evaluation of supply quality in passenger transport as a basis for
TRANSCRIPT
Stuttgart, 27.09.2012
DB Netz AG – I.NVT 7
Univ.-Prof. Dr.-Ing. Andreas Oetting, Angela Rio
presenter: Michael Rittner
Evaluation of supply quality in passenger transport
as a basis for the assessment of railway infrastructure measures
2
Outline
Introduction & Requirements
Empirical Methods
Findings & Conclusions
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Outlook
3 DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Regarding future traffic growth the project „FreeFloat 1“ supports
the objectives of DB Netz AG as well as the requirements of politics
objectives of DB Netz AG requirements of politics
manage the future traffic growth in an
economically efficient and sustainable way
strengthen competitiveness of the railway
ensure an efficient and modern railway
infrastructure
strengthen the economic position of Germany
enable traffic growth
use investment ressources optimally
defuse bottlenecks in the network
make system reserves useable
enhance punctuality
FreeFloat 1 funded by:
State of the art:
The evaluation of quality measures is limited
4
travel costs
scheduled travel time
transport demand models generally regard
increase of capacity
reduction of in-vehicle-times
improvement of timetable quality
improvement of operational quality
benefits of an infrastructure measure
scheduled
travel time
travel time
travel time
travel costs
travel costs
travel costs
example
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
5
benefits
economic evaluation of quality improving
measures (in particular delay reducing measures)
comparison of capacity and quality measures by a
common criterion
identification of the most suitable set of measures
to increase the attractiveness of the railway
system
consistent evaluation of measures along the value
chain of the infrastructure manager
ne
ga
tive
va
luatio
n
delays
scheduled
in-vehicle-times
revenues
costs
profit
contribution
€
The model to be developed by Evalinfra provides the basis for an
economic evaluation of the quality impacts of infrastructure measures
„Evalinfra“ project contents
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
deduction of a quantitative relationship between the
valuations of different components of supply quality
integration into transport demand models
example: delays
Transport demand models have to consider the basic characteristics
of supply quality that can be influenced by the infrastructure manager
6
impact of infrastructure measures on supply quality
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
reduction of scheduled waiting times
impact on supply quality
reduction of in-vehicle-times
no impact on supply quality, economic
evaluation by number of saleable trainpaths
reduction of delays leading to
- fewer waiting times at the originating station
- fewer in-verhicle-times
- shortened or extended changing times
increase of capacity
reduction of in-vehicle-times
improvement of timetable quality
improvement of operational quality
7 DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Evalinfra is devided into 4 work packages (WP)
WP 1
Survey freight transport
WP 4
Implementation and application on a first example
WP 3
Demand models for the valuation of supply quality
WP 2
Survey passenger transport
network planning (macro) network planning (micro) scheduling disposition
8
Outline
Introduction & Requirements
Empirical Methods
Findings & Conclusions
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Outlook
9
example: conjoint measurement matrix
objective:
deduction of a quantitative relationship between the
valuations of different components of supply quality with
focus on delays
method: nationwide survey in long-distance and local
passenger trains
main survey period: October / November 2009
utilizable return:
approx. 3,200 of 5,000 questionnaires (64%)
survey content:
- core of the questionnaire: conjoint measurement matrix
(5 different, jointly analyzable matrices)
- supplementary information on the customers’ perceptions
and opinions
- information on „today‘s“ journey
- basic information on the person
different questionnaires for:
- long-distance and local passenger trains
- three travel distances (50 km, 130 km, 400 km)
A survey was conducted to gain more specific information on
how passengers value different components of travel time
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
delay at
originating station
waiting time caused by
missed connecting train
changing time
travel costs
delay at
orig. station
waiting time caused by
missed connecting train
changing
time
travel
costs
method of the survey in passenger transport
10
example: conjoint measurement matrix
Conjoint analysis was used to deduce a quantitative relationship
between the valuations of different characteristics of supply quality
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
method of analysis
e.g. additional time
(ne
ga
tive)
sta
nd
ard
ized
pa
rt w
ort
h u
tilit
y
regression line
utility attribute 1
regression line
utility attribute 2
marginal utility attribute 2
marginal utility attribute 1 conjo
int
ana
lysis
*
delay at
originating station
waiting time caused by
missed connecting train
changing time
travel costs
delay at
orig. station
waiting time caused by
missed connecting train
changing
time
travel
costs
* in the project conducted by Prof. H.E. Jochim, FH Aachen University of Applied Sciences,
Fachbereich Bauingenieurwesen, Lehrgebiet Bahnanlagen und öffentlicher Verkehr;
literature: Backhaus, K., Erichson, B., Plinke, W., Weiber, R.: Multivariate Analysemethoden;
10. Aufl., Berlin/Heidelberg 2003
11
Outline
Introduction & Requirements
Empirical Methods
Findings & Conclusions
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Outlook
12 DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Passengers generally seem to value stops more negatively
than in-motion times
Part worth utilities for in-vehicle-times and the
number of stops both are represented very well by
linear functions.
