pdeng final reportspecification of the range of allowable cardinalities that a set may assume [iso...
TRANSCRIPT
PDEng – Final report
Modelling utilities by developing a domain ontology
PDEng Candidate Date Version
Ramon ter Huurne 11-01-2019 2.0
PDEng – Final Report I
Colophon
PDEng Candidate
Name
Ir. R.B.A. (Ramon) Huurne
Organization
Name
Faculty
Department
Contact details
University of Twente
Engineering Technology (ET)
Construction Management & Engineering
P.O. Box 217
7500 AE Enschede
Client
Name
Contact details
Campus and Facility Management University of Twente
Dienstweg 5
7522 ND Enschede
Graduation Committee
Dr. drs. J.T. (Hans) Voordijk
Prof. dr. ir. A.G. (André) Dorée
Dr. ir. L.L. (Léon) olde Scholtenhuis
Ing. R. (Ray) Klumpert
H.J.M. (André) de Brouwer
Dr. C.P.L. (Carl) Schultz
C. (Caroline) Groot
University of Twente
University of Twente
University of Twente
Campus & Facility Management
Campus & Facility Management
Aarhus University
Kadaster
PDEng Programme Director
Thesis Supervisor
Daily Supervisor
Company Supervisor
Company Supervisor
External Examiner
External Examiner
PDEng – Final Report II
Document characteristics
Title
ID
Author
Date
Subject
Publisher
Type
Format
Language
Version
PDEng – Final report
PDEng_Final_Report.pdf
Ramon ter Huurne
Date of last adjustment – 11-01-2019
Modelling utilities by developing a domain ontology
University of Twente
Text
Portable Document Format (PDF)
English
2.0
Changelog
Version Date Changes
made by
Sections
changed
Change(s) made
2.0 11-01-2019 Ramon ter
Huurne
All
Chapter 6
Chapter 7
Appendix VI
Textual changes throughout the whole report
Added section about the gradual development of the
ontology
Added proof of concept
Added possible BSC / MSc research subjects
1.0 05-12-2018 Ramon ter
Huurne
All First complete version of the report
List of figures
Figure 1 – Engineering cycle by Wieringa (2014) applied to the project context and its objectives 9
Figure 2 – Stakeholder mapping following Alexander’s (2005) onion model 16
Figure 3 – Distribution of attribute types within elicited realities 19
Figure 4 – The place of a domain ontology within the industry (left as-is, right with a domain ontology included) 25
Figure 5 – UML notation (ISO 19103:2015 Geographic Information – Conceptual Schema Language) 33
Figure 6 – Example of a UML class model (from ‘Operations and Maintenance – Data specification’) 35
Figure 7 – Network components UML class model (from ‘Operations and Maintenance – Data Specification’) 38
Figure 8 – Utility data modelled according to the domain ontology’s format in QGIS 46
PDEng – Final Report III
List of tables
Table 1 – Project characteristics 1
Table 2 – Stakeholders within the project 15
Table 3 – Snap of empirical analysed utility data from the telecom domain 18
Table 4 – Snap of empirical analysed utility data from the gas domain 18
Table 5 – Comparison between and overview of utility digital modelling standards 22
Table 6 – Goals, needs and requirements following the INCOSE method 31
Table 7 – UML stereotypes applied in the domain ontology 34
Table 8 – UML values types applied in the ontology 34
Table 9 – Reporting style applied within Feature Catalogue for feature types and data types 39
Table 10 – Reporting style applied within Feature Catalogue for codelist and enumerations 39
Table 11 – Changelog of ontology development process 42
List of publications
All publications can be found in Appendix VII of this report. Publications included are the following:
ter Huurne, R.B.A. (2018). Digitization for integration: fragmented realities in the utility sector. 92-100.
Paper presented at ARCOM 2018, Belfast, United Kingdom.
ter Huurne, R.B.A. (2018). Introductie van een uniform objectmodel voor het beheer en onderhoud van
ondergrondse infrastructuur. 1-9. Paper presented at CRWO Infradagen 2018, Arnhem, Netherlands.
PDEng – Final Report IV
Foreword and acknowledgements
In front of you lies the report of my PDEng project. A project I have been working on for the past two
years. A project that gave me lots of new insights and knowledge, but also brought me new friendships and
good memories. A project that not could have been accomplished without the help of many people.
Therefore, I want to take the opportunity here in the report to thank all of those.
First of all, I have to thank all of my supervisors for their time, enthusiasm, and thoughtful advices and
recommendations. In specific, I want to thank prof. dr. ir. André Dorée and dr. ir. Léon olde Scholtenhuis
as my supervisors from the University of Twente and ing. Ray Klumpert and André de Brouwer as my
supervisors from the Campus and Facility Management of the University of Twente.
Second, I want to thank the experts being part of my expert panel for their feedback and useful advice,
allowing me to shape the ontology in the way it is presented in this report. In specific, I want to thank the
representative(s) from CityGML UtilityNetwork ADE, Siers Infraconsultancy, Geonovum, and the
Kadaster.
Third, I am very grateful to all my colleagues and friends from the department of Construction
Management and Engineering I have been working with for the past two years. Even though the workload
was quite high several times, it never felt as a burden for me to work, due to the amazing atmosphere at
our office. Even more, I learned a lot from all of your projects and work, and I want to thank you for the
support and encouragement you gave me.
Fourth, but definitely not last, I wish to extent a huge thank to my girlfriend Evelien, my family, and my
friends who all supported me throughout these two years. I cannot thank you all enough for your
encouragement and understanding.
PDEng – Final Report V
Terms and definitions
[1.] Aggregation
Special form of association (3) that specifies a whole-part relationship (24) between the aggregate (whole)
and a component (7) part [UML 1]
[2.] Application
Manipulation and processing of data in support of user requirements [ISO 19101-1:2014]
[3.] Association
Semantic relationship (24) that can occur between classes [UML 2]
[4.] Association role
Explanation of how a class participates in the association.
[5.] Attribute
Feature (14) within a classifier that describes a range of values that instances of the classifier may hold
[UML 1]
[6.] Class
Description of a set of objects (21) that share the same attributes (5), operations, methods, relationships
(24), and semantics [UML 1]
[7.] Component
Representation of a modular part of a system that encapsulates its contents and whose manifestation is
replaceable within its environment [UML 2]
[8.] Composition
Aggregation (1) where the composite object (21) (whole) has responsibility for the existence and the storage
of the composed objects (parts) [UML 2]
[9.] Conceptual model
Model that defines concepts of a universe of discourse [ISO 19101-1:2014]
[10.] Conceptual schema
Formal description of a conceptual model [ISO 19101-1:2014]
[11.] Data specification
Detailed description of a data set or data set series together with additional information that will enable it
to be created, supplied to and used by another party [ISO 19131:2007]
[12.] Data type
Specification of a value domain with operations allowed on values in this domain [ISO 19103:2015]
[13.] Depth
Distance of a point from a chosen reference surface measured downward along a line perpendicular to
that surface [ISO 19111:2007]
PDEng – Final Report VI
[14.] Feature
Abstraction of a real world phenomenon [ISO 19101-1:2014] In UML also defined as the property of a
classifier [UML 2]
[15.] Feature catalogue
Catalogue containing definitions and descriptions of the feature (14) types, feature attributes, and feature
relationships occurring in one or more spatial data sets, together with any feature operations that may be
applied [ISO 19101-1:2014]
[16.] Feature type
Class of features (14) having common characteristics [ISO 19156:2011]
[17.] Generalization (inheritance)
Taxonomic relationship (24) between a more general element and a more specific element of the same
element type [UML 2]
[18.] Infrastructure asset management
Set of activities performed and strategies implemented with the goal to preserve and extend the service life
of the utilities
[19.] Interoperability
Capability to communicate, execute programs, or transfer data among various functional units in a manner
that requires the user to have little or no knowledge of the unique characteristics of those units [ISO/IEC
2382:2009, 212317]
[20.] Multiplicity
Specification of the range of allowable cardinalities that a set may assume [ISO 19103:2015]
[21.] Object
Entity with a well-defined boundary and identity that encapsulates state and behavior [UML 1]
[22.] Ontology
Formal and explicit specifications of shared conceptualizations. In the report also defined as a digital
modelling standard
[23.] Operations and maintenance
See ‘Infrastructure asset management’ (18)
[24.] Relationship
Semantic connection among model elements [UML 1]
[25.] Stereotype
Extension of an existing metaclass that enables the use of platform or domain specific terminology or
notation in place of, or in addition to, the ones used for the extended metaclass [UML 2]
[26.] Utility infrastructure
Infrastructure with the purpose of transporting commodities.
