Content uploaded by Elias Pajares
Author content
All content in this area was uploaded by Elias Pajares on Jan 30, 2023
Content may be subject to copyright.
Technische Universität München
TUM School of Engineering and Design
Development, application, and testing of an accessibility
instrument for planning active mobility
Elias Pajares
Vollständiger Abdruck der von der TUM School of Engineering und Design der Technischen
Universität München zur Erlangung eines
Doktors der Ingenieurwissenschaften (Dr. -Ing.)
genehmigten Dissertation.
Vorsitz: Prof. Dr.-Ing. Rolf Moeckel
Prüfer*innen der Dissertation:
1. Prof. Dr.-Ing. Gebhard Wulfhorst
2. Prof. Dr.-Ing. Liqiu Meng
3. Assoc. Prof. Dr. Anders Larsson
Die Dissertation wurde am 29.09.2022 bei der Technischen Universität München eingereicht
und durch die TUM School of Engineering und Design am 06.12.2022 angenommen.
Elias Pajares I
Acknowledgments
This dissertation emerged from my time as a researcher at the Chair of Urban Struc-
ture and Transport Planning. I feel infinitely grateful for having the opportunity to work
in this marvelous environment. Their excellent work sparked my interest in urban and
transport planning, and they introduced me to the wonderful world of accessibility. The
openness and support to kick off the development of GOAT were exceptional. In par-
ticular, I would like to thank my Doktorvater, Prof. Dr. -Ing. Gebhard for believing in
me and providing critical and inspiring input, which continuously helped me to move
forward. Furthermore, I would like to thank my dear second supervisor, Prof. Dr. -Ing.
Liqiu Meng, for her honest and exceptionally innovative feedback.
I also want to express my special thanks and admiration to Dr. -Ing. Benjamin
Büttner. His research group was not only an unbeatable academic home but a joyful
spring of new friendships. Therefore, my thanks also extend to the whole team. We
spent prime time during work, in pubs, and at parties. Particularly endless thankfulness
goes to Ulrike Jehle and Majk Shkurti, who believed in the idea of GOAT from day
one. Without Ulrike’s brilliant and inclusive mind, the numerous workshops and many
improvements in the software would not have been possible. I have never met an
equally talented and motivated programmer as Majk. We spent nights coding and with
rich discussions. Your support for developing the frontend of GOAT and friendship is
invaluable.
Furthermore, I would like to thank all the students who helped with their theses in
bringing the development of GOAT forward. Your support opened new horizons and
helped the project grow. It was an honor to supervise you all. I am very grateful for the
precious support from the cities of Fürstenfeldbruck, Freising, Freiburg, and Munich to
test GOAT and for providing their rich experience from practice in co-creative develop-
ment. My family and friends were an endless source of support and motivation to bring
my research forward. I am so happy you provided me with moments of distraction and
love during these intense times. Finally, I would like to thank a thousand times the most
important person in my life – my wife. Without your endless patience and support, this
dissertation would have been impossible.
Elias Pajares III
Abstract
Our transport system is both an enabler and a threat to sustainable development in
cities. While providing opportunities and prosperity, urban transportation is accom-
panied by severe consequences such as land seizure, global warming, and pollution.
There is increasing awareness that travel demand is a derived phenomenon, and land-
use patterns strongly influence our daily mobility choice. The concept of accessibility
is promising by providing a holistic framework to interrelate transport with land-use.
More than any other form of transport, active mobility (e.g., walking and cycling) de-
pends on high local accessibility to destinations. Therefore, emerging ideas, such as
the 15-Minute City propagate high proximity to destinations and the provision of ap-
propriate infrastructure. Accessibility instruments function as a specialized Planning
Support System (PSS) and can create high value for planning practice when shap-
ing sustainable cities. However, research has shown that accessibility instruments are
rarely adopted in practice. In addition to institutional barriers, it was found that acces-
sibility instruments still do not meet expectations in practice, and the availability of data
and resources remains an open concern.
Therefore, this dissertation presents a novel accessibility instrument called Geo Open
Accessibility Tool (GOAT) that focuses on active mobility and local accessibility. It is
a web-based tool combining different web and Geographic Information Systems (GIS)
technologies and was developed in a co-creative and iterative process. The instrument
is developed open source and was applied in numerous German and international case
studies.
The tool requirements were derived by exploring the existing tool landscape through
testing and literature review of 26 accessibility instruments. Accessibility instruments
have undergone significant developments in the last few years. Improvements in GIS,
new web technologies, and the rising availability of (open) data have facilitated the fast
development of these new tools. Nevertheless, the review still identified a lack of open
source and transferable instruments that allow on-the-fly scenario building while focus-
ing on active mobility. Furthermore, a clear gap was identified between fully-featured
proprietary desktop and (open) web tools. Therefore, by focusing on an open, interac-
Elias Pajares V
tive, and transferable web tool, the typical shortcomings of existing accessibility instru-
ments were addressed during the development of GOAT.
One focus was developing data strategies to enable powerful and affordable acces-
sibility instruments. A novel population disaggregation procedure was developed that
produces population data on the building access level with low data requirements. Fur-
thermore, Point of Interest (POI) data originating from OpenStreetMap (OSM) was re-
fined and fused with additional data sets. In addition, data was contributed to OSM
through different means, such as mapping events with volunteers and a mapping mode
in GOAT. The different data strategies made it possible to apply the tool in practice.
Finally, the usefulness of GOAT in practice was studied. In particular, the tool was used
for real-world planning questions and assessed by 42 planning professionals in work-
shops. Four core use cases were identified and users appreciated the tool’s interactivity
and easy-to-use interface. However, further improvements to the tool are necessary,
including enhancing the implemented indicators, increasing its usability and integrating
it with existing GIS software.
The developed tool offers a valuable contribution to the landscape of accessibility instru-
ments. It is recommended that future research focus on the open technical challenges
that GOAT and other accessibility instruments still face. Furthermore, tool developers
should strive for continuous exchange with the planning practice to meet real-world re-
quirements, support teaching accessibility, and aim to use accessibility for sustainable
development.
Zusammenfassung
Unser Verkehrssystem ist sowohl ein Motor als auch eine Gefahr für die nachhaltige
Entwicklung in Städten. Während es Entwicklungschancen und Wohlstand ermöglicht,
geht städtischer Verkehr mit schwerwiegenden Folgen wie Flächenverbrauch, Erder-
wärmung und Umweltverschmutzung einher. Es gibt ein wachsendes Bewusstsein
dafür, dass die Verkehrsnachfrage eine abgeleitete Größe ist, und die Raumstruktur
unsere täglichen Mobilitätsentscheidungen stark beeinflusst. Das Konzept der Erreich-
barkeit ist vielversprechend, da es einen ganzheitlichen Blick auf die Verknüpfung von
Verkehr und Raumstruktur bietet.