An increasing marginal utility cannot be confirmed
by the given results for both attributes.
Passengers don‘t seem to distinguish between
„expected“ and additional in-vehicle-times.
Passengers generally seem to value stops 1.5 to
2.5 times more negatively than in-motion times.
valuation of stop times
va
luati
on
of
sto
p t
ime
s
in r
ela
tio
n t
o in
-ve
hic
le-t
ime
s
[min
]
travel distance
A valuation of about 2.5 to 3 for most travel purposes
could be confirmed for delays on the train in relation to in-vehicle-times
13 DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
valuation of delays as a prolongation of in-vehicle-time
The valuation of delays in relation to in-vehicle-times of JOCHIM [1] could be confirmed for delays
that passengers undergo on the train:
valuation by long-distance passengers on business or private trips between 2.5 and about 3
valuation by vacation travellers a little higher, yet at the same time scheduled in-vehicle-times are
attached less value to by this group
A valuation for long-distance commuters could be deduced, comparable very well to the valuation by
business and private travellers.
A weighting factor of 2.5 could be deduced for travel distances of less than 100 km, which can be
used for mass passenger transport.
[1] Jochim, H. E.: Verkehrswirtschaftliche Ermittlung von Qualitätsmaßstäben im Eisenbahnbetrieb; Veröffentlichungen des
Verkehrswissenschaftlichen Institutes der RWTH Aachen, Heft 54 (1999).
It can be assumed that all waiting times at a train station
caused by delays are valued by passengers in the same way
14 DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
valuation of delays at a station
The means for the negative standardized part worth utilities indicate a common function with a constant to diminishing
marginal utility for delays at the originating station and waiting times caused by a missed connecting train.
assumption: All waiting times at a train station caused by delays are valued by passengers in the same way (e.g. also
waiting times at a station because of a delayed connecting train).
A simplified approach was used to roughly estimate a constant factor for the time being.
Quelle: FH Aachen
A comprehensive approach could be deduced for
the valuation of transfers
15 DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
valuation of transfers
An average transfer with unknown platform
(including up to 10 min walking) is valued by
passengers like about 15 to 30 minutes additional
in-vehicle-time.
An average transfer includes an average need of
changing platforms. (assumption: With a probability
of 90% a transfer requires a change of platforms.)
Long-distance passengers attach a negative value
of about 15 to 20 minutes of in-vehicle-time to the
need of changing platforms for a transfer.
The value of a transfer can be reduced to about 10
to 15 minutes of in-vehicle-time, when it is ensured
that a change of platforms is not required (same
platform).
same
platform
unknown
platform
0
5
10
15
20
25
30
35
> 200 km100 - 200 km< 100 km
va
luati
on
of
tra
nsfe
rs
in r
ela
tio
n t
o in
-ve
hic
le-t
ime
[min
]
travel distance
90% • (valuation of change of platforms)
16
Outline
Introduction & Requirements
Empirical Methods
Findings & Conclusions
DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Outlook
17 DB Netz AG, Michael Rittner, I.NVT 7, 27.09.2012
Outlook
next steps
Within the project Evalinfra DB Netz AG has integrated the deduced factors in a travel
demand model.
The model has been implemented in the network planning system “WiZug/MOSES”.
An implementation of the model in further systems is pursued in order to achieve a consistent
assessment of qualitative infrastructure measures.
The deduced weighting factors and coherences can be integrated in further travel demand
models; yet, attention should be paid to the fact that if parameters are newly integrated in a
system of parameters, the whole system should be jointly adjusted according to its strengths
and weaknesses.
further research
Further research should concentrate on a further detailing of the weighting factors by the
purpose of the journey for the single attributes.
Particularly for the valuation of delays causing additional waiting times at a station a non-linear
approach should be pursued.