PDEng – Final Report VII
Abbreviations
AEC
Architecture, Engineering and Construction
ADE
Application Domain Extension
BIM
Building Information Modeling
CFM
Campus and Facility Management
GIS
Geographic Information System
GML
Geography Markup Language
IFC
Industry Foundation Classes
IMKL
Informatiemodel Kabels en Leidingen (ENG – Information model Cables and Pipes)
INSPIRE
Infrastructure for Spatial Information in Europe
ISO
International Organization for Standardization
OGC
Open Geospatial Consortium
UML
Unified Modelling Language
XML
Extensible Markup Language
PDEng – Final Report VIII
Graphical summary
PDEng – Final Report IX
Executive summary
The utility sector is shifting more towards a life cycle oriented management approach. This means utility
owners increasingly want to have a comprehensive set of digital information about their utilities. This need
is also recognized at the Campus and Facility Management (CFM) of the University of Twente. The CFM
– client within the PDEng project – is responsible for the installation, operations and maintenance of their
utilities. Consequently, the CFM wants to be able to rely on uniform and consistent digital information
about these utilities.
However, the problem, and thereby the starting point of the PDEng project, is that the utility sector
currently lacks a digital modelling standard with the emphasis on lifecycle management, i.e. operations
and maintenance. The lack of a standard leads to confusion and misunderstandings while exchanging data
during the preparation and execution of utility works. Therefore, it is necessary to develop a sector-wide
digital modelling standard that covers the main concepts and relations for utilities’ operations and
maintenance. Digital modelling standards are also called ‘ontologies’ in information science. An ontology
can be seen as a formal and explicit model of shared conceptualizations about reality. Once adopted and
shared amongst end-users, they represent knowledge in a unified, simplified, and consistent way. If
ontologies are focusing on the conceptualization of knowledge in a specific domain – such as operations
and maintenance of utilities – they are called domain ontologies.
Main goal during the two-year course of the PDEng project has been the development of a domain
ontology that does include those concepts and relations required within operations and maintenance
related activities. The domain ontology was designed while keeping the PDEng assessment criteria in
mind. These criteria elaborate on the design process (construction, realisiblity, functionality, impact and
presentation) and design process (organization and planning, problem analysis and solution,
communication and social skills).
To structure the ontology development process, I applied the engineering cycle of Wieringa (2014),
including the following four phases: (1) problem investigation, (2) treatment design, (3) treatment
validation, and (4) treatment implementation. Since ontology development is necessarily an iterative
process, I used the concept of concurrent design from systems engineering. This meant treatment design
and validation were iterated multiple times, sometimes even leading to a situation in which a next iteration
was already started while the previous iteration one was still running. Ultimately, this iterative and
concurrent fashion resulted in the creation of eight versions of the domain ontology, becoming increasingly
sophisticated every iteration, up till the current version.
The problem within the PDEng project was investigated through the means of a stakeholder analysis,
qualitative case study, a literature review, a document study and input of an expert panel. I found evidence
that for utilities’ operations and maintenance, the main concepts and relations are not covered within a
digital modelling standard. Consequently, organizations create and use their own organizational standards
to model their utility information. Ultimately, this leads to the current widespread of utility modelling
practices in the utility sector, and consequently, confusion and miscommunication when exchanging utility
information. The problem, thereby, justifies the development of the domain ontology.
In order to design the ontology, I first established an ontology design method through an in-depth study
of literature and existing ontology design methods. This method followed the steps of defining competency
PDEng – Final Report X
questions, defining the ontology’s requirements, choosing an ontology development tool and ontology
language, and building the ontology. I executed these steps within the treatment design phase of the
engineering cycle. The main principle I followed during the ontology development process was active end-
user engagement. End-users, in specific the CFM, actively participated in the development of the ontology.
An overview of all meetings conducted during the two-year course of the PDEng can be found in Appendix
II. Through the active involvement of end-users, it was ensured that the real information need for
operations and maintenance on utilities was captured within the ontology.
To validate that the developed ontology did cover the concepts and relations of the domain of operations
and maintenance, I applied four validation techniques: (1) assessment against a source of domain data, (2)
check against class modelling rules, (3) assessment against expert panel input and (4) assessment against
competency questions. As a result of the validation, the final version of the ontology was proven capable
of delivering the utility information as required by the end-user, in specific, the CFM. In addition,
validation proved the ontology to be ready for implementation in (spatial) asset management and GIS
(Geographic Information System) software.
Implementation of the ontology was limited to a proof of concept. I was able to model utility data within
a GIS environment following the ontology’s format. However, this proof of concept only covered a small
piece of the ontology. As a consequence, the ontology has not been evaluated. Therefore, full
implementation of the entire ontology in (spatial) asset management and GIS software is still required. To
ease this process, a brief guideline was written on how to implement the ontology. Moreover, the ontology
and all documentation and data have been made available through an online repository. Via this
repository, potential users have open access to all materials required for implementation and use of the
ontology.
In order to achieve full-scale implementation within the utility sector as a whole, adoption is necessary.
Whereas the purpose of the ontology is to improve the system interoperability of utility data in the sector,
main prerequisite is then that the ontology is also used within the sector. Whereas the PDEng project did
not focus on widespread adoption of the model, is it recommended for future work to make this one of
the main drivers, since once adopted, the domain ontology does provide the following potential:
To the client – the CFM – the domain ontology provides a comprehensive structure alongside
utility data can be stored. The domain ontology is a vast improvement over the existing utility
modelling practice. The domain ontology provides the CFM a more detailed insight in its utilities
and can support visualization, up to the creation of 3D utility models. In addition, the ontology
can facilitate the uptake towards digital maintenance practices.
To the utility sector as a whole, the ontology provides a domain concept overview that is useful to
bridge the various organizational standards within the domain. This may facilitate the transfer
between formats and shape standardized object descriptions improving system interoperability of
utility data.
To software developers the domain ontology provides the support of smarter reasoning within
utility models. This may lead to the digital maintenance practices, and possibly even the creation
of cyber-physical systems.
Altogether, the PDEng project delivers a domain ontology that can be used to model utilities in a uniform
and consistent manner. To this end, the domain ontology does fulfil to the requirements of the CFM and
stakeholders in general, while making the first step towards improved system interoperability utility data.
PDEng – Final Report XI
Product summary
Within the PDEng project, I developed a domain ontology, i.e. a digital modelling standard. The domain
covered by the ontology is the domain of operations and maintenance of utilities. The name given to this
ontology is the ‘Operations and Maintenance conceptual schema’. A conceptual schema is a formal
description of a model that defines a universe of discourse.
The Operations and Maintenance conceptual schema is empirically grounded and provides an overview
of those concepts and relations relevant for operations and maintenance of utilities. Use of the conceptual
schema allows consistent processing and exchange of operations and maintenance related utility data,
thereby improving system interoperability of utility data. The Operations and Maintenance conceptual
schema is developed with support of the developers of the internationally recognized CityGML
UtilityNetwork ADE (Application Domain Extension). As a result, the schema builds upon the CityGML
UtilityNetwork ADE conceptual schema (CityGML, 2016) and adds those concepts and relations relevant
for the domain of operations and maintenance.
The Operations and Maintenance conceptual schema is developed within the system Enterprise Architect.
The schema consist of classes formatted in the Unified Modelling Language (UML). UML is an often
applied language in class modelling due to its graphical notation, and moreover, also applied by the
developers of CityGML UtilityNetwork ADE. In total the whole conceptual schema consist of eight UML
class models. These class models describe the following domains of interest:
[1.] Core and geometry
[2.] Functional characteristics
[3.] Network properties
[4.] Network components
[5.] Component properties
[6.] Operations and maintenance properties
[7.] Performance properties
[8.] Hollow space
The targeted use of the conceptual schema lies in asset management related applications, covering the
domain of operations and maintenance of utilities. The Operations and Maintenance conceptual schema
is applicable to the following utility disciplines: (1) electricity, (2) oil, (3) gas, (4) chemicals, (5) sewage, (6)
water, (7) thermal and (8) telecommunication. The use of the schema has no geographic boundary, but is
based on utility networks of developed countries.
The documentation of the Operations and Maintenance conceptual schema contains three documents:
(1) a data specification, (2) a feature catalogue, and (3) an overview of the class models in UML. The data
specification is the main document, which includes all eight class models together with a detailed
description for each of these and the conceptual schema as a whole. The feature catalogue contains the
complete vocabulary as applied within the ontology, and provides a definition and explanation for all used
terms and names. The overview of UML class models contains solely the class models of the domain
ontology in UML.
