Analysis of automated driving diffusion: Potential
diffusion paths into the German Market
Dr. Christine Eisenmann
SIP-adus Meeting, 12/11/2020
In collaboration with:
Dr. Christian Winkler, Dennis Seibert, Nina Thomsen (DLR)
Prof. Tobias Kuhnimhof, Michael Schrömbges (RWTH)
Dr. Thomas Meissner, Dr. Peter Phleps (BMW)
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020DLR.de • Chart 2
Research objectives and contents
Diffusion of connected and automated
driving (CAD) (focus: car market)
Changes in car ownership, individual
travel behaviour and collective travel
demand
Reserch
objectives
Potential
diffusion
paths
The private autonomous car Autonomous Mobility as a Service MaaS
(resp. ODM, ridehailing)
Potential
diffusion
paths
DLR.de • Chart 3
Research objectives and contents
Research
questions
in this
regard
• How do those diffusion paths affect car ownership, the new car market, the car stock and
travel demand?
• Which effects do the different diffusion paths have on transport sustainability and on the car
industry?
• Is the narrative (told in Germany) realistic that private cars will be replaced by automated
services and that automation will lead to a reduction of the car fleet? And if so, under which
conditions?
•
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 4
To analyse CAD diffusion from a quantitiative perspective, a model chain is
being implemented
Eff
ec
tso
n
Market entry of CAD (connected and automated driving)
Vehicle ownership
in households
New car market
and car stockTravel demand
and CAD services
Resulting impacts of CAD diffusion
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 5
Key questions on market entryE
ffe
cts
on
Market entry of CAD (connected and automated driving)
Vehicle ownership
in households
New car market
and car stockTravel demand
and CAD services
Resulting impacts of CAD diffusion
When will automated cars be available on the German market?
When will we have automated MaaS vehicles on the streets?
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 6
Model assumptions on the entry of level 4 automated vehicles into the
German market are derived from interviews with industry experts
2030 2035 2040 2045
Level 4 luxury class vehicles
Level 4 medium class vehicles
Mobility as a Service
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 7
Key questions on vehicle ownership in householdsE
ffe
cts
on
Market entry of CAD (connected and automated driving)
Vehicle ownership
in households
New car market
and car stockTravel demand
and CAD services
Resulting impacts of CAD diffusion
How does automation affect the car ownership decisions of private households on
(i) the private cars diffusion path or (ii) the MaaS diffusion path?
Under which conditions might the availability of MaaS lead to household decisions
against an own car?
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 8
A discrete choice model for car ownership in households is being developed
Two successive
logistic regression
models
Diffusion paths
- private car
diffusion path
- MaaS diffusion
path
GoalModel car
ownership in private
households
Household
attributes
Socio-economics
Built environment
Transport supply
Personal attributes
Future
?
Model estimation for future scenarios
Model calibration with the national travel survey „Mobility in Germany“
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 9
Key questions on the new car market and the vehicle stockE
ffe
cts
on
Market entry of CAD (connected and automated driving)
Vehicle ownership
in households
New car market
and car stockTravel demand
and CAD services
Resulting impacts of CAD diffusion
How many automated cars will we have in the German new car market and car stock
in 2030, 2040 and 2050 at (i) the private car diffusion path or (ii) the MaaS diffusion
path?
How will automation affect the used car market?
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 10
An agent-based vehicle stock model is being used to study the effects of
automation to the new car market and the vehicle stock
The German vehicle stock is modeled dynamicly
1.1
.1995
1.1
.1996
1.1
.1997
1.1
.1998
1.1
.1999
1.1
.2000
1.1
.2001
1.1
.2002
1.1
.2003
1.1
.2004
1.1
.2005
1.1
.2006
1.1
.2007
1.1
.2008
1.1
.2009
1.1
.2010
1.1
.2011
1.1
.2012
1.1
.2013
1.1
.2014
1.1
.2015
1.1
.2016
removals (export & scrapping)
new vehicle registrations
ownership transfers
Annual, agent-based simulation of household
choices
Purchase decision Used vehicle market decision
BadAverageGood
Vehicle technologydecision
Vehicle segment
Drivetrain
Automation-level
No purchase
New vehicle (commercial)
New vehicle (private)
Used vehicle
Yearlyhousehold
choices
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 11
Ac rising demand for automated cars on the new car market has a ladded
effect on the car stock
Exemplary, preliminary results for the private car diffusion path
0
1
2
3
4
5
2020 2025 2030 2035 2040 2045 2050
Ca
rs [
mio
.]
New car sales
Level 0-3 Level 4+ optimistic Level 4+ pessimistic
0
10
20
30
40
50
2020 2025 2030 2035 2040 2045 2050
Ca
rs [
mio
.]
Car stock
Level 0-3 Level 4+ optimistic Level 4+ pessimistic
? ?
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 12
Key questions on travel demand and CAD servicesE
ffe
cts
on
Market entry of CAD (connected and automated driving)
Vehicle ownership
in households
New car market
and car stockTravel demand
and CAD services
Resulting impacts of CAD diffusionHow will future travel mode choice change on (i) the private car diffusion path or (ii)
the MaaS diffusion path?
How does the operation of MaaS fleets work?
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 13
A german-wide transport demand model is being extended, to analyze the
impacts of automation on travel demand
Macroscopic travel demand model for Germany (DEMO)
Diffusion of CAD within
the private vehicle fleet
Autonomous MaaS
vehicles
Trip generation
Trip distribution
Mode choice
Route assignment
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 14
In the travel demand model MaaS is being applied in large cities
MaaS will be available in cities
with more than 100,000
inhabitants (marked in orange)
The city of Weimar is used
as a reference for the
algorithm development• Only 3 to 6 MaaS vehicles per 1,000 inhabitants
are needed to enable a MaaS modal share of 7%
to 11%
• MaaS vehicles cover about eight times more
distance per day than private cars
• The occupancy rate indicates that only a few
MaaS trips are shared
Modal
Share
MaaS fleet Daily mileage Occupancy rate
[vehicles per 1,000
inhabitants]
[km per vehicle] [passengers per
trip]
7 % 3.8 269 1.1
11 % 5.2 261 1.3
Exemplary results on MaaS for the Weimar region
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 15
Key questions on resulting impacts of CAD (at a later stage of the project)E
ffe
cts
on
Market entry of CAD (connected and automated driving)
Vehicle ownership
in households
New car market
and car stockTravel demand
and CAD services
Resulting impacts of CAD diffusion
What are the resulting effects of CAD diffusion (on transport, the environment,
industry) at (i) the private car diffusion path or (ii) MaaS diffusion path?
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
DLR.de • Chart 16
Conclusion and outlook
• With this project we are applying a unique and solid model chain to adequately display two likely diffusion
paths of automated vehicles into the German market and their effects on sustainability and the industry
• The applied model chain enables us to shed light into the discussion (in Germany), whether and under which
conditions automation might lead to a reduced car fleet.
• The Japanese-German collaboration:
• The joint reflection of assumptions, scenarios and model approaches is very beneficial for the project
activities.
• Comparisons on CAD diffusion in Japan and Germany given structural similarities but also
geographical, social and regulatory differences are helpful for the development and implementation of
CAD
• A major question remains as to what impact the current development and the corona pandemic will have on
the probability of either of the diffusion paths occurring
> SIP-adus > Christine Eisenmann • CAD Diffusion Germany > November 12, 2020
Thank you for your attention!
Dr. Christine Eisenmann
DLR Institute of Transport Research
DLR.de • Chart 17