Die aktive Mobilität (z. B. Fußverkehr und Radverkehr) ist mehr als jede andere
Art von Fortbewegung auf eine gute lokale Erreichbarkeit zu Zielen angewiesen. Da-
her propagieren neue Ideen, wie die 15-Minuten-Stadt kurze Wege zu Zielen und die
Bereitstellung von geeigneter Infrastruktur. Erreichbarkeitsinstrumente fungieren als
spezialisierte Planning Support System (PSS) und können einen großen Mehrwert für
die Planungspraxis in der Gestaltung nachhaltiger Städte schaffen. Die Forschung hat
jedoch gezeigt, dass Erreichbarkeitsinstrumente selten in der Praxis eingesetzt wer-
den. Neben institutionellen Barrieren wurde auch festgestellt, dass Erreichbarkeitsin-
strumente in der Praxis noch immer nicht den Erwartungen entsprechen, und die Ver-
fügbarkeit von Daten und Ressourcen bleibt ein offenes Problem.
Daher wird in dieser Dissertation ein neuartiges Erreichbarkeitsinstrument namens
Geo Open Accessibility Tool (GOAT) entwickelt, das sich auf die aktive Mobilität und
lokale Erreichbarkeit konzentriert. Es ist ein webbasiertes Tool, das unter Nutzung
von verschiedene Web- und Geoinformationssysteme (GIS) in einem ko-kreativen und
iterativen Prozess entwickelt wurde. Das Instrument wird Open Source entwickelt und
wurde in zahlreichen deutschen und internationalen Fallstudien eingesetzt.
Die Anforderungen an das Instrument wurden abgeleitet, indem 26 bestehende In-
strumente in Tests und durch Literaturrecherche untersucht wurden. Erreichbarkeitsin-
strumente wurden in den letzten Jahren stark weiterentwickelt. Verbesserungen in GIS,
neue Webtechnologien und die zunehmende Verfügbarkeit von (offenen) Daten haben
die rasche Entwicklung neuer Instrumente begünstigt. Dennoch wurde in der Unter-
Elias Pajares VII
suchung ein Mangel an quelloffenen und übertragbaren Instrumenten, mit dynamischer
Szenarienentwicklung und einem Fokus auf der aktiven Mobilität festgestellt.
Darüber hinaus wurde eine deutliche Lücke zwischen voll funktionsfähigen propri-
etären Desktopsoftware und (offenen) Web-Tools identifiziert. Durch die Fokussierung
auf ein offenes, interaktives und übertragbares Webtool wurden daher während der En-
twicklung von GOAT die Herausforderungen bestehender Erreichbarkeitsinstrumente
adressiert. Ein Schwerpunkt war die Entwicklung von Datenstrategien, um leistungs-
fähige und erschwingliche Erreichbarkeitsinstrumente zu ermöglichen.
Es wurde eine neue Methode zur Disaggregierung von Bevölkerung entwickelt,
die mit geringem Datenaufwand Bevölkerungsdaten auf der Ebene des Gebäudeein-
gangs bereitstellt. Darüber hinaus wurden Point of Interest (POI) aus OpenStreetMap
(OSM) verfeinert und mit zusätzlichen Datensätzen fusioniert. Darüber hinaus wurde
die OSM-Datenbank auf verschiedene Weisen ergänzt, z. B. durch Mapping Partys mit
Freiwilligen und einem Mapping Mode in GOAT. Die verschiedenen Datenstrategien
ermöglichten die Anwendung des Tools in der Praxis.
Schließlich wurde der Mehrwert von GOAT in der Praxis untersucht. Insbeson-
dere wurde das Tool für reale Planungsfragen eingesetzt und von 42 Planungsex-
perten:innen in Workshops bewertet. Es wurden vier Hauptanwendungsfälle identi-
fiziert, Benutzer:innen schätzten die Interaktivität und die einfache Nutzeroberfläche
des Tools. Es sind jedoch weitere Verbesserungen des Tools erforderlich, darunter die
Verbesserung der implementierten Indikatoren, die Erhöhung der Benutzerfreundlichkeit
und die Integration in bestehende GIS-Software.
Das entwickelte Tool stellt einen wertvollen Beitrag zur bestehenden Landschaft von
Erreichbarkeitsinstrumenten dar. Es wird empfohlen, dass sich zukünftige Forschung
auf die offenen technischen Herausforderungen konzentriert, denen GOAT und andere
Erreichbarkeitsinstrumente noch gegenüberstehen. Darüber hinaus sollten sich die
Entwickler:innen von Instrumenten um einen kontinuierlichen Austausch mit der Pla-
nungspraxis bemühen, um den Anforderungen der Praxis gerecht zu werden. Zudem
sollten Erreichbarkeitsinstrumente das Erreichbarkeitskonzept lehren und für die nach-
haltige Entwicklung eingesetzt werden.