PDEng – Final Report XII
In summary, the documentation includes the following documents:
[1.] Operations and Maintenance – Data Specification version 4.1
o OM_Data_Specification.pdf
[2.] Operations and Maintenance – Feature Catalogue 2.1
o OM_Feature_Catalogue.pdf
[3.] Operations and Maintenance – Overview of class models in UML version 4.1
o OM_UML_Overview_pdf
To implement the conceptual schema in (spatial) asset management and GIS (Geographic Information
System) software, it has first been converted to an Extensive Markup Language (XML) schema file. XML
is a widely adopted standard language for exchanging information. The XML schema file allows mapping
of the utility data according to the Operations and Maintenance conceptual schema. However, before the
XML schema can be utilized, it has to be programmed within the software FME, which still has to be
done. Besides programming in FME, all preparation in order to use the domain ontology in (spatial) asset
management and GIS software is finished. An online GitHub repository is created to make all the
documentation and data on the Operations and Maintenance conceptual schema available:
https://github.com/RamonTerHuurne/UtilityNetwork-OperationsAndMaintenance
A list of the main features of Operations and Maintenance conceptual schema are:
Geospatial data model for operations and maintenance on utilities
Based on the CityGML UtilityNetwork ADE conceptual schema
Representation of utilities in both 2D and 3D
Representation of a subnetwork and superordinate network
Representation of supply areas
Representation of detailed utility objects, including:
o Spatial attributes
o Material attributes
o Dimensional attributes
o Shape attributes
o Cost attributes
o Performance attributes
o Impact attributes
o Use attributes
o Surrounding soil attributes
o Status attributes
Integration of the utility domain with CityGML models:
o Buildings (CityGML Building module)
o City furniture (CityGML CityFurniture module)
o Land use (CityGML LandUse module)
o Vegetation (CityGML Vegetation module)
o Water objects (CityGML WaterBody module)
o Transportation (CityGML Transportation module)
o Tunnels (CityGML Tunnel module)
o Bridges (CityGML Bridge module
PDEng – Final Report XIII
Table of contents
1 Introduction 1
1.1 Project description 1
1.2 ReDUCE 2
1.3 PDEng criteria 2
1.4 Role of document 3
1.5 Reading guide 3
2 Project goal, scope, questions and objectives 5
2.1 Project goal and scope 5
2.2 Project questions 6
2.3 Project objectives 7
3 Project design and methodology 8
3.1 Problem investigation 10
3.2 Treatment design 11
3.3 Treatment validation 13
3.4 Treatment implementation 14
4 Problem investigation and analysis 15
4.1 Stakeholder analysis 15
4.2 Phenomena 17
4.3 Problem statement and opportunities 24
5 Treatment design 27
5.1 Methodological steps 27
5.2 Requirements engineering 28
5.3 Ontology design 30
6 Treatment validation 40
6.1 Validation techniques 40
6.2 Development of ontology 42
7 Treatment implementation 44
7.1 Proof of concept 44
7.2 Online repository 46
8 Discussion 47
9 Conclusions and recommendations 49
9.1 Conclusions 49
PDEng – Final Report XIV
9.2 Recommendations 52
10 Bibliography 53
Appendix I – Overview of educational activities 56
Appendix II – Overview of meetings 58
Appendix III – Collected empirical data 66
Appendix IV – Overview of existing standards 71
Appendix V – Expert panel sessions handouts (Dutch) 73
Appendix VI – Possible BSc / MSc research subjects 82
Appendix VII – Publications 83
Section 1 – Project initiation 1
1 Introduction
This chapter introduces the PDEng project, starting with a project description and explanation of the
programme in which the project was executed. Thereafter, a list of PDEng criteria is presented, followed
by an explanation about the role of this document. The chapter concludes with a reading guide for this
PDEng report as a whole.
1.1 Project description
A PDEng (Professional Doctorate in Engineering) programme is a two years post-Master design
programme focussing on the direct needs of the industry. Part of the PDEng programme is a tailor-made
design project in close collaboration with an industry partner. The ‘industry’ partner within the project is
the Campus and Facility Management (CFM) of the University of Twente. An overview of the main project
characteristics are included in Table 1.
Table 1 – Project characteristics
Project title
Project type
Project client
Project period
Modelling utilities by developing a domain ontology
Professional Doctorate in Engineering (PDEng)
Campus and Facility Management University of Twente
October 2016 – January 2019
The CFM has a unique role as the manager of the campus of the University of Twente. She is one of the
few park and terrain managers in the Netherlands with full ownership rights over her utilities.
Consequently, the CFM is responsible for the installation, operations and maintenance of – amongst
others – cables and pipes. Being part of her maintenance duty, the CFM collects and registers all sorts of
information about her utilities. This information covers, for example, location, material and status related
properties. In addition to being an asset manager, the CFM is also client of works on the campus. During
these works she wants to be able to rely on uniform and consistent information about the as-built situation
of their utilities. However, how registration of the as-built information occurs often depends on personal
preferences of utility owners and operators, with as a result, inconsistency in how the information is stored.
In the process of modelling as-built information, construction assets are often digitized by defining their
concepts and relations (El-Diraby & Osman, 2011). Digitized asset information is, consequently, stored in
digital databases and information models. The use of such digital databases and information models in
the Architecture, Engineering and Construction (AEC) industry to manage asset information is a much-
discussed topic in literature (e.g. Bradley et al., 2016; Lu et al., 2015; Pauwels et al., 2016). For example,
within the design of buildings Building Information Modeling (BIM) software in combination with digital
modelling standards such as the Industry Foundation Classes (IFC) is used.
Digital modelling standards are also called an ‘ontology’ in information science. Ontologies describe the
world as seen by a group of people at a certain time according to a school of thought that is based on a set
Section 1 – Project initiation 2
of fundamental propositions or world views (El-Diraby & Osman, 2011). An ontology can be defined as
formal and explicit specifications of shared conceptualizations (Studer, Benjamins, & Fensel, 1998).
Conceptualization refers to the universe of discourse (El-Diraby & Osman, 2011). Shared refers to the
multiple views an ontology should be able to represent. Formal and explicit refers to the fact that the
concepts within the ontology should be described in a clear computer-interpretable format (Olde
Scholtenhuis & Hartmann, 2015).
Ontologies promise a shared and common understanding of a particular domain that can be
communicated across people and computers. In other words, ontologies model ‘realities’. They form the
knowledge base of information systems and allow sharing and reuse of knowledge. Once adopted and
shared amongst end-users, they represent domain knowledge in a unified, simplified, and consistent way.
An ontology offers the CFM – and utility owners, utility operators and contractors in general – a structure
alongside they can store the utility asset information to utilize it for installation, operations and
maintenance purposes. Moreover, in the context of asset management, an ontology can enable digital
maintenance practices, and even the creation of cyber-physical systems. However, an ontology covering
the domain of operations and maintenance of utilities has not yet been developed. Therefore, to support
operations and maintenance of utilities, there is need for an ontology covering those operations and
maintenance related concepts and relations.
1.2 ReDUCE
The PDEng project is part of the ReDUCE (Reduction of Damage to Utilities & Careful Excavation)
programme. ReDUCE is a collaboration between Dutch utility operator KPN (previously Reggefiber) and
the University of Twente. The aim of ReDUCE is to improve design, construction, installation, and
maintenance of (buried) utilities. ReDUCE wants to promote the awareness within the excavation supply-
chain regarding reduction of damages and careful excavation through the development of new methods
and techniques for excavation (ZoARG, 2018).
An important step towards reduction of damages to utilities and careful excavation is to have a uniform
and consistent view about the information of utilities. However, as previously explained, a standard for the
classification of utilities’ concepts and relations in the domain of operations and maintenance is missing.
Consequently, owners and operators of utilities classify and name similar concepts in different ways. This
leads to confusion and misunderstandings at the preparation and execution of utility works. Therefore,
the development of an ontology describing how concepts are classified and which data elements are used
to describe them is necessary. With a good and useful ontology the installation, operations and
maintenance of works in the future can be arranged more precise, making the relevance of the project to
ReDUCE explicit.
1.3 PDEng criteria
The PDEng project is assessed on criteria for its design process and design product. To fulfil to these
criteria, I decided to make them explicit for myself right from the start of the project. These criteria follow
from the ‘Study Guide PDEng programme in Civil Engineering University of Twente’ and are presented
below. I refer back to the PDEng criteria in the conclusion of this report, chapter 9, to assess how I did
fulfil to these.
Section 1 – Project initiation 3
Design product:
Construction – Structuring, inventivity, convincingness.
Realisability – Technical and economic realisability.
Functionality – Satisfaction, ease of use, reusability.
Impact – Societal impact, risk.
Presentation – Completeness, correctness.
Design process:
Organization and planning – Project planning, plan realization, structure and consistency,
conducting meetings.
Problem analysis and solution – Analysis, understanding of impact, creativity, genericity.
Communication and social skills – Reporting, knowledge management, stakeholder motivation,
social skills, independency, reflection and critical attitude.
Structure and attitude – Structure and consistency, reflection and critical attitude, independency.
In addition to the design process and product, educational activities were conducted to fulfil the required
ECTS within a PDEng project. An overview of the educational activities is provided in Appendix I.
