Vehicle Transport System by Sharing, Electrification and

Automatization

Toshiyuki YamamotoNagoya University

1

CASE

• Connected

• Autonomous

• Shared

• Electric

2

C

A

S

E

“Connected, Autonomous, Shared, Electric: Each of these has the power to turn our entire industry upside down. But the true revolution is in combining them in a comprehensive, seamless package.” by Dr. Dieter Zetsche (Chairman of the Board of Management of Daimler AG)

Shared autonomous vehicles

3

A

S

Automation levels (SAE)

4

Source: https://www.nhtsa.gov/technology-innovation/automated-vehicles-safety

Level3

2020

Level 4

2025Target year in Japan

Car sharing• The fleet is made available for use by members of

the car sharing organization• Merits: Rational mode choice, decrease car

dependency, fuel efficient vehicle, save parking space, etc.

5Car sharing in Japan

(Source: Foundation for Promoting Personal Mobility and Ecological Transportation)

0200,000400,000600,000800,0001,000,0001,200,000

05,000

10,00015,00020,00025,00030,000

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

Mem

bers

hip

(per

son)

Flee

t size

(veh

icle

)

Year

Fleet size (vehicle)Membership (person)

A

S

Weak point of car sharing

• One-way system is more convenient for users than return-only system

• But, one-way operation causes imbalance of fleet, deteriorating efficiency

6

Autonomous vehicle can relocate by themselves

Objective

Forecast supply and demand of shared autonomous vehicles

• Probability for sharing private cars• Potential demand for driverless taxi• Required fleet size

7

Study area

Meito Ward, Nagoya, Japan• Area: 19.45 km2

• Population: 164,570• East-end of Nagoya City• Residential area• Good access to CBD by

subway

8

Sharing of private autonomous cars

9

Uber:Private car driven by

human driver as taxi

Shared private autonomous

car:private

autonomous car used as taxi at spare time

Now Future

Framework

Car ownershipand shared use

Mode choice Traffic network

10

Car availability

Traffic demand

CongestionWaiting timeFleet supply

• Interaction is roughly considered• Equilibrium state is not rigorously calculated

Revenue

Intention for autonomous vehicle ownership & shared use (N=803)

• 70,000 Private cars in the area -> 9100 potential shared cars-> Driverless taxis system can be organized by them 11

Own car w/o share55%

Own car & share it

13%

Don't own & use shared

car32%

Age difference in intention

12

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

20s

30s

40s

50s

60+

Own car w/o share Own car & share it Don't own & use shared car

Current car use (or non-use)

• 14 & 13 hrs. at garage on weekday and weekend on average• 10 & 14% of households don’t use car on weekday and

weekend 13

0 50 100 150 200 250

Almost everyday4-5 days/week2-3 days/week

1 day/week1 day/2-3 weeks

1 day/month1 day/2-3 months

1 day/half year1 day/year

Less than aboveNo drive

No license

Drive frequency

0 50 100 150

Not fixed

3

6

9

12

15

18

21

24

Car at garage (hours/day)

Weekend Weekday

Z Z z…

Expected monthly income by sharing

14

3,000JPY13% 5,000JPY

11%

10,000JPY17%

15,000JPY11%

20,000JPY36%

30,000+JPY9%

Other3%

• Assumed to provide your private car for 5 hrs. each day

Nested logit model of intra-zonal travel mode choice

15By Chukyo person trip survey data in 2011

Estimation results (N=4542)Generic variable Coef.

Travel cost [100JPY] -0.126**

Travel time [hour] -0.998**

Waiting time [hour] -2.211**

16

Alternative specific variable Rail Bus Taxi Car 2wheel

Male 0.106 -0.328 0.253 0.071* -0.059*Child (<16) -2.241** -0.348* -0.785**Student -0.350* 0.335**Old (65+) 0.197 2.419** 0.805**Unemployed -0.605* 0.224 -0.163 -0.129** -0.161**Commute 1.007** 1.711** -0.141**Constant -2.199 -3.907 -0.71 0.273 -0.099

• Adjusted rho-square = 0.158• Value of time = 792 JPY/hr• Inclusive value = 0.245** (for NMT)

=1.0 (fixed for MT)

• Walk as base alternative * 5% significance, ** 1% significance

Potential demand scenarios• Cost is assumed as 55 JPY/km (slightly less than private car)• Waiting time is assumed as 1 minute• Those who own car w/o share will not use other share cars