Contents
Contents
ListofFigures ................................... XII
ListofTables .................................... XV
ListofAbbreviations ................................ XVII
I Introduction, state of the art, and research design 1
1 Introduction 3
1.1 Background.................................. 3
1.2 Motivation................................... 4
1.3 Researchobjectives ............................. 5
1.4 Thesisstructure ............................... 7
2 Literature review 9
2.1 Accessibilitytheory.............................. 9
2.2 Accessibility measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Accessibility instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3 Research questions and goals 15
3.1 Maingoal ................................... 15
3.2 RQ1 Tool requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 RQ2Datamanagement ........................... 16
3.4 RQ3Toolassessment ............................ 17
4 Research design 19
4.1 Tooldevelopment............................... 20
4.2 Toolapplication................................ 22
4.3 Co-creative involvement . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.4 Link between research questions and scientific papers . . . . . . . . . . 25
Elias Pajares IX
II Scientific papers 27
5 Accessibility by proximity: Addressing the lack of interactive accessibility
instruments for active mobility 29
5.1 Introduction.................................. 30
5.1.1 The role of active mobility and accessibility . . . . . . . . . . . . 30
5.1.2 Objectives, research questions and structure . . . . . . . . . . . 31
5.2 Literaturereview ............................... 32
5.2.1 Quantitative planning support for active mobility . . . . . . . . . . 32
5.2.2 Literature review accessibility instruments . . . . . . . . . . . . . 33
5.3 Methodology ................................. 35
5.3.1 Overview accessibility instrument landscape . . . . . . . . . . . 36
5.3.2 Tooldevelopment........................... 37
5.3.3 Involvement planning practice . . . . . . . . . . . . . . . . . . . . 37
5.4 Current scene accessibility instruments . . . . . . . . . . . . . . . . . . 39
5.4.1 Overview current landscape of accessibility instruments . . . . . 39
5.4.2 Application gap active mobility . . . . . . . . . . . . . . . . . . . 43
5.5 Development of Geo Open Accessibility Tool (GOAT) . . . . . . . . . . . 44
5.5.1 Scope of the instrument . . . . . . . . . . . . . . . . . . . . . . . 44
5.5.2 Technical architecture . . . . . . . . . . . . . . . . . . . . . . . . 45
5.5.3 Routing algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.5.4 Contour-based accessibility measures . . . . . . . . . . . . . . . 49
5.5.5 Gravity-based accessibility measures . . . . . . . . . . . . . . . 52
5.5.6 Data.................................. 54
5.5.7 Input practitioners . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.6 Reflection tool development . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.7 Conclusion .................................. 58
6 Population Disaggregation on the Building Level Based on Outdated Cen-
sus Data 61
6.1 Introduction.................................. 62
6.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.2.1 StudyContext ............................ 65
6.2.2 Data.................................. 66
6.2.3 Software ............................... 68
6.3 Results .................................... 69
6.3.1 Tableschema............................. 70
6.3.2 Fusion of building data and dasymetric mapping . . . . . . . . . 72
6.3.3 Detection of building entrances and new developments areas . . 73
Contents
6.3.4 Updating of census tracts . . . . . . . . . . . . . . . . . . . . . . 75
6.3.5 Population distribution . . . . . . . . . . . . . . . . . . . . . . . . 76
6.3.6 Comparison with the municipal population registry . . . . . . . . 77
6.4 Discussion .................................. 81
6.5 Conclusions.................................. 83
7 Identification and discussion of use cases of an interactive accessibility
instrument for active mobility planning 87
7.1 Introduction.................................. 88
7.2 Literaturereview ............................... 89
7.2.1 Planning support systems in practice . . . . . . . . . . . . . . . . 89
7.2.2 Accessibility instruments and their potential . . . . . . . . . . . . 91
7.3 Accessibility instruments GOAT . . . . . . . . . . . . . . . . . . . . . . . 92
7.3.1 Overview GOAT project . . . . . . . . . . . . . . . . . . . . . . . 92
7.3.2 Technical architecture . . . . . . . . . . . . . . . . . . . . . . . . 93
7.3.3 Implemented indicators . . . . . . . . . . . . . . . . . . . . . . . 93
7.4 Methodology ................................. 95
7.4.1 Overview user involvement . . . . . . . . . . . . . . . . . . . . . 97
7.4.2 Application workshops . . . . . . . . . . . . . . . . . . . . . . . . 97
7.4.3 Usefulness assessment . . . . . . . . . . . . . . . . . . . . . . . 99
7.5 Results .................................... 102
7.5.1 Infrastructure planning walking . . . . . . . . . . . . . . . . . . . 102
7.5.2 Infrastructure planning cycling . . . . . . . . . . . . . . . . . . . 104
7.5.3 Location planning . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.5.4 Housing development . . . . . . . . . . . . . . . . . . . . . . . . 111
7.5.5 Overall assessment . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.6 Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 113
III Synthesis and discussion 119
8 Synthesis and discussion 121
8.1 RQ1 - Tool requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
8.1.1 Components of accessibility instruments . . . . . . . . . . . . . . 121
8.1.2 Observed trends in the current accessibility instrument landscape 122
8.1.3 Potential for the development of new accessibility instruments . . 123
8.1.4 Summary - tool requirements . . . . . . . . . . . . . . . . . . . . 124
8.2 RQ2 - Data management . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.2.1 Workflow data preparation . . . . . . . . . . . . . . . . . . . . . . 124
Elias Pajares XI
8.2.2 Data refinement and fusion . . . . . . . . . . . . . . . . . . . . . 126
8.2.3 VGI contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
8.2.4 Summary - data management . . . . . . . . . . . . . . . . . . . 130
8.3 RQ3 - Tool assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
8.3.1 Utility Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8.3.2 Usability assessment . . . . . . . . . . . . . . . . . . . . . . . . 132
8.3.3 Summary - tool assessment . . . . . . . . . . . . . . . . . . . . . 133
8.4 Maingoal ................................... 133
8.4.1 Tool characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . 134
8.4.2 Summary - main goal . . . . . . . . . . . . . . . . . . . . . . . . 135
9 Conclusions 137
9.1 Limitations .................................. 137
9.2 Future development path and research needs . . . . . . . . . . . . . . . 138
9.3 Finalreflections................................ 143
Bibliography 144
A Reviewed accessibility instruments 165
B Used software and programming languages 168
C Mapping table schema 169
D Table schema 170
E Core variables in data configuration file 175
List of Figures
List of Figures
1.1 Researchaim................................. 6
1.2 Thesisstructure ............................... 8
2.1 Four accessibility components . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Comparison of the most common impedance functions . . . . . . . . . 13
4.1 Connections between main goal, research questions, and methods . . . 19
4.2 Development timeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.3 Applied GOAT version worldwide . . . . . . . . . . . . . . . . . . . . . . 23
4.4 Different involvement formats . . . . . . . . . . . . . . . . . . . . . . . . 25
5.1 Researchwork-flow ............................. 36
5.2 Countries of origin reviewed instruments . . . . . . . . . . . . . . . . . . 39
5.3 Implemented transport modes . . . . . . . . . . . . . . . . . . . . . . . . 40
5.4 Tool type and scenario building capabilities . . . . . . . . . . . . . . . . 41
5.5 Access of accessibility instruments . . . . . . . . . . . . . . . . . . . . . 42
5.6 Data sources accessibility instruments . . . . . . . . . . . . . . . . . . . 43
5.7 Key strategic aims of GOAT . . . . . . . . . . . . . . . . . . . . . . . . . 44
5.8 Simplified technical architecture of GOAT . . . . . . . . . . . . . . . . . 45
5.9 Cycling impedance factor street gradient . . . . . . . . . . . . . . . . . . 48
5.10 Reached network and single isochrone . . . . . . . . . . . . . . . . . . 50
5.11 Multi-isochrone with served population by supermarkets . . . . . . . . . 51
5.12 Computation of accessibility values using Modified Gaussian impedance
functions ................................... 53
5.13 Hexagonal grid as spatial unit for the heatmap . . . . . . . . . . . . . . 53