1.4 Role of document
Aim of this document is to explain the complete ontology development process I conducted within the
PDEng project. The rationale behind the ontology development process, as well as all the choices I made
within this process are explained. This document, however, does not include the actual domain ontology
itself. It was chosen to create a separate document – a data specification – in which the domain ontology
itself is provided. This data specification contains a detailed description of the domain ontology, and does
not elaborate on the design process. In addition to the data specification, also a feature catalogue is
developed. This document contains the complete vocabulary as applied within the ontology, and provides
a definition and explanation for all used terms and names. Having these separate documents for the
domain ontology eases external publication of the product. The third document provides a sole overview
of the ontology’s class models, for those not requiring all additional explanation and information. Besides
this report, the complete documentation of my PDEng project therefore includes the following three
documents:
[1.] Operations and Maintenance – Data Specification version 4.1
OM_Data_Specification.pdf
[2.] Operations and Maintenance – Feature Catalogue version 2.1
OM_Feature_Catalogue_pdf
[3.] Operations and Maintenance – Overview of class models in UML version 4.1
OM_UML_Overview_pdf
1.5 Reading guide
This document is structured in nine chapters – of which this being the first – divided over three sections.
The first section involves the project initiation. This section exists of three chapters, of which this being
the first, and elaborates on the kick-off of the PDEng project. This chapter involves the introduction on
Section 1 – Project initiation 4
the project. Chapter 2 clarifies the project goal, questions, scope and objectives. The first section ends with
the project methodology in chapter 3.
The second section involves the engineering cycle. The engineering cycle is a rational problem-solving
process proposed by Wieringa (2014). The engineering cycle starts in chapter 4 with the problem
investigation assessing which phenomenon should be improved and why. Chapter 5 includes the treatment
design, describing how the product at hand – the domain ontology – is designed and thereby could treat
the problem. After the design, the engineering cycle continues with the treatment validation in chapter 6.
Validation assesses whether the designed ontology actually treats the problem as it should. Last part of the
engineering cycle is the treatment implementation, discussed in chapter 7.
The third section involves concluding words. Chapter 8 elaborates on various points of discussion.
Chapter 9 concludes the report with conclusions and recommendations.
Section 1 – Project initiation 5
2 Project goal, scope, questions and objectives
This chapter elaborates on the project goal, scope, questions and objectives, as is presented in the following
three sections.
2.1 Project goal and scope
This section elaborates on the project goal and scope of the PDEng project.
2.1.1 Goal
Having uniform and consistent information about utilities is essential to (1) effectively utilize it for
installation, operations and maintenance purposes, (2) carefully excavate, and (3) thereby reducing
damages to utilities. As explained in the project description the lack of a digital modelling standard of
utilities leads to confusion and misunderstandings at the preparation and execution of works. Therefore,
the main goal within the PDEng project is to develop an ontology that offers a structure alongside utility
information can be stored in a uniform and consistent manner. The goal is formulated as follows:
Development of a domain ontology that includes the relevant concepts and relations for the operations
and maintenance of utilities.
2.1.2 Scope
In the context of the PDEng project, utilities are defined as buried utility infrastructure in shallow
subsurface, together with the utility infrastructure at the surface itself. With shallow, a depth of maximum
two meters is meant. Goal of the project is to develop an ontology relevant for the domain of operations
and maintenance. Operations and maintenance refers here to the set of activities performed and strategies
implemented with the goal to preserve and extend the service life of the utilities. The meaning of
operations and maintenance in the context of this report is, thereby, comparable with what is known as
distinct phases in infrastructure asset management.
The utility sector is a multi-disciplinary domain in which various utility disciplines are present. To ensure
proper coverage of the complete (physical) domain of utilities, the developed domain ontology includes
the following utility disciplines:
[1.] Electricity
[2.] Oil
[3.] Gas
[4.] Chemicals
[5.] Sewage
[6.] Water
[7.] Thermal
[8.] Telecommunication
The coverage of these utility disciplines is based on utility networks from developed countries, and
moreover, draws upon the European INSPIRE (Infrastructure for Spatial Information in Europe)
directive (European Commission, s.d.). INSPIRE aims to provide an integrated infrastructure in which
interoperability and exchangeability of the spatial data of the Member States of the European Union is
guaranteed. The decision to keep the utility disciplines relatively abstract was made to allow modelling of
Section 1 – Project initiation 6
every kind of utility network. Further detailing of the utility disciplines, for example towards a high-voltage
electricity network, is thereby possible through the addition of attributes to each discipline.
The domain ontology is developed in close collaboration with the client, the CFM. Consequently, the
domain ontology is specifically developed for a Dutch utility owner. However, the ontology was built in
such a way that it allows global coverage of utility networks, as long as the utilities are within the subsurface
or at the surface. The outreach of the ontology is, therefore, likely to be greater than just the Netherlands.
To create support of the ontology, connection was sought with the developers of the internationally
recognized CityGML UtilityNetwork ADE. CityGML UtilityNetwork ADE is an application domain
extension (ADE) on the CityGML city model standard. The UtilityNetwork ADE adds heterogeneous
modelling of utility networks to the 3D city models of CityGML (CityGML, 2016). Both CityGML and
CityGML UtilityNetwork ADE are part of the Open Geospatial Consortium (OGC). The developed
domain ontology in the PDEng report builds upon the existing conceptual schema of the UtilityNetwork
ADE. The domain ontology adds the concepts and relations relevant to the domain of operations and
maintenance and, therefore, is called the ‘Operations and Maintenance conceptual schema’. By building
on an already known and OGC approved standard, the potential for greater support is there.
Implementation of the ontology is limited to a proof of concept. The proof of concept is performed to
demonstrate how the ontology can be implemented in (spatial) asset management and GIS (Geographic
Information System) software.
2.2 Project questions
In order to achieve the project goal, within the boundaries of the project scope, various questions had to
be asked. In total, four questions were formulated:
[1.] What are the methodological steps required to develop a domain ontology?
A domain ontology is built on multiple principles. Key is to develop an ontology that ensures system
interoperability, the linkage across domains, and logical inferences and proofs (Pauwels, Zhang, & Lee,
2017).
[2.] Which (international) digital modelling standards for utilities exist and how do these relate to the
domain of operations and maintenance?
The utility sector knows digital modelling standards as the Dutch IMKL (Informatiemodel Kabels en
Leidingen) and European INSPIRE. Interesting is to analyze how operations and maintenance – i.e.
infrastructure asset management – is covered within these existing standards. It is analysed which concepts
and relations are covered within their conceptual schemas, possibly providing leads for the development
of the domain ontology.
[3.] How do end-users within the utility sector classify and register their utility data?
Classification and registration of utility data by end-users – such as utility owners, utility operators and
contractors – can greatly differ comparing theory and practice. Whereas the existing digital modelling
standards may not be able to suffice to the end-users’ information need, end-users use their own way of
classifying their concepts and relations. This results in a widespread of utility modelling practices.
Section 1 – Project initiation 7
[4.] Which concepts and relations relevant to the domain of operations and maintenance must the
domain ontology cover?
Coverage of the right concepts and relations by the ontology is crucial for it being widely accepted and
adopted. End-users in the utility sector all have specific information uses and needs concerning the
operations and management of their utilities. These uses and needs are direct input for the concepts and
relations of the domain ontology. Compared to the third question, which analyzes what in practice is
registered, this fourth question analyzes what the information uses and needs of the utility sector are and
how these can be translated to the concepts and relations to be covered by the domain ontology.
2.3 Project objectives
To answer the project questions and progress towards the project goal within the boundary of its scope,
five project objectives have been defined.
[1.] Define the methodological steps and structural requirements for the design of the domain
ontology.
In order to develop a domain ontology that ensures system interoperability, linkage across domains, and
logical inferences and proofs (Pauwels, Zhang, & Lee, 2017) the definition of structural requirement to be
met is a necessity. Within this objective it is (1) decided which ontology design method will be used for
the development of the ontology to and (2) defined what the domain ontology’s structural requirements
are. The first project objective, thereby, aligns to the first project question.
[2.] Define the functional requirements for the design of the domain ontology.
Functional requirements define the required functionality by the domain ontology. Functional
requirements are elicited from (1) the end-users, i.e. the end-user and (2) from existing digital modelling
standards. Since current practices show confusion and misunderstandings at the preparation and execution
of works in the utility sector, the ontology must be based on a shared set of conceptualizations. Engagement
of the end-user is necessary to capture these conceptualizations, and therefore crucial to make the ontology
functionally sound. The second project objective aligns with the second, third and fourth project questions.
[3.] Design the domain ontology.
After making the requirements for the domain ontology explicit in the first two objectives, the third
objective involves the design and modelling of the domain ontology. Design of the domain ontology
includes modelling following the methodological steps and structural requirements of the first objective,
while ensuring the functional requirements from the second objective are covered.
[4.] Validate the domain ontology.
The fourth objective is to validate the domain ontology on its structural and functional soundness. Aim of
the objective is to justify that the designed ontology does meet the requirements from the first and second
objective.
[5.] Implement the domain ontology.
Implementation in the context of this PDEng project is restricted to a proof of concept. Although not
directly related to the project questions nor the project goal, implementation, in a sense, acts as a final
evaluation measure.