17

Rail1.2%

Bus2.1% Taxi

0.2%

Car22.3%

Bike14.6%

Walk39.7%

SAV19.9%

Mode share

Those who own car w/o share

Waiting time won’t use will use

1 minute 22455 trips 43307 trips5 minutes 18107 trips 34005 trips

Trip demand by scenario

Agent-based simulation

18

Empty taxi En route to customer

Occupied taxi Waiting customer

• Trip demand:– Generated based on

actual OD pattern

• Vehicle agent:– Distributed based on

population distribution

• Vehicle speed:– 18.9 km/h (peak hour)– 24 km/h (off-peak)

System behavior by scenario

19

0

1

2

3

4

5

100% 75% 50%

Prob

abili

ty (%

)

Proportion of SAV supply

Probability of waiting time over 1 minutes

Waiting time: 1 min, limited user Waiting time: 5min, limited usersWaiting time: 1min, all users Waiting time: 5 min, all users

1 minute of waiting time is satisfied at 95+%

Relationship between supply and income

20

05000

100001500020000250003000035000400004500050000

30% 40% 50% 60% 70% 80% 90% 100%

MO

NTH

LY IN

CO

ME

OF

SAV

S (J

PY/C

AR

)

PROPORTION OF SAV SUPPLY

Cumulative distribution of expected income

Electrification of university car fleet: A case of Nagoya University

21

S

E

Electric vehiclesEVs (Electric vehicles)

BEVs(Battery Electric

Vehicles)

PHEVs(Plug-in Hybrid

Electric Vehicles)

HEVs(Hybrid Electric

Vehicles)

FCEVs(Fuel-Cell

Electric Vehicles)

Charging Level Type* Power

supplyCharging time

(24kWh battery) Typical use Standard

Level 1 Normal 120V AC 16 hours Home SAE J1772

Level 2 Normal 240V AC 8 hours Home or public places SAE J1772

Level 3 Fast 480V DC 30 minutes Public places CHAdeMO/CCS/Tesla

Battery charging

*Normal charging aka slow chargingFast charging aka quick or rapid charging

S

E

Anxiety about electric vehicle

• Shorter drive distance than gasoline vehicle

• Increase in electricity demand

23

Mostly short distance trips are served by car sharing

V2G can contribute peak cut of demand

Vehicle to grid (V2G)

• Batteries in EV could be used to let electricity flow from the car to the electric grid

• Provide power to help balance loads by “valley filling” and “peak-shaving”

24Source： http://www.cenex.co.uk/vehicle-to-grid/

Objective• To evaluate the reduction of CO2 emission by

replacing university car fleet with EV• To quantify electricity supply with V2G for

campus use

25

Method• Fitting the Daily Travel Distance (DTD) data with

different distribution functions• Based on the distribution function, determining the

vehicles that can be replaced by EV• Calculating electricity supply with V2G considering

usage and charging pattern

DataNagoya University’s fleet system• Observations: Oct. 2014 to Sept. 2015 with 54 vehicles• Item: department, vehicle ID, vehicle type, time of

check-out and check-in, etc.

26

Graduate School of Env. Studies 12School of Agricultural Science 10Research Institutes 10Secretariat 8Faculty of Science 5Faculty on Liberal Arts 3School of Engineering 2School of Informatics and Science 2Museum 1Physical Education Center 1

Number of vehicles by Faculty

47

25

Vehicle Type

ordinary gasolinevehicle

diesel vehicle

hybrid vehicle

Vehicle use rate by time of day

27

0%

1%

2%

3%

4%

5%

6%

7%

8%

9%

10%0

1:00

2:00

3:00

4:00

5:00

6:00

7:00

8:00

9:00

10:0

0

11:0

0

12:0

0

13:0

0

14:0

0

15:0

0

16:0

0

17:0

0

18:0

0

19:0

0

20:0

0

21:0

0

22:0

0

23:0

0

The

use

rate

Time of day

SUN MONTUE WEDTHU FRISAT

Two peaks (10:00 and 13:00) and significant drop at 12:00Consistent with people’s daily work schedule

Trip distance per check-out

2016/03/29 28

0%

10%

20%

30%

40%

50%

60%

10 20 30 40 50 60 70 80 90 100110120130140150160170180190200km

0%2%4%6%8%

10%12%14%16%

1 2 3 4 5 6 7 8 9 10km

• Mostly, vehicles are used for short distance trips

The number of vehicles used at the same time

• At maximum, 21 vehicles (out of 48) are used at the exact same hour, and it happened only once in a year. 29

0

500

1000

1500

2000

2500

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Num

ber o

f hou

rs p

er y

ear

Number of vehicles in use

The number of vehicles in use at the exact same hour

Significant number of vehicles can be reduced by implementing car-sharing system

Distribution of daily travel distance

Log-normal56.3%

Gamma18.8%

Weibull12.5%

Normal12.5%• Best fitted distribution

is chosen for each vehicle although 20 vehicles didn’t fit any at 95% confidence level