5.14 Deriving travel times per point of interest . . . . . . . . . . . . . . . . . . 54
6.1 Studied municipalities. . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.2 Core spatial data used. . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.3 GOATArchitecture.............................. 68
Elias Pajares XIII
6.4 Overviewprocedure.............................. 70
6.5 Required input tables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6.6 Optional input tables. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.7 Data structure output tables. . . . . . . . . . . . . . . . . . . . . . . . . 71
6.8 Fusion building data and dasymetric mapping. . . . . . . . . . . . . . . 72
6.9 Binary classification buildings. . . . . . . . . . . . . . . . . . . . . . . . . 73
6.10 Detection of building entrances and New development areas. . . . . . . 74
6.11 New development areas and gross floor area per building entrance. . . 75
6.12 Update census tracts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.13 Population distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.14 Population distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.15 Comparison disaggregated and recorded population data on the build-
inglevel..................................... 79
6.16 Comparison disaggregated and recorded population data correlation ex-
amples. .................................... 79
6.17 Correlation disaggregated and recorded data on (a) building-level and
(b) grid-level (100 m ×100m). ....................... 80
6.18 Comparison disaggregated and recorded population on grid-level. . . . 80
6.19 Comparison census population from 2011 and recorded population 2020. 81
7.1 Technical architecture GOAT . . . . . . . . . . . . . . . . . . . . . . . . 94
7.2 CoreindicatorsGOAT ............................ 96
7.3 Main user groups involved in the development of GOAT . . . . . . . . . 98
7.4 Application workshop in Freising and Fürstenfeldbruck . . . . . . . . . . 98
7.5 Worksheet planning workshops . . . . . . . . . . . . . . . . . . . . . . . 100
7.6 Framework assessment usefulness . . . . . . . . . . . . . . . . . . . . . 100
7.7 Bridge scenario and changes in connectivity . . . . . . . . . . . . . . . . 105
7.8 Scenario new pedestrian bridge over a river . . . . . . . . . . . . . . . . 105
7.9 Scenario new barrier-free crossing . . . . . . . . . . . . . . . . . . . . . 107
7.10 Analyses and data visualization for planning cycling infrastructure . . . 107
7.11 Analyses and data visualization for planning cycling infrastructure . . . 109
7.12 Location planning social facilities - nurseries in Fürstenfeldbruck . . . . 109
7.13 Population density heatmap, Fürstenfeldbruck . . . . . . . . . . . . . . . 110
7.14 Comparison of accessibility and population density heatmap, Fürsten-
feldbruck ................................... 110
7.15 Scenario with buildings uploaded from a building development plan and
new road infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.16 New buildings and kindergartens . . . . . . . . . . . . . . . . . . . . . . 112
List of Figures
8.1 Components of accessibility instruments . . . . . . . . . . . . . . . . . . 122
8.2 Workflow data preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 125
8.3 Schema of common custom tables . . . . . . . . . . . . . . . . . . . . . 125
8.4 Mapping mode in GOAT . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
8.5 OSMmappingregions............................ 128
8.6 Feature types mapped . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
9.1 Potential pedestrian flows to primary schools for home-based trips of 6-
10-year-oldchildren ............................. 141
9.2 Accessibility instruments as a contribution to sustainable development . 143
Elias Pajares XV
List of Tables
List of Tables
4.1 Applied GOAT versions . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.2 Link between scientific papers and research questions . . . . . . . . . . 26
5.1 Participants of the application workshops, clustered by profession and
sector ..................................... 38
5.2 Default excluded OSM highway categories . . . . . . . . . . . . . . . . . 47
5.3 Cycling impedance factor street surface . . . . . . . . . . . . . . . . . . 49
5.4 Datasets used in the Munich Region . . . . . . . . . . . . . . . . . . . . 55
5.5 Collection of requested features and their current development status . 56
6.1 Data used for population disaggregation. . . . . . . . . . . . . . . . . . . 66
7.1 Datasetsused ................................ 94
7.2 Agenda planning workshops . . . . . . . . . . . . . . . . . . . . . . . . 101
7.3 Overview planning questions . . . . . . . . . . . . . . . . . . . . . . . . 103
7.4 User feedback - general . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7.5 User feedback - usability . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7.6 User feedback - utility . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Elias Pajares XVII
List of Abbreviations
List of Abbreviations
ALKIS Amtliches Liegenschaftskatasterinformationssystem.
API Application Programming Interface.
ATKIS Amtliches Topographisch-Kartographisches Informationssystem.
COVID-19 Coronavirus Disease 2019.
EPSG European Petroleum Survey Group / Nowadays EPSG Geodetic Parameter
Dataset.
EU European Union.
GIS Geographic Information Systems.
GOAT Geo Open Accessibility Tool.
GTFS General Transit Feed Specification.
JOSM Java OpenStreetMap Editor.
LOD Level of Detail.
MVT Mapbox Vector Tile.
OGC Open Geo Spatial Consortium.
OSM OpenStreetMap.
OSS Open Source Software.
PL/pgSQL Procedural Language/PostgreSQL.
POI Point of Interest.
Elias Pajares XIX
PSS Planning Support System.
RMSE Root-Mean-Square Error.
SQL Structural Query Language.
TUM Technical University of Munich.
USA United States of America.
VGI Volunteered Geographic Information.
WebGIS Web Geographic Information System.
WFS Web Feature Service.
WMS Web Map Service.
XML Extensible Markup Language.
Part I
Introduction, state of the art, and
research design
Elias Pajares 1
Chapter 1. Introduction
Chapter 1
Introduction
1.1 Background
A critical debate about the role of transportation in urban contexts is taking place in
cities worldwide. It is undeniable that transportation networks and technology are key
enablers of economic prosperity. However, they are accompanied by an array of exter-
nalities. Transportation can place local communities at severe risk and threaten local
and global sustainability. Globally, it is estimated that 3.7 million people die prematurely
because of outdoor air pollution, largely because of transport emissions (WHO, 2019).
Meanwhile, it is estimated that 1.35 million people die annually due to crashes in road
traffic (World Health Organization et al., 2019). In the European Union (EU), 26% of
the carbon dioxide emissions were related to road transportation in 2018 (Statistisches
Bundesamt, 2021). At 62% passenger cars and motorcycles are the main contributors.
Across multiple sectors, carbon dioxide emissions fell by 23% in the EU between 1990
and 2018. However, in the same period, carbon dioxide emissions caused by road
transportation rose by 24% (Statistisches Bundesamt, 2021). The adoption of electric
cars in Europe will accelerate fast until 2030 (IEA, 2021). In 2030, 7% of the global
vehicle fleet in the stated policy scenario, and 12% of the global vehicle fleet in the sus-
tainable development scenario will have an electric engine (IEA, 2021). At the same
time, large portions of the vehicle fleet will still be powered by a combustion engine,
and even if electric, energy will often come from fossil fuels. Meanwhile, a change in
vehicle technology does not answer the scarcity of space in cities. In cities like Munich,
17% of the area is dedicated to transportation (Geisser & Lenk, 2017). Additionally,
traffic congestion severely affects the quality of life in cities.
There is increasing awareness that simply expanding transport infrastructure or pro-
viding additional mobility options cannot solve the described problems. Instead, it is
suggested that mobility and therefore the need for transport is understood as a derived
Elias Pajares 3
phenomenon. Since people strive for accessibility to destinations, increasing people’s
accessibility should be the primary concern (Handy, 2020;Levine, 2019;Silva et al.,
2019;Venter, 2016). Meanwhile, Silva and Larsson (2019) have highlighted the risks
of accessibility planning that addresses accessibility problems solely by transport or
mobility measures. Instead, a stronger focus is suggested on increasing proximity to
destinations (Silva & Larsson, 2019). Strengthening proximity might be particularly rel-
evant since, during the last decades, a general decline in local accessibility could be
observed in many western countries due to the concentration of activities (Silva, 2020).