Section 1 – Project initiation 8
3 Project design and methodology
An ontology – i.e. digital modelling standard – related to the domain of operations and maintenance for
utilities missing. The ontology to be designed should comprehend those concepts and relations relevant
to the particular domain of operations and maintenance. In order to design the ontology, the engineering
cycle as proposed by Wieringa (2014) was taken as a basis for the overall design process. Within product
engineering - in the context of this project ontology engineering – the engineering cycle is a well-known
approach to follow and was, therefore, considered as most appropriate. The engineering cycle is a rational
problem-solving process consisting of five phases being (1) problem investigation, (2) treatment design, (3)
treatment validation, (4) treatment implementation, and (5) implementation evaluation.
Within the scope of the project the fifth phase is not performed. The domain ontology is only partially
implemented. Evaluation of the implemented domain ontology requires full implementation within
multiple organizations and presumably the sector as a whole. To this end, the remaining four engineering
cycle phases are coupled to the project context and its objectives, resulting in the following structure:
[1.] Problem investigation –which phenomena must be improved and why?
[2.] Treatment design – design of ontology to treat the problem.
Define the methodological steps and structural requirements for the domain ontology.
Define the functional requirements for the domain ontology.
Design the domain ontology.
[3.] Treatment validation – assessment whether the ontology treats the problem.
Validate the domain ontology.
[4.] Treatment implementation – actual treatment of the problem with the ontology.
Implement the domain ontology.
For each of the four phases of the engineering cycle – including the project objectives – the following
sections assess (1) what the goal within the phase is, (2) what methods and strategies are followed to reach
that goal and (3) what the outcome of the phase is. As part of the methods and strategies, it is also explained
how data was collected and analysed.
The engineering cycle does not prescribe a rigid sequence of the phases. However, ontology development
is necessarily an iterative process (Noy & McGuinnes, 2001). Therefore, I made use of systems engineering
and its concept of concurrent design (Blanchard & Fabrycky, 2014). Before implementation, the steps of
treatment design and validation are iterated multiple times, until the project goal is fully met. This also
means that a next iteration may be started when the previous iteration is still running, i.e. concurrent design
of the engineering cycle. This, according to Wieringa (2014), leads to “increasingly sophisticated versions
of the artefact, all aimed at treating the same complex problem”.
Figure 1 shows how the engineering cycle and the principle of concurrent design were applied to the
project context and its objectives.
Section 1 – Project initiation 9
Design
Validation
Requirements
engineering
[2.] Treatment design
[3.] Treatment validation
Sufficient requirements?
Implementation
[1.] Problem investigation
Meets requirements?
Implementation?
Problem statement
[4.] Treatment implementation
Domain ontologyPerform stake-
holder analysis
Assess
expert input
Asess against source of
domain data
Perform literature
review
Methodological
steps
Validation
techniques
Structural and
functional requirements
Perform document
study
Perform document
study
Perform qualitative
case study
Perform qualitative
case study
Activity
Project objective
Phase in engineering cycle
Outcome
Assess against
competency questions
Check against class
modelling rules
Assess expert input
Perform document
study
Define competency
questions
Perform literature
review
Figure 1 – Engineering cycle by Wieringa (2014) applied to the project context and its objectives
Section 2 – The engineering cycle 10
3.1 Problem investigation
Goal:
Identifying, describing, explaining and analysing the problem to be treated, before design of the treatment
and identification of the requirements.
Methods and strategies:
In order to investigate the problem to be treated, multiple methods and strategies are applied, as is
explained below:
First a stakeholder analysis was conducted, in which the stakeholders and their goals were mapped.
Mapping occurs following the onion model of Alexander (2005). This model focuses on the relations of
the stakeholders to the final product, rather than on power and interest. The stakeholder mapping
provides immediate value in identifying and validating stakeholder roles in the requirements elicitation.
Second the phenomena were investigated. Phenomena of interest are (1) existing utility modelling
practices and (2) standardization efforts. It was analysed what the mechanism behind these phenomena
were to occur. I conducted a qualitative case study to gain insights in existing utility modelling practices
(Yin, 1989). In specific, the utility modelling practices within a Dutch utility engineering consultancy firm
were studied. This company registers and processes the data of twelve major utility owners. The twelve
utility owners cover the disciplines of water, gas, electricity, telecommunication and thermal. It was studied
how standards and data protocols were used by these utility owners. The data collection approaches
followed within the case study include observations of works practices and study of asset management
guidelines. To this end, seven information managers were observed whose daily task was to model utility
networks. During the observations, the information managers showed and explained how utility owners
model their utility networks in digital environments. While keeping notes on this, there were dialogues in
which the information managers explained their asset management work routines. Within the case study
also six asset management guidelines were collected, used by utility owners as a guideline to verify the
utility data modelled within the digital environments. More asset management guidelines were either not
available or made accessible due to privacy reasons. The collected data were analysed in a two-staged
process: (1) qualitative coding to abstract and compare the utility modelling practices, and (2) identification
of similarities and differences between the practices. Within the qualitative coding (Saldaña, 2015), the
typology of infrastructure objects of El-Diraby and Osman (2011) was applied. This typology distinguishes
between component level, subsystem level, and system level objects and captures, for example, spatial,
dimensional and material attributes. To justify the development of the domain ontology, I also identified
similarities and differences between the utility modelling practices, following the typology of Gasevic et al.
(2009). This typology essentially describes three elements of an ontology, being: (1) taxonomy and
hierarchy, (2) vocabulary, terms and names, and (3) semantics, the linguistic meaning of the representation.
The taxonomy and hierarchy refer to the hierarchical categorization of the concepts within the ontology.
The vocabulary refers to the set of terms and names that are used in the subject area captured by the
ontology. Semantics refers to the linguistic meaning of the applied terms and names, often prone to
ambiguity and multi-interpretability.
Assessment of existing standardization efforts concerning utility modelling happened through a document
study. The INSPIRE, IMKL, and CityGML UtilityNetwork ADE standard were studied in depth, whilst
also other standards such as the Utility Survey Standard, PAS 128 and ASCE 38-02 were taken into
account. Insights were retrieved in how existing standards are capable of modelling operations and
Section 2 – The engineering cycle 11
maintenance related utility information, providing leads for the development of the domain ontology. The
infrastructure typology from El-Diraby and Osman (2011) was applied to qualitatively analyze covered
infrastructure objects and attributes.
Third the phenomena were reflected upon the stakeholder goals assessing whether the phenomena do
contribute or extract from the goals. This led to the problem statement.
Outcome:
Problem statement based on phenomena investigation and assessment of stakeholder goals.
3.2 Treatment design
Objective 1: Define the methodological steps and the structural requirements for design of the domain
ontology
Goal:
Identifying and describing the methodological steps and structural requirements for the design of the
domain ontology.
Methods and strategies:
In order to elicit the methodological steps and structural requirements for the domain ontology, both a
literature review and document study were conducted, as is explained below:
Through a literature review, both the methodological steps and structural requirements were identified.
Various search engines were used to gain access to appropriate literature: (1) FindUT (search engine of
the University of Twente, (2) Scopus, (3) Google Scholar, and (4) ScienceDirect. I searched for the
following keywords, either individually or combined: ontology, domain ontology, classification, taxonomy,
class-modelling, representation, methodology, requirements and criteria. To this end, I analysed the
following literature on ontology design methods and principles:
Corcho et al. (2002) – Methodologies, tools and languages for building ontologies.
Fernández-López et al. (1999) – Building a chemical ontology using Methondology and the
Ontology Design Environment.
Gasevic et al. (2009) – Model Driven Engineering and Ontology Development.
Jakus et al. (2013) – Concepts, Ontologies, and Knowledge Representation.
Noy and McGuinness (2001) – Ontology Development 101: A Guide to Creating Your First
Ontology.
Pinto et al. (2009) – Ontology Engineering and Evolution in a Distributed World using
DILIGENT.
Sure and Studer (2002) – On-To-Knowledge – Final Version.
Sure et al. (2009) – Ontology Engineering Methodology.
Tempich (2006) – Ontology Engineering and Routing in Distributed Knowledge Management
Applications.
Elicited methodological steps and structural requirements were assessed on the applicability within the
project context. The methodological steps act as a framework for the ontology design, and are discussed
in chapter 5. In addition, I documented the acquired knowledge from the literature review in a separate
Section 2 – The engineering cycle 12
report called Designing and ontology. A proposed framework. This report includes a thorough assessment
on ontology design methods and can be made available on request.
The document study included analysis of existing standardization efforts, performed simultaneously with
the document study of the problem investigation. Whereas in the problem investigation it was assessed
how the domain of operations and maintenance is captured within the standards, within this step it was
analysed how the standards were actually designed. This involves analysis of the class-modelling techniques
applied within the standardization efforts. I documented the acquired knowledge from the document study
in a separate report called Utility modeling. This report provides an overview of utility modelling methods
and standardizations practices and can be made available on request.
The structural requirements are written down following the method proposed by INCOSE (International
Council of Systems Engineering) (2015). They distinguish requirements from five different viewpoints,
being: (1) enterprise, (2) business management, (3) business operations, (4) system and (5) system element.
In accordance with these five viewpoints, the structural requirements of the domain ontology were
established.
