio-demographic profile of 11,037 PEV adopters in

California. Latent class cluster analysis reveals four heterogeneous groups of PEV buyers. 49% are High

income families, 26% Mid/high income old families, 20% Mid/high income young families, and about 5%

are Middle income renters. Using the latent classes as input factors in Bass diffusion models we show

that high income families may not continue to be the largest group of PEV adopters, while high income

families are 49% of the PEV market today, they only represent 3.6% of California households. For …

10. An early look at plug-in electric vehicle adoption in disadvantaged communities in California

Authors: Kathryn Canepa, Scott Hardman, Gil Tal

Publication date: 2019/6/1

Journal: Transport Policy

Volume: 78

Pages: 19-30

Publisher: Pergamon

Description

Prior research on plug-in electric vehicle (PEV) adoption has revealed that early adopters tend to be

wealthy consumers, this may mean that the benefits of PEVs are not being equitably distributed.

Extensive research has shown that low-income and minority commutes are disproportionately impacted

by environmental and transportation injustice. PEVs can contribute to importing air quality and could

provide lower cost and more reliable transportation to low-income and minority communities if they are

deployed there. This paper takes an early quantitative look at PEV adoption in disadvantaged

communities (DACs), which are census tracts in California that suffer from a combination of economic

barriers and environmental burden. We use six datasets to examine PEV market share, socioeconomic

characteristics of PEV owners, and PEV charging infrastructure. Analysis confirms that adoption of both

new and used…

Appendix A

158

11. An Examination of the Impact That Electric Vehicle Incentives Have on Consumer Purchase Decisions

Over Time

Authors: Alan Jenn, Jae Hyun Lee, Scott Hardman, Gil Tal

Publication date: 2019/5/1

Description:

We investigate the impacts of a combination of incentives on the purchase decisions of electric vehicle

(EV) buyers in California from 2010 through 2017. We employ a comprehensive survey on over 14,000

purchasers of EVs in California. The survey covers a range of purchase intentions, general demographics,

and the importance of various incentives. Our results indicate that the most important incentives for

plug-in electric vehicle (PEV) owners are the federal tax credit, the state rebate, and HOV lane access. In

addition, the importance of the incentives and their associated effect on purchase behavior has been

changing over time: respondents are more likely to change their decisions and to not buy a vehicle at all

as time passes and the technology moves away from early adopters.

12. Characterizing Plug-In Electric Vehicle Driving and Charging Behavior: Observations from a Year Long

Data Collection Study

Authors: Seshadri Srinivasa Raghavan, Gil Tal

Publication date: 2019

Source: Transportation Research Board 98th Annual Meeting Transportation Research Board

Issue: 19-03573

Description:

This paper analyzes driving and refueling/charging data using high-resolution GPS for plug-in electric

vehicles (PEV) in households. This will help in understanding the energy efficiency and emission

reduction associated with PEV usage. Consumer perceptions of PEVs’ ability to meet driving needs as

opposed to the ability of internal combustion engine vehicles may contribute to the low market share in

the overall light-duty vehicle market. The results of this study can be used to help policymakers develop

policies that encourage PEV adoption.

13. Exploring the Value of Clean Air Vehicles High Occupancy Lane Access in California

Authors: Wei Ji, Gil Tal

Publication date: 2019

Source: Transportation Research Board 98th Annual Meeting Transportation Research Board

Issue: 19-03755

Description:

This paper estimates the value of the California program that allows single-occupant use of high

occupancy vehicle lanes by plug-in vehicles and to benefit from reduced tolls same as high occupancy

vehicles. A survey was conducted that targeted PEV owners in California to understand their attitudes

towards EV-related incentive policies, as well as their commute routes and the frequency of their access

Appendix A

159

to high occupancy vehicle/toll (HOV/T) lanes. In San Francisco Bay Area for example, 32% reported that

they are paying reduced tolls of about $540 dollars per year and 28% save on commute time while

driving alone. The value of a program was estimated based on the toll savings reported and the value of

travel time savings for each survey respondent while commuting, and the estimated value was

compared with the corresponding Clean Vehicle Rebate for which the respondent was eligible. The

authors also examined the spatial heterogeneity of CAV decal value across different regions and tested

the impact of local HOV/T lane accessibility to the value of CAV decals.

14. Who are the early adopters of fuel cell vehicles?

Authors: Scott Hardman, Gil Tal

Publication date: 2018/9/13

Journal: International Journal of Hydrogen Energy

Volume: 43

Issue: 37

Pages: 17857-17866

Publisher: Pergamon

Description:

All modern technologies, including automotive technologies, are first purchased by early adopters.

These consumers are currently posed with the choice of purchasing a fuel cell vehicle (FCV) or a variety

of other alternatively fueled vehicles, including battery electric vehicles (BEVs). For FCVs to be

commercially successful they need to carve out their own niche in the automotive market, something

which may prove challenging in the face of strong BEV market growth. The results in this paper come

from a questionnaire survey of 470 FCV owners and 1550 BEV owners. The paper explores the socio-

economic profile, travel patterns, and attitudes of FCV buyers and compares them to the buyers of BEVs.

The result suggests that the adopters of BEVs and FCV are similar in gender, level of education,

household income, and have similar travel patterns. They have differences in age, ownership of previous

alternative fuel …

15. Estimating the Longest Trip for Plug-In Electric Vehicle Households

Authors: Rosaria M Berliner, Gil Tal, Alan Jenn

Publication date: 2018

Source: Transportation Research Board 97th Annual Meeting Transportation Research Board

Issue: 18-04792

Description:

Long distance road trips are underreported and underestimated in many travel behavior studies. These

infrequent trips of several hundred miles account for a non-trivial percentage of vehicle and household

vehicle miles traveled (VMT), yet many studies tend to overlook, underreport, or misrepresent them.

Overall, for households that own a new plug-in vehicle, a single trip (the longest in the last 12 month)

accounts for 10% of the household’s annual VMT for almost 95% of households. In terms of greenhouse

Appendix A

160

gas (GHG) emissions, 10% of the household’s GHG emissions are accounted for by that trip for

approximately 90% of households in the sample. The authors explore the variables and characteristics

that effect the distance of the longest trip. They use data collected in California during June and July

2017 as part of a study that focused on plug-in electric vehicle (PEV) households. The authors estimate a

log-linear model to understand the factors that influence the length of the longest road trip made in the

previous 12 months. The number of household vehicles, the presence of low-range battery electric

vehicles, and the number of passengers on the longest trip have the greatest impact on trip-length.