The fast trending concept of the 15-Minute City and variations such as the 10-Minute
or 20-Minute City are part of a counter-movement. The 15-Minute City concepts pro-
vides a clear message that is widely understood and can be defined as an "urban setup
where locals are able to access all of their basic essentials at distances that would not
take them more than 15 min by foot or by bicycle" (C. Moreno et al., 2021).
1.2 Motivation
Overall, the concept of accessibility has great potential to provide an integrated frame-
work for sustainable transportation and land-use planning (Bertolini et al., 2005). Car-
fully increasing accessibility can be understood as the pre-condition for sustainable
mobility (Couclelis, 2000). By creating proximity to opportunities and providing access
to sustainable forms of transportation, alternatives to the personal car can facilitate
more sustainable transport and land-use patterns. Applying accessibility, therefore,
has the high potential to make transportation and urban development more sustainable
(Bertolini et al., 2005;Bertolini & Silva, 2019;Levine, 2019). While there is hope that
accessibility is increasingly adopted (Handy, 2020), Planning Support System (PSS)
in general (Geertman et al., 2017b;Russo et al., 2017;te Brömmelstroet, 2017) and,
more specifically, accessibility instruments to operationalize the theoretical concept are
not yet widely used in practice (Bertolini & Silva, 2019;Boisjoly & El-Geneidy, 2017b;
Hull et al., 2012;Papa et al., 2015;te Brömmelstroet et al., 2016;te Brömmelstroet et
al., 2014;Wulfhorst et al., 2017). This phenomenon is commonly described as an im-
plementation gap, and past research has identified different explanatons for it. Levine
(2019) underlined that classical mobility measures persist because transport engineers
and spatial planners are asked to use them. Furthermore, there is evidence that ac-
cessibility is often conceptually misinterpreted (Levine, 2019).
te Brömmelstroet et al. (2016) highlighted that tool development is often discon-
nected from the users, running the risk that the actual problems in the planning practice
are not addressed. Furthermore, a lack of data makes the application of accessibility
instruments challenging and expensive (Boisjoly & El-Geneidy, 2017a;Papa et al.,
Chapter 1. Introduction
2015;te Brömmelstroet et al., 2014). Others have claimed that there is a lack of knowl-
edge (Boisjoly & El-Geneidy, 2017b) and resources (te Brömmelstroet et al., 2014) to
use accessibility analyses in practice. Users asked for more interactive instruments
providing real-time scenario calculations (Silva et al., 2017;te Brömmelstroet, 2017;
te Brömmelstroet et al., 2014;Wulfhorst et al., 2017). Another concern is that acces-
sibility instruments are often perceived as "complex, inflexible, incomprehensible, and
rigid black boxes" (Papa et al., 2015).
1.3 Research objectives
This research aims to develop a novel accessibility instrument. This work is an ap-
plied research project that makes use of ongoing technological development. Previous
research identified the close involvement of potential users as a core requirement for
developing a usable and useful accessibility instrument (Bertolini & Silva, 2019;Silva
et al., 2017;te Brömmelstroet et al., 2016;te Brömmelstroet et al., 2014). Meanwhile,
numerous cities worldwide are willing to foster active mobility and adopt concepts such
as the 15-Minute City (C40 Cities, 2020). The concept of the 15-Minute City aims to
increase local accessibility. Local accessibility, in contrast to regional accessibility fo-
cuses on short and frequent trips to different destinations (Handy, 1993). Silva and
Altieri (2022) described local accessibility as proximity-based and regional accessibility
as mobility-based.
As active mobility relies on the ability to travel short distances to destinations, high
local accessibility is the basis for a high share of active mobility. Active mobility typically
groups the transport modes of walking and cycling (Koszowski et al., 2019a). However,
also other typically non-motorized modes (e.g., skateboards) fall under this category.
From the own perspective all forms of mobility that require substantial physical activity
can be classified as active mobility. However, this research focuses on walking and
cycling.
The willingness to gear mobility toward a higher share of active mobility from prac-
tice, paired with the low availability of planning support when shaping active mobility,
provides a unique opportunity to bridge the implementation gap, particularly for active
mobility and local accessibility. Consequently, this dissertation proposes a tool de-
velopment that focuses on creating an interactive, transferable, web-based, and open
solution called Geo Open Accessibility Tool (GOAT), which focuses on planning active
mobility and local accessibility. It aimed to make accessibility analyses for active mo-
bility applicable to a much wider group of users. Accordingly, the GOAT aims to serve
as a valuable contribution to the landscape of accessibility instruments and help shape
sustainable mobility.
Elias Pajares 5
Figure 1.1: Research aim
In sum, the contributions of this research is the development of a useful (combining
utility and usability) accessibility instrument, which can be applied with a reasonable
amount of resources. Consequently, a widely applicable accessibility instrument should
be developed. A more straightforward application of accessibility in practice ultimately
can contribute to a more sustainable reality in transport and land-use development (see
Figure 1.1). While, an assessment of how the development influences transport and
land-use development is beyond the scope of the thesis, it serves as the vision and
underlying motivation of the thesis. More specifically, the research objectives of this
dissertation are as follows:
• Identification of areas in which accessibility instruments should be developed fur-
ther to meet the shortcomings of existing accessibility instruments.
• Development and application of an interactive, transferable, web-based, and open
accessibility instrument for active mobility and local accessibility.
• Development and application of novel and open data strategies to collect, pre-
pare and fuse spatial data to facilitate more powerful and affordable accessibility
instruments.
• Assessment and discussion of the usefulness of the developed instrument for the
planning practice.
• Collection and sharing of the expertise of the co-creative development of an ac-
cessibility instrument with practice and other tool developers.
Chapter 1. Introduction
1.4 Thesis structure
This dissertation is comprised of nine chapters. It starts with an introduction to the
research topic in Chapter 1, followed by a short literature review in Chapter 2 on the
essentials of accessibility research. Afterward, the three research questions and main
goals are presented in Chapter 3. While the primary goal is the development of the
novel accessibility instrument GOAT, the three research questions address tool require-
ments, data management, and tool assessment. In Chapter 4 the research design is
presented, characterized by the tool development, tool application, and the co-creative
involvement process of practitioners, researchers, and students. As a paper-based
dissertation, the following three chapters present each one scientific paper:
• Paper 1: Accessibility by proximity: Addressing the lack of interactive accessibility
instruments for active mobility. Pajares, E., Büttner, B., Jehle, U., Nichols, A.,
Wulfhorst, G.