Outcome:
Methodological steps and list of structural requirements for the design of the domain ontology.
Objective 2: Define the functional requirements for design of the domain ontology
Goal:
Identifying and describing the functional requirements to design the domain ontology.
Methods and strategies:
In order to elicit the functional requirements, multiple methods and strategies were applied, as is explained
below.
First, I applied the principle of end-user engagement. Crucial for the ontology to be successful is to create
a shared set of conceptualizations about the domain. As such, the principle of end-user engagement was
one followed throughout the whole project, but first and foremost within the functional requirements
elicitation. Through active end-user engagement I was able to elicit tacit knowledge. End-user engagement
was realized through (1) a qualitative case study, (2) an expert panel and (3) regular contact with the client,
the CFM, being an end-user herself. The qualitative case study is the study referred to within the problem
investigation, and allowed insights in the actual information need of the end-user. The expert panel is a
multi-disciplinary panel consisting of the CFM, information managers, the cadastre and standardization
establishments. Two plenary sessions were held with the panel, of which the first session provided most
input concerning requirements engineering. In the first session, the use cases of the ontology were defined.
In addition to the plenary sessions, I had up to ten sessions with experts individually. The regular talks
with the CFM – ranging weekly to monthly – were held to guarantee proper coverage of the domain of
operations and maintenance within the ontology. Being an end-user herself, the CFM provided lots of
useful insights in the actual information need for operations and maintenance on utilities. Moreover, the
regular talks kept the expectations from me as a designer and the CFM as a client aligned.
Second, the stakeholders’ goals as defined earlier on during the problem investigation were translated
towards requirements. Capturing these goals through requirements within the ontology’s design, allows
justification to the stakeholders’ goals.
Section 2 – The engineering cycle 13
Third, competency questions were defined. Competency questions describe questions that the domain
ontology should be able to answer. Two types of competency questions can be distinguished, being (1)
informal and (2) formal (Obrst, Ceusters, Mani, Ray, & Smith, 2008). To elicit the functional requirements
informal competency questions are used. Informal competency questions are not yet expressed in the
formal computer-interpretable ontology language. Therefore, informal questions act as a qualitative
assessment of the requirements in terms of the ontology’s expressiveness (Grüninger & Fox, 1995).
Formal competency questions are expressed in a computer-interpretable format, and are also referred to
as queries. These only can be asked once the ontology is fully implemented. However, this is not part of
the PDEng project, and therefore, not executed.
Similar to the structural requirements, the functional requirements are written down following the method
proposed by INCOSE (2015).
Outcome:
List of functional requirements for the design of the domain ontology.
Objective 3: Design the domain ontology
Goal:
Design of the domain ontology.
Methods and strategies:
To ensure the domain ontology’s structural and functional soundness, the methodological steps and
requirements as defined within the first and second objective are taken into account during the design. In
addition, a mix between a bottom-up and top-down approach is applied to model the ontology’s concepts
and relations. The top-down approach is an often applied approach used within ontology development,
and well-known for its delivery of high-quality ontologies (Sure, Staab, & Studer, 2009). Within the top-
down approach the more generic and abstract concepts are defined first and related to each other. This
assures coverage of the broader context of the ontology. However, to assure that the ontology really
delivers the information as required by the end-user, the bottom-up approach is applied as well. Through
the involvement of end-users within the ontology development process, as explained in the previous
section, is it ensured that the real information need for operations and maintenance on utilities is captured.
Outcome:
Designed domain ontology. The design of the ontology has been redesigned multiple times after
validation, characterizing the iterative design process followed.
3.3 Treatment validation
Objective 4: Validate the domain ontology
Goal:
Justifying that the developed domain ontology does contribute to the stakeholder goals while meeting the
structural and functional requirements.
Methods and strategies:
In order to validate the domain ontology on its structural and functional requirements, four validation
techniques were applied, as is explained below:
Section 2 – The engineering cycle 14
First, I compared the ontology against sources of domain data. The sources of domain data were retrieved
during the qualitative case study as well as the document study of existing digital modelling standards. I
compared the domain ontology with the elicited utility modelling practices and digital modelling standards
by following the ontology typology of Gasevic et al. (2009). To this end, I could validate whether the
ontology was capable of reproducing the utility concepts and relations as seen within these data sources.
Second, I checked the ontology on its class modelling rules. In the project context, the domain ontology
was modelled within the Unified Modelling Language (UML). By applying UML rules, I could validate
the domain ontology on its structural soundness from a modelling perspective.
Third, I used input from an expert panel to validate the ontology. The expert panel sessions held for
validation, both the plenary and individual ones, are the same sessions as were held to define the
requirements during the design phase. This illustrates the concurrent design process followed during the
project. To this end, two plenary expert panel sessions were held of which the second provided most input
concerning validation. In this session, it was validated whether the information need belonging to the use
cases and competency questions could actually be modelled by the domain ontology. The individual
sessions with the experts were in general more in-depth, discussing, for example, the use of a single concept
within the ontology.
Fourth, I assessed the ontology against the (informal) competency questions defined during the design
phase. Even though the ontology was already evaluated within the second plenary expert panel session on
its use cases and competency questions, not all of them were validated because of a limited amount of
time. Therefore, I once more validated the ontology against the competency questions.
Outcome:
Validated domain ontology.
3.4 Treatment implementation
Objective 5: Implement the domain ontology
Goal:
Partial implementation of the domain ontology in asset management related GIS software.
Methods and strategies:
The designed domain ontology cannot directly be implemented in (spatial) asset management and GIS
related software. UML is a graphically oriented ontology language, and consequently, not directly
interpretable by (spatial) asset management and GIS software. Therefore, I converted the domain ontology
in UML to an XML (Extensive Markup Language) schema file. XML is a widely adopted standard
language for exchanging information. Through the XML schema file, I was able to write existing utility
data towards a GML (Geography Markup Language) file, which is a format widely supported by GIS
related software. To this end, the domain ontology can be applied in GIS software.
Outcome:
Partially implemented domain ontology.
Section 2 – The engineering cycle 15
4 Problem investigation and analysis
This chapter elaborates on phase one of the engineering cycle, the problem investigation. To investigate
the problem, first a stakeholder analysis is conducted. Second, the phenomena of interest are investigated,
including an investigation of existing utility modelling practices and standardization efforts. The chapter
concludes with the problem statement and opportunities.
4.1 Stakeholder analysis
This section includes the stakeholder analysis. First, the stakeholders of interest to the PDEng project are
identified. Second, the identified stakeholders are mapped according to the onion model of Alexander
(2005).
4.1.1 Stakeholder identification
A stakeholder is a person, group of persons or an institution affected by the to be treated problem. They
are the source of goals and constraints within the project context, which in their turn are the source for the
requirements (Wieringa, 2014). Therefore, identification of stakeholders is regarded highly relevant to
ensure all requirements are taken into account. The stakeholders of interest to this PDEng project are
described within Table 2, together with their goals and constraints.
Table 2 – Stakeholders within the project
# Stakeholder Description Goal(s) Constraint(s)
1 Designer PDEng trainee of the
University of Twente,
responsible for development
of the domain ontology.
Acquiring PDEng certificate. Project duration of two years.
2 Client (CFM) Industry partner within the
PDEng project. Park and
terrain manager of the
campus of the University of
Twente with full ownership
over her utilities.
Real-time insight in the as-built
information of their utilities.
Easy utilization of ontology.
Improving effectiveness and
efficiency of operations and
maintenance related activities.
As-built information currently
CAD based.
Limited extensiveness of as-
built information.
3 Utility owners Organizations with
ownership over utilities.
This includes public domain
authorities as well.
Real-time insight in the as-built
information of their utilities.
Easy utilization of ontology.
Organizational (utility
modelling) routines.
Privacy of utilities information.
4 Utility
operators
Organizations responsible
for operations and
maintenance related
activities on utilities.
Improving effectiveness and
efficiency of operations and
maintenance related activities.
Easy utilization of ontology.
Organizational (utility
modelling) routines.
Privacy of utilities information.
5 Excavation
contractors
Organizations responsible
for excavation works.
Reducing excavation damages to
utilities.
Easy utilization of ontology.
-
6 Authorities Organizations concerned
with class-modelling of
utilities (such as the
cadastre).
Improving system
interoperability of utility data.
Conformance with existing
class-modelling standards.
Section 2 – The engineering cycle 16
7 Software
developers
Organizations developing
the (spatial) asset
management related (GIS)
software.
Development of sector-wide
(spatial) asset management
related (GIS) software.
Application within existing
(spatial) asset management
related (GIS) software.
8 Utility users End-user of utilities. Reliable use of the services by
utilities.
-
9 Investors Organizations funding the
project.
Project accomplishment within
budget.
Project budget.
4.1.2 Stakeholder mapping
Mapping of the stakeholders follows the onion model of Alexander (2005). This model focuses on the
stakeholders’ relationship to the final product rather than on the stakeholders’ power and interest. The
onion model builds a stakeholder taxonomy and provides insights in the relevance of the stakeholders
towards the domain ontology. Mapping of the stakeholders according to the onion model is shown in
Figure 2.