30

0

0.1

0.2

0.3

0.4

Prob

abilit

y

km

Vehicle 4637Observation Log-normal

Replace by EV

31

5 EV Scenarios• Type 1: Mitsubishi i-MiEV (100km, 16kWh)• Type 2: Chevrolet Spark EV (130km, 21kWh)• Type 3: Kia Soul EV (150km, 27kWh)• Type 4: Nissan Leaf (170km, 30kWh)• Type 5: Tesla Model 3 (320km, 60kWh)

22 23 25 2630

05

1015202530

Type 1 Type 2 Type 3 Type 4 Type 5

Num

ber o

f veh

icle

s

Type 1:i-MiEV

Type 2:Spark

Type 3:Soul

Type 4:Leaf

Type 5:Tesla

The number of vehicles that can be replaced by each type of EV

satisfying 95% of travel can be covered

CO2 reduction

14% 14% 15% 16%20%

0%

5%

10%

15%

20%

25%

Type 1:i-MiEV

Type 2:Spark

Type 3:Soul

Type 4:Leaf

Type 5:Tesla

Reduction of CO2 emission

32

Emission factorGasoline Electricity

2.3 kg/L 0.39 kg/kWh

�𝐶𝐶𝐶𝐶2 from 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 =𝐷𝐷𝐷𝐷𝐷𝐷

𝐷𝐷𝑒𝑒𝑒𝑒𝐷𝐷𝑒𝑒𝐷𝐷𝐷𝐷 𝑒𝑒𝑟𝑟𝐷𝐷𝐷𝐷𝑒𝑒× 𝐶𝐶 × 0.39

𝐶𝐶𝐶𝐶2 from 𝐷𝐷𝑟𝑟𝑔𝑔𝑔𝑔𝑒𝑒𝑒𝑒𝐷𝐷𝑒𝑒 = 𝐷𝐷𝐷𝐷𝐷𝐷/𝐽𝐽𝐶𝐶08 × 2.3

Vehicle to grid

33

• University provides real time data of electricity use

• Collected data from Jun 20th

2017 to Jul 9th 2017, and used the average as reference

• According to NU announcement, even 0.3% reduction at peak could be helpful for contract

02000400060008000

10000120001400016000kW

Time of Day

13561

Charge/discharge speed

34

� 𝑒𝑒𝑡𝑒𝑒 𝑔𝑔𝑠𝑠𝑒𝑒𝑒𝑒𝑠𝑠 𝑔𝑔𝑜𝑜 𝑒𝑒𝑡𝑟𝑟𝑒𝑒𝐷𝐷𝑒𝑒𝐷𝐷𝐷𝐷𝑒𝑒𝑡𝑒𝑒 𝑔𝑔𝑠𝑠𝑒𝑒𝑒𝑒𝑠𝑠 𝑔𝑔𝑜𝑜 𝑠𝑠𝑒𝑒𝑔𝑔𝑒𝑒𝑡𝑟𝑟𝑒𝑒𝐷𝐷𝑒𝑒𝐷𝐷𝐷𝐷 ⇒ 10𝑘𝑘𝑘𝑘 𝑠𝑠𝑒𝑒𝑒𝑒 𝑡𝑔𝑔𝑜𝑜𝑒𝑒S𝑠𝑠𝑒𝑒𝑒𝑒𝑠𝑠

Vehicle Battery Size

Level 2 LCS-30 5.8kW

Level 2 LCS-30 7.7kW

Level 2 LCS-30 9.6kW

Level 2 LCS-30 11.5kW

Mitsubishi i-MiEV 16kWh 5 5 5 5Chevrolet Spark EV 21kWh 7 7 7 7

Kia Soul EV 27kWh 4.5 4 4 4Nissan Leaf 30kWh 5 4.5 4.5 4.5

Tesla Model 3 60kWh 10.5 8 6.5 5

Estimated Electric Vehicle Charge Times

Source: https://www.clippercreek.com/charging-times-chart/

• Available electricity from EV is calculated according to charge/discharge speed considering vehicle usage pattern

Average available electricity by V2G

35

0

50

100

150

200

250

300

Avai

labl

e el

ectri

city

(kW

h)

Time of day

TYPE 1 TYPE 2 TYPE 3 TYPE 4 TYPE 5

Type 1:i-MiEV

Type 2:Spark

Type 3:Soul

Type 4:Leaf

Type 5:Tesla

1.21% 1.28% 1.40% 1.47% 1.72%

Peak hour

Final remark: Image of Car

36

• Status symbol• Independence• Pollution emitter

• Mobility tool• Fleet (cars)• Part of energy

management

Now Future