• Paper 2: Population Disaggregation on the Building Level Based on Outdated
Census Data. Pajares, E., Muñoz Nieto, R., Meng, L., Wulfhorst, G.
• Paper 3: Assessment of the usefulness of the interactive accessibility instrument
GOAT for the planning practice. Pajares, E., Jehle, U., Hall, J., Miramontes, M.,
Wulfhorst, G.
The synthesis and discussion of the main research findings are presented in Chap-
ter 8. The results are split into sections per research question and one for the main
goal. Finally, in Chapter 9 the conclusion is presented, consisting of the study limi-
tations, the discussion of future development paths, and final reflections. Figure 1.2
visualizes the structure of the thesis.
Elias Pajares 7
Figure 1.2: Thesis structure
Chapter 2. Literature review
Chapter 2
Literature review
2.1 Accessibility theory
Accessibility has faced the challenge that it was hardly tangible for a long time. Gould
(1969) described the dilemma as "Accessibility (...) It is a slippery notion, however;
one of those common terms that everyone uses until faced with the problem of defining
and measuring it!". The first definition in literature dates back to Hansen (1959), who
defined accessibility as "the potential of opportunities for interaction".
It is essential to differentiate mobility and transportation from accessibility in this
context. Mobility can be defined as the "potential for movement" (Handy, 2002), while
transportation is the "movement of goods and persons from place to place" (Britannica,
The Editors of Encyclopaedia, 2019). Scholars agree that transportation is a derived
phenomenon, that arises from the need of a person to realize activities that are not at
the person’s origin (Levine et al., 2012;Meyer & Miller, 1984). This need can be met by
increasing mobility, such as expanding the highway network or increasing proximity to
destinations (Levine, 2019;Levine et al., 2012). There are places (e.g., the city center)
where mobility is low due to slow travel speeds, but accessibility is high thanks to the
high density of opportunities. Meanwhile, in suburban or rural contexts, accessibility
is lower despite the commonly higher travel speeds (by car) due to the low density of
opportunities (Handy & Niemeier, 1997;Levine et al., 2012). Therefore, accessibility is
related to the performance of transportation and land-use (Bertolini et al., 2005).
Overall, optimizing accessibility can be more appropriate than widespread mobility-
focused transportation planning. As it serves the actual needs of people, which is
realizing activities instead of traveling fast nowhere. Since Hansen’s early definition,
researchers have focused on developing appropriate ways to measure accessibility
(Handy & Niemeier, 1997;Ingram, 1971;Koenig, 1980). Geurs and van Wee (2004)
systematized and categorized different accessibility measures and proposed four com-
Elias Pajares 9
ponents: transport, land-use, temporal and individual.
Figure 2.1: Four accessibility components
(Geurs & van Wee, 2004)
2.2 Accessibility measures
Ideally, an accessibility measure should include all four accessibility components (Geurs
& van Wee, 2004). Accessibility measures can be classified as infrastructure-based,
location-based, person-based, and utility-based (Geurs & van Wee, 2004). Infrastructure-
based measures do not have a land-use dimension and instead solely capture the per-
formance indicators of the transport system (e.g., travel speed). Person-based mea-
sures show accessibility from an individual standpoint. Therefore, personal constraints
such as the daytime availability or the person’s physical capabilities are considered
(Geurs & van Wee, 2004). Person-based accessibility measures are based on the
space-time geography defined by Hägerstrand (1970). Utility-based accessibility mea-
sures aim to picture the benefits (e.g., social and economic) associated with acces-
Chapter 2. Literature review
sibility (Ben-Akiva & Stevan R., 1979;Geurs & van Wee, 2004;Handy & Niemeier,
1997).
The most widely used set of accessibility measures is location-based. These in-
clude contour and gravity-based measures (Geurs & van Wee, 2004;Handy & Niemeier,
1997;Iacono et al., 2010). A typical contour measure is the travel time isochrone,
showing how far one can travel in a certain length of time. The first examples of
isochrones were published in studies of the 19th century such as Isochronic Passage-
Charts by Galton (1881) and Isochronenkarte der österreichisch-ungarischen Monar-
chie by Penck (1889). Nowadays, isochrones are often intersected with spatial data
(e.g., points of interest or population) to compute the sum of reached opportunities in
a given travel time, called cumulative opportunities (Geurs & van Eck, 2001;Geurs &
van Wee, 2004). The isochrone as an accessibility measure has clear benefits as they
are relatively easy to calculate and understandable by a wider audience (Bertolini et al.,
2005;Geurs & van Eck, 2001). At the same time, one limitation of using cumulative
opportunities is that there is no differentiation between different travel times within the
cutoff range (Bertolini et al., 2005;Bhat et al., 2002;El-Geneidy & Levinson, 2006).
Gravity-based accessibility measures were first introduced by Hansen (1959). Un-
like contour-based measures, accessibility to opportunities is evaluated by the gener-
alized travel cost using an impedance function (Bhat et al., 2002;Geurs & van Eck,
2001;Handy & Niemeier, 1997). In this method, opportunities located further away are
weighted lower than closer opportunities. This approach is in line with the first law of
geography, which states that: ëverything is related to everything else, but near things
are more related than distant things" (Tobler, 1970). Accordingly, gravity-based acces-
sibility measures can be expressed using the following formula:
Ai=PjOj∗f(tij)
In which Aiis the accessibility at location i,Ojare the opportunities at location j
and tij is the travel cost between iand j. While the most common type of travel cost
is travel time, different types of cost can be utilized to compute impedances such as
monetary (Büttner, 2016) or emissions (Kinigadner, 2020). Furthermore, it is decisive
which function is used for computing the impedance. The impedance function generally
involves the traveltime and a sensitivity parameter that is commonly represented with
the letter β. The following frequently used impedance functions were summarized by
Kwan (1998):
•Inverse Power
Elias Pajares 11
f(tij) =
1for tij <1
tij−βelse
•Negative Exponential
f(tij) = e−β×tij
•Modified Gaussian
f(tij) = et2
ij /β
•Cumulative Opportunities Rectangular
f(tij) =
1for tij ≤tmax
0else
•Cumulative Opportunities Linear
f(tij) =
(1 −tij/tmax)f or tij ≤tmax
0else
As shown in Figure 2.2 different functions significantly influence the computed impedance.
The shown impedance functions using the indicated sensitivity parameters and travel
times in minutes. The negative exponential functions is frequently used (Iacono et al.,
2010;Vale & Pereira, 2017). Vale and Pereira (2017) stated that "the literature sup-
ports the concept of a certain indifference toward marginal distances." This occurrence
favors functions that are less sensitive to changes in the travel time in the beginning
and fall off more quickly with increasing travel time. The presented functions, espe-
cially the Gaussian one, fall under this category. Already in early accessibility research,
the Gaussian function was already considered superior to other functions (Bhat et al.,
2002;Ingram, 1971).