Figure 2 shows the onion model
including the stakeholders within
the project. The inner circle
comprises ‘The Product’, which is
the product of design, in this project
the domain ontology for utilities.
The next circle, ‘The System’ circle,
involves the stakeholders that do
have an influence on the design of
this domain ontology and / or
eventually will be the end-user of the
product. Following the earlier
explained principle of end-user
engagement, this circle includes the
client, utility operators and utility
owners, in addition to the designer.
Creating a shared set of
conceptualizations amongst these
stakeholders is of great importance.
The next circle, ‘The Containing
System’, involves beneficiaries and
others, who are not directly involved in the design of the product. This circle includes the excavation
contractors, software developers, investors and authorities. Albeit they do have an influence on the design,
their influence is indirect, whilst they don’t have a direct say in the design choices to be made. In the last
circle, ‘The Wider Environment’, all other stakeholders that are recognized within the project are
included. This circle includes the utility user, which might benefit from less down-time of utilities, due to
more effective and efficient operations and maintenance of utilities.
Figure 2 – Stakeholder mapping following Alexander’s (2005) onion model
Section 2 – The engineering cycle 17
4.2 Phenomena
This second elaborates on the phenomena of interest to the PDEng project. To this end, first utility
modelling practices are explored. Second, international and national utility modelling standardization
efforts are analysed.
4.2.1 Utility modelling practices
A notable shift within the utility sector – following the general trend within the construction industry – is
the increasing attention to life cycle management (Bradley, Li, Lark, & Dunn, 2016). Therefore, utility
owners and operators increasingly want to have an uniform and consistent set of digital information about
their utility objects. Uniform and consistent information about utility objects is essential to (1) effectively
utilize it for installation, operations and maintenance purposes, (2) carefully excavate and (3) thereby
reducing damages to utilities. Since operations and maintenance of the utility objects requires more
attribute data than only the location, dimensions and materials – nowadays mainly registered – other
attribute types become relevant, such as shape, costs, performance, surrounding, dependency,
redundancy, state of operation and impact (El-Diraby & Osman, 2011).
The shift to life cycle management ideally assumes that domain knowledge is captured in a uniform and
consistent set of concepts and relations. However, in practice this overview is missing whereas
organizations (utility owners, utility operators, contractors) describe concepts/objects using different (1)
attributes, formats, relationships and (2) semantic terms. Altogether, these may result in misunderstanding
(hypernyms (different subordinates), homonyms (pronounced the same, different meaning), synonyms
(different words, equal meaning)) when utility data is communicated.
To explore how utilities are modelled in practice, a qualitative case study was conducted. As explained in
chapter 3, I studied the utility modelling practices within a Dutch utility engineering company. Findings of
the case show that the elicited digital models include a comprehensive set of utility asset data. To this end,
around twenty to forty infrastructure objects within a digital model were found, depending on the type of
utility discipline. For the telecommunication discipline, for example, around twenty utility infrastructure
objects were found, whereas the gas discipline covered up to even forty objects. Following the typology of
utility infrastructure objects of El-Diraby and Osman (2011) – which distinguishes between component
level, subsystem level and system level top-level objects – most of the utility infrastructure objects found
belong to the component level. Below the typology of utility infrastructure objects is briefly explained:
Component level – Lowest level of utility infrastructure objects. Objects that in essence cannot be
decomposed any further.
Subsystem level – Intermediate level of utility infrastructure objects. Objects that can be
decomposed, but also are composed of objects themselves.
System level – Top level of utility infrastructure objects. Objects that are on top of the hierarchy.
To illustrate how this typology applies to practice, Table 3 provides the found system, sub-system and
component level objects for the telecom domain. The objects as presented are already analysed, meaning
that similar objects from multiple digital models (and multiple utility owners) are grouped together.
Section 2 – The engineering cycle 18
Table 3 – Snap of empirical analysed utility data from the telecom domain
System level object Subsystem level object Component level object
Telecom distribution system Fiber line system
Coax line system
Cable
Connection
Station/center
Patch box
Handhole
Point of Presence
Tube
Transition
Cable
Connection
Station/center
Sleeve
Amplifier
Splitter
Moreover, I found extensive lists of attribute information for the infrastructure objects at hand. Within
elicited digital models from the gas discipline, for example, up to eighteen attributes for the component
level object ‘pipe’ were found as illustrated in Table 4. A comparable number of attributes were found for
other infrastructure objects. The attributes as presented are already analysed, meaning that similar
attributes from multiple digital models (and multiple utility owners) are grouped together.
Table 4 – Snap of empirical analysed utility data from the gas domain
System level
object
Subsystem level
object
Component
level objects
Attributes
Gas distribution
system
Gas low
pressure
network
Pipe Location, sub net, date of installation, pressure level,
material, implementation, coating, connection type, cathodic
protection, tensile resistance, length, date of rehabilitation,
accuracy, owner, operator, financial characteristic,
controlled drilling
To classify the infrastructure attributes, I used the typology of El-Diraby and Osman (2011) which
distinguishes the following twelve attribute types: (1) dimensional attributes, (2) spatial attributes, (3)
material attributes, (4) shape attributes, (5) cost attributes, (6) performance attributes, (7) surrounding soil
attributes, (8) dependency attributes, (9 redundancy attributes, (10) state of operation attributes, (11)
impact attributes and (12) use attributes. The attribute types are briefly explained below:
[1.] Dimensional – Describing measurable dimensions of a component, for example, the interior width
or exterior diameter.
[2.] Spatial – Describing a location-related attribute of the component, for example, the x- or y-
coordinate.
[3.] Material – Describing one or more attributes of the component material, for example, the type or
ductility.
[4.] Shape – Describing the shape of a component, for example, circular or rectangular.
[5.] Cost – Describing a monetary value associated with the component, for example, capital or
maintenance costs.
Section 2 – The engineering cycle 19
[6.] Performance – Describing the performance of a component, for example, the engineering or
serviceability performance.
[7.] Surrounding soil – Describing the soil-related properties surrounding the component, for example,
the type and permeability.
[8.] Dependency – Describing the interdependencies between components, for example, physical or
geographical.
[9.] Redundancy – Describing the need of a particular component within the greater network in case
of a breakdown of that component.
[10.] State of operation – Describing the state of operation of a component at a particular state in time,
for example, functional or under construction.
[11.] Impact – Describing the impact of a component, for example, environmental or economic.
[12.] Use – Describing use related properties of a component, for example, the function or use of a
component.
Within the elicited digital models mostly use attributes (12) were found. Use attributes for example cover
information about its function or use, but also about the date of installation or related party. Attributes
concerning material (3), performance (6) and state of operation (10) follow with a fairly comparable
distribution. The number of
attributes found per attribute type is
presented in Figure 3. Reflected
upon the domain of operations and
maintenance the lack of cost (5),
impact (11), shape (4) and
surrounding soil (7) attributes is
important to acknowledge. Even
though these attribute types
provide necessary information to
effectively utilize operations and
maintenance related activities, they
are (almost) not represented in the elicited realities. For a complete overview of the found infrastructure
object and attributes within the elicited digital models I refer to Appendix III.
A comparison between the elicited digital models showed that in general similar objects and attributes
were captured by the utility owners. The digital models include comparable system, subsystem and
component level objects. This finding also applies to the type of attributes captured. However, a more
detailed comparison between the digital models shows subtle differences. For each of the three elements
of the typology of Gasevic et al. (2009), one difference is presented in detail, while some other examples
are briefly noted. Since the data has been collected in Dutch, for both differences in vocabulary and
semantics, the original Dutch term is given in addition to the English translation to secure its intended
meaning. I elaborate this below:
[1.] Taxonomy and hierarchy
An example of a taxonomical difference is visible within the telecom discipline. Within the discipline,
subsystem level and system level objects to model a telecom network are applied in the following two ways.
First, two digital models use subsystem level objects to model both coaxial and fiber networks as individual
subsystems. The main telecom network, including both the coaxial and telecom network, is modelled on
Figure 3 – Distribution of attribute types within elicited realities
20
9 9 8
5 5
2 2 1 1 0 0
Section 2 – The engineering cycle 20
top of the taxonomy at the system level. Second, another digital model does not distinguish subsystems
level objects for the coaxial and fiber network. Here, the particular network type is specified as an attribute
of the subsystem object, i.e. the main network.
Another taxonomical difference, for example, is visible within the modelling of the object 'station' (station)
when comparing three digital models within the gas discipline. Where two digital models model a station
as one single component level object and define its type (internal, external) at the attribute level, one other
digital model defines the type of station already at the component level object.
[2.] Vocabulary, terms and names
An example of a difference in vocabulary is the similar use of the words 'meting' (measurement) and
'nauwkeurigheid' (accuracy) when comparing two digital models within the electricity discipline. Both
digital models use these words to describe the type of measurement of coordinates (either analogous or
digital) in order to estimate how accurate the coordinates of component level objects are.