Chapter 2. Literature review
Figure 2.2: Comparison of the most common impedance functions
(Own visualization based on (Higgins, 2019))
Meanwhile, still little application of the Gaussian function is observed (Vale & Pereira,
2017). The literature agrees that selecting a suitable impedance function should be in-
fluenced by the studied transport mode and trip purpose (Iacono et al., 2010;Vale &
Pereira, 2017). Besides finding a suitable impedance function, appropriate sensitivity
parameters strongly influences the inverse power, negative exponential, and modified
Gaussian function. It has also been suggested that the sensitivity parameter should
be calibrated with data from real-world travel behavior (Geurs & van Eck, 2001;Handy
& Niemeier, 1997;Iacono et al., 2008). Meanwhile, suitable data for calibration is of-
ten unavailable. Classical household surveys are often not granular enough. Ideally,
calibration would differentiate between POIs and population groups, as the impedance
might vary significantly for different demographic attributes (e.g., age, gender, and so-
cial status) (Handy & Niemeier, 1997). Chapter 5.5.5 presents the gravity-based mea-
sure in GOAT; due to the lack of suitable data for calibrating the impedance functions,
the sensitivity parameter can be adjusted. It was decided to use the modified Gaussian
function, which was calculated with travel times in seconds.
Elias Pajares 13
2.3 Accessibility instruments
Accessibility instruments are usually GIS-based tools that aim to translate the accessi-
bility concept into operational planning support. Papa et al. (2015) state that:
"Accessibility instruments (...) are a type of planning support system (PSS) de-
signed to support integrated land-use transport analysis and planning through
providing explicit knowledge on the accessibility of land uses by different modes
of transport at various geographical scales."
Generally, accessibility instruments simplify the complex and often subjective na-
ture of accessibility. The term accessibility instrument is comparatively new; it was
mainly consolidated in the COST Action project on "Accessibility Instruments for Plan-
ning Practice in Europe" (Hull et al., 2012;te Brömmelstroet et al., 2014). Accordingly,
there might be cases in which another term is used. Accessibility instruments are often
referred to as accessibility tools. However, the literature does not clearly distinguish
between occasional accessibility analyses, commonly created using GIS software, and
an accessibility instrument. An accessibility instrument is understood as a specialized
tool mainly designed to perform accessibility analyses in this dissertation. Accordingly,
other tools capable of conducting specific accessibility analyses, such as macroscopic
transport models, trip planners, or multi-purpose GIS software do not fall under this
category. In Chapter 8.1.1, a classification of the different technical components is pro-
posed. A wide range of accessibility instruments exist for different purposes (Silva et
al., 2019;TUM - Chair of Urban Structure and Transport Planning, 2021) and analyses
are conducted for different transport modes and spatial scales (Papa et al., 2015;TUM
- Chair of Urban Structure and Transport Planning, 2021). The concrete planning ques-
tions and data availability might influence the tool’s design and application in practice.
Furthermore, tools are developed as desktop applications, software extensions, or web
tools (TUM - Chair of Urban Structure and Transport Planning, 2021). So far, most of
the tools have been developed in academic projects and have shown dynamic devel-
opment but, to some extent, also immaturity. Furthermore, academic developments
always risk being discontinued when project funding ends. A richer insight into the tool
landscape and potential for new developments is provided in Chapter 5.
Chapter 3. Research questions and goals
Chapter 3
Research questions and goals
This chapter presents the research questions and the main goal. Previous research
has underlined the great potential of accessibility-based planning. Moreover, it has
been emphasized that accessibility instruments are not frequently used in the planning
practice. Therefore, a collaborative development process through the involvement of
planning practitioners has the potential to develop useful instruments. Furthermore,
the rapid growth of web technologies and GIS and the increasing availability of (open)
data facilitate faster and more affordable development. Accordingly, there is a high
potential for developing a new generation of interactive accessibility instruments based
on open technology and transparent indicators.
3.1 Main goal
The main goal is the development and application of the accessibility instrument GOAT.
As discussed in Chapter 1.3 the development and application of the development tools
should ultimately contribute to more sustainable development. The particular focus of
the development can be summarized as follows:
Development of an open source, interactive, and transferable accessibility instru-
ment for planning local accessibility by active mobility.
Due to its deterministic nature, the development is defined as a goal rather than
a research question. The main goal is addressed throughout the research process
and can be understood as an important deliverable of the presented work. It reacts
to the identified gaps in accessibility research and instruments. Moreover, it is closely
interrelated with the three research questions, and it is the basis for answering the
three research questions. RQ1 is starting ground for tool development. In comparison,
Elias Pajares 15
the tool development provides the basis for RQ2 and RQ3. Overall, this dissertation
understands the development as an enabler to answering the research questions and,
more importantly, to help answer real-world planning questions. The tool development
should be understood as a vehicle to support sustainable urban and transport planning
by using accessibility as a holistic framework.
3.2 RQ1 Tool requirements
In which areas do accessibility instruments have to be developed further to better sup-
port planning practice when planning for active mobility?
There is a rising awareness of the need to promote active mobility and local accessi-
bility. At the same time, PSS are rarely applied in practice (see Chapter 2.3 and 7.6).
While the implementation gap is a general trend across all transport modes, the lack
of other planning tools (e.g., macroscopic transport models) for active mobility and the
rising importance of providing favorable conditions for active mobility show a particular
need. Consequently, this dissertation concentrates on the implementation gap in active
mobility planning. This research question focuses on studying and classifying existing
accessibility instruments. In particular, it assesses if highly desired features by the plan-
ning practice in the past, such as interactive scenario building, are implemented and if
the tools support walking and cycling. Furthermore, the tool access, transferability, and
type are examined to understand the ease of applying the tool in practice.
3.3 RQ2 Data management
How can data refinement and fusion strategies enable powerful and affordable acces-
sibility instruments?
Accessibility instruments are data-driven tools; therefore, the availability of data paired
with the need to harness its full potential is essential. The ever-increasing availability
of suitable data offers increased opportunities for a new generation of accessibility in-
struments, facilitating analyses on the local scale. Overall, the data is the basis for the
development of powerful accessibility instruments. At the same time, high data collec-
tion or acquisition costs often make accessibility instruments unaffordable. The data
needed for an accessibility instrument is diverse and strongly influenced by the study
area and concrete planning questions. In the context of the dissertation, the availability
of population data on the building level and high-quality POIs data are particular chal-
lenges. Therefore, this research question focuses on developing a novel population
Chapter 3. Research questions and goals
disaggregation method based on widely available data. Moreover, data fusion strate-
gies are designed to combine diverse sets of POIs. This is accompanied by exploring
how collecting of Volunteered Geographic Information (VGI) in OpenStreetMap (OSM)
can increase the data quality and availability.