Another notable difference in use of vocabulary, for example, is the similar use of the words 'ligging' (area)
and 'locatie' (location) to describe the x and y coordinates of a component level object.
[3.] Semantics
A semantic difference in the elicited digital models is the ambiguous meaning of the word 'uitvoering'
(implementation). The word is used to describe various characteristics of objects on mainly the component
level. Six digital models across the telecom, gas and electricity discipline, show the use of the word to
describe divergent characteristics such as diameter, manufacturer, material type, and installation type. This
difference was even found in individual digital models of the gas discipline. In addition, in three digital
models of the electricity discipline, the word is used to describe multiple characteristics at once, including
manufacturer, material and diameter.
Other examples of semantic differences, for example, include the ambiguous meaning of both the words
'sort' (soort) and 'type' (type). Similar to 'uitvoering' (implementation), these words are used to describe
various characteristics of objects, being mainly the installation type and material type.
Main reason of the existence of these differences is the existence of ‘organizational’ standards. The findings
of the case study showed that the modelled digital models were represented in: (1) international, (2)
national and (3) organizational standards. Both through observations of the work practices and study of
asset management guidelines, it was revealed that, for example, each of the utility owners prescribed their
own asset data registration standard, based on their own work practices. According to one of the
information managers spoken with during the case study, these ‘organizational’ standards were used more
frequently that the international and national standards. The illustrated examples also follow from these
organizational standards. In conclusion, one can argue that practice shows a widespread of ‘digital realities’.
Based on a set of fundamental propositions or world views, utility owners develop their own perceived
view on the reality, and model utility networks accordingly.
The findings from the qualitative case study were also taken as a basis for the conference paper
‘Digitization for integration: fragmented realities in the utility sector’ (ter Huurne & olde Scholtenhuis,
2018). I presented this paper at the 34th
ARCOM conference in Belfast, September 2018. The premise
tackled in the paper is the general belief that digitization leads to integration, by arguing the existence of
fragmented realities in the utility sector. The paper is enclosed to this report and can be found in Appendix
VII.
Section 2 – The engineering cycle 21
4.2.2 Standardization efforts
The limited use of international and national standards by utility owners raises the question whether
existing international and international standards are capable of representing the comprehensiveness of
utility information as required by their users. This question was answered through a document study
covering multiple existing utility modelling standards. A complete overview of the standards examined is
provided is Appendix IV.
Before a standard, also utility modelling standards, emerges, it goes through the process of standardization.
The aim of standardization, in general, is to capture realities and construct uniformities across cultures,
time and geography on the basis of agreed-upon rules (Timmermans & Epstein, 2010). These agreed-
upon rules are thereby captured in standards (Bowker & Star, 1999; Timmermans & Epstein, 2010). The
types of standards relevant to the project context are digital modelling standards. Standardized data formats
and structures help to achieve integration under the condition that these are accepted and represent end-
users’ shared conceptualizations.
Drawing upon Howard and Björk (2008) standardization is critical to facilitate communication between
stakeholders in a fragmented industry – like the AEC industry and in specific the utility sector. Where
standards for representing construction data and geospatial data, such as the Industry Foundation Classes
(IFC) and CityGML are gradually accepted and used as the predominant way to exchange data between
engineering software, common data formats for other types of infrastructure such as utilities are less
developed (Bradley, Li, Lark, & Dunn, 2016).
Well-known digital modelling standards for utilizes that do exist – from an international and national
perspective – are, for example, INSPIRE (European Union) and IMKL (the Netherlands). By reflecting
these and other standards upon the project context – i.e. operations and maintenance of utilities – and the
domain data collected within the case study, insights were retrieved in how capable these standards are in
modelling operations and maintenance related utility information. Standards analysed include INSPIRE,
IMKL, CityGML UtilityNetwork ADE, Utility Survey Standard, PAS 128 and ASCE 38-02. More
standards were analysed, but due to brevity reasons not included in this chapter. For more information on
these standards, I refer to Appendix IV or the report Utility modeling. The latter can be made available
on request.
A systematic comparison between the analysed digital modelling standards is provided in Table 5 (page
22). I analysed the six standards on six areas:
[1.] Focus – Focus (purpose) of the digital modelling standard.
[2.] Scope – Scope of the digital modelling standard.
[3.] Users – Targeted end-users of the digital modelling standard.
[4.] Data – Coverage of operations and maintenance related utility data within digital modelling
standard.
[5.] Origin – Origin of digital modelling standard.
[6.] Reference – Reference to digital modelling standard.
Section 2 – The engineering cycle 22
Table 5 – Comparison between and overview of utility digital modelling standards
INS
PIR
E
IMK
L
Cit
yGM
L
Utility
Netw
ork
AD
E
Utility
Su
rvey
Sta
nd
ard
PA
S 1
28
AS
CE
38-0
2
1 Focus Create an integrated
European infrastructure to
provide cross-border
interoperability and
exchangeability of spatial
data.
Improve interoperability
and exchangeability of
utility data to prevent
excavation damages.
Develop a homogenized
3D network model for
multi-utility simulation
including the relevant
thematic attribution.
Create a common
understanding of the
acquisition and production
of utility data within utility
surveying through
guidelines.
Set out clear and
unambiguous provisions to
those engaged in
the detection, verification
and location of active,
abandoned, redundant or
unknown utilities.
Develop a consensus
standard for classifying the
quality of utility location
and attribute information
that is placed on plans,
allowing users to
developed strategies to
reduce risk by improving
the reliability of utility
information in a defined
manner.
2 Scope Applies to the utility
disciplines of electricity,
oil, gas, chemicals, sewage,
telecommunication and
water. Provides basic
information on a wide
range of administrative
and social services of
public interest.
Applies to the utility
disciplines of interest to
the Netherlands and
Belgium. Specifically
aimed at preventing
damages during excavation
activities.
No predefined utility
disciplines. Disciplines are
modelled through
properties set. Specifically
modelled as an extension
to CityGML for failure
simulation of utility
networks.
Applies to the utility
disciplines of electricity,
gas, sewage,
telecommunication, water
and drainage. Specifically
developed for utility
surveying.
Applies to the utility
disciplines of drainage,
gas, water, electricity and
telecommunication.
Specifies requirements for
detection, verification and
location of existing and
new underground utilities.
Applies to existing utilities
of the disciplines
electricity, gas, oil, crude,
water, thermal,
telecommunication,
sewage or any similar
commodity. Specifies
guidelines on how utility
information can be
obtained, what
technologies are available
to obtain the information,
and how that information
can be conveyed to the
users.
3 Users Targeted end-users are
map requesters, utility
owners, and public
domain authorities.
Targeted end-users are
map requesters, utility
owners, and public
domain authorities.
Targeted end-users are
(mainly) utility owners and
public domain authorities.
Targeted end-users are all
involved in utility
surveying (can be a map
Targeted end-users are
practitioners (surveyor,
geophysicist or utility
engineer) but also clients
Targeted end-users are
contractors, project
owners, utility owners and
engineers. Therefore,
Section 2 – The engineering cycle 23
requestor, utility owner or
public domain authority).
(such as engineers,
contractors, project
owners and utility owners).
Therefore, users involve
both map requestors and
utility owners.
users involve both map
requestors and utility
owners.
4 Data Application schema
(UML) that contains
system, subsystem and
component level objects.
Material, performance and
use attributes are partially
covered, whereas cost,
surrounding soil, and
impact attributes are
lacking.
Application schema
(UML) that contains
system, subsystem and
component level objects.
Material, performance and
use attributes are partially
covered, whereas cost,
surrounding soil, and
impact attributes are
lacking.
Application schema
(UML) that contains
system, subsystem and
component level objects.
Material, performance and
use attributes are partially
covered, whereas cost
surrounding soil, and
impact attributes are
lacking.
Guidelines and templates
prescribing which utility
data to be collected and
how. No taxonomy or
modelling of relations
between objects. Standard
is not object-based but
does describe subsystem
and component level
objects. Material and use
attributes are partially
covered, whereas cost,
performance, surrounding
soil, dependency,
redundancy, and impact
attributes are lacking.
Requirements for the
detection, verification and
location of existing and
new activities. Standard
does not provide a
taxonomy nor vocabulary.
Standard is not object-
based but does describe
component level objects.
Material, performance and
use attributes are partially
covered, whereas shape,
cost, surrounding soil,
dependency, redundancy,
state of operation and
impact attributes are
lacking.
Guidelines on collection
and depiction of utility
information on four
quality levels. Each level
prescribes what utility
information must be
collected and how this
must be conveyed to the
users. The guidelines do
not provide a taxonomy
nor vocabulary. Standard
is not object-based but
describes component level
objects. Attributes are en-
closed in (CAD) drawings
through labelling, symbols,
colour coding, line types
or text. Dimensional and
use attributes are partially
covered, whereas material,
shape, cost, performance,
surrounding soil