3.4 RQ3 Tool assessment
Is the accessibility instrument GOAT of useful support in the planning practice?
This research question critically examines whether the developed accessibility instru-
ment GOAT is useful when planning active mobility and local accessibility. Usefulness
is differentiated into two components utility and usability. In this context, utility describes
whether the developed tool provides relevant analyses for concrete planning questions
and usability focuses on the ease and intuitiveness of GOAT for the planning questions.
If possible, the results are broken down into individual use cases (e.g., infrastructure
planning walking) and discussed on a more aggregated level. Areas where the instru-
ment already meets the requirements in practice and where it does not are highlighted.
Accordingly, the tool assessment serves as the basis for the reflection on the quality of
the instrument and future development paths.
Elias Pajares 17
Chapter 4. Research design
Chapter 4
Research design
Figure 4.1: Connections between main goal, research questions, and methods
The dissertation follows the research design visualized in Figure 4.1. The starting point
was the identified need for a new accessibility instrument, and concrete development
needs were further expanded in RQ1. Additionally, the requirements identified in the
literature and the research process were translated into the presented main goal (see
Chapter 3.1). Three research questions were also developed that summarize the sci-
entific contribution of the thesis. An iterative software development cycle characterizes
the research design. The tool was first developed with basic functionality; it was then
applied and intensively tested through the co-creative involvement of practitioners. Ac-
cordingly, the tool development was accompanied by an early application of the tool.
First, the tool was applied in one study context and, later, in several case studies. The
Elias Pajares 19
open access provision of the application facilitated early involvement of the developer
community and especially of the planning practice, which was the target user group of
the development.
4.1 Tool development
The development of the tool GOAT (see Figure 4.2) was first initialized in the author’s
master’s thesis (Pajares, 2017). During the master’s thesis, a prototype was developed,
serving as a minimum viable product that facilitated online testing of its functionality. In
early 2018, the tool development was continued at the Chair of Urban Structure and
Transport Planning at TUM. Accordingly the development in the scope of this disserta-
tion started in early 2018. Until October 2019, the development was realized without
external funding, and therefore progress was relatively slow. Nevertheless, by the end
of 2019, GOAT 0.1 was released, with many features that GOAT still contains today.
Figure 4.2: Development timeline
From November 2019 until February 2021, the GOAT development was supported
by the mFUND initiative (BMDV, 2022). This project funding facilitated the develop-
ment and implementation of new features such as accessibility analysis for cycling and
provided the project context for the co-creative involvement of practitioners (see Chap-
ter 4.3). The project ended with the release of GOAT version 1.0 (GOAT-Community,
2021c). Since early 2021, the development of GOAT has been continued by the startup
Plan4Better and in joint research and application projects with TUM, as well as other
practice and research partners. However, this dissertation focuses on the tool’s de-
velopment from early 2018 to the release of GOAT 1.0 in March 2021. The develop-
ment was realized as an open source project (GOAT-Community, 2021c) that involved
other researchers and developers. The author led the development team, translating
Chapter 4. Research design
the needs identified through co-creative involvement into concrete technical require-
ments. Furthermore, the author conceptualized and implemented the tool architecture
and the backend development, which contained most of the tool logic and indicators.
The frontend software development and DevOps were supported by additional devel-
opers most notably Majk Shkurti (Majk Shkurti, 2022). The contributions of different
collaborators to GOAT’s codebase can be explored in the project’s Github repository
(GOAT-Community, 2021d).
Elias Pajares 21
4.2 Tool application
Table 4.1: Applied GOAT versions
Study context Project Year Live
deployment
Municipality of Munich First version and mFUND 2018 Yes
Municipality of Fürstenfeldbruck mFUND 2020 Yes
Municipality of Freising mFUND 2020 Yes
County of Freising and
Erding
Mobility concept for
Mittlere Isarregion
and Ampertal
2019 No
Municipality of Freiburg Plan4Better project 2021 Yes
Municipality of Boca Raton Plan4Better project 2021 no
Municipality of Istanbul Plan4Better project 2021 No
Municipality of Bogotá Master’s thesis
(Munoz Nieto, 2020)
2020 Yes
Municipality of San Pedro Garza García Master’s thesis
(Ivanov, 2021)
2021 Yes
Municipality of Matosinhos Study project
(Viertler, 2020)
2020 Yes
District of Hasenbergl (Munich) Master’s thesis
(Jehle, 2020)
2020 Yes
District of Neuperlach and
Maxvorstadt (Munich)
Bachelor’s thesis
(Ben Hassine, 2019)
2019 No
County of Starnberg Study project
(Schott, 2020)
2020 No
Municipality of Augsburg, Wolfratshausen
and Münsing
Bachelor’s thesis
(Diepolder, 2020)
2020 No
Municipality of Magdeburg Bachelor’s thesis
(Zuckriegl, 2021)
2021 No
Municipality of Berlin, Heidelberg, Cottbus,
Flensburg, Neustadt a.d. Donau, Eppel-
born, Oschersleben, Nordkirchen, Zorned-
ing, Ronneburg, Seeshaupt and Rackwitz
Master’s thesis
(Viertler, 2020)
2020 No
Municipality of Magdeburg Bachelor’s thesis
(Zuckriegl, 2021)
2021 No
Municipality of Atlanta Study project
(Hanekamp, 2021)
2021 No
Chapter 4. Research design
The different deployments varied in focus (see Chapter 4.1). Some experimented with
new features, others tested the transfer to a new study context, and others were used
in workshops or real-world consulting projects. While various versions of GOAT were
used for the application over the years, they can all be considered pre-releases of GOAT
version 1.0. Not all versions were deployed live on a server; many were only set up
locally for testing and application. The deployed version were provided open access
so everyone could test the tool. Furthermore, as the tool is open source, other parties
transferred it to the author’s knowledge to additional study contexts.
Figure 4.3: Applied GOAT version worldwide
This dissertation primarily focuses on the application of GOAT in Munich, Freising,
and Fürstenfeldbruck. Therefore, most of applications happened in the Munich region
(see Figure 4.3) due to the dedicated funding that facilitated the development of most
features within the mentioned municipalities as a study case. Furthermore, the funding
allowed the version to remain online for a longer time frame and to receive more exten-
sive feedback. The different versions were usually equipped with different data sets. In
addition, the data was enriched with data collected in OpenStreetMap and Mapillary. A
complete overview of the core data sets used can be found in Chapter 5.
Elias Pajares 23
4.3 Co-creative involvement
Different groups have tested and used the tool since its first release in early 2018 (see
Figure 4.4). Feedback and new ideas were considered whenever possible during the
development process. The open access provision of the tool on the project website
allowed individuals to test the tool independently and spontaneous and unstructured
feedback was obtained either via e-mail or orally. Furthermore, a website with a blog
and activities on social media (i.e. LinkedIn and Twitter) was used to communicate the
project’s progress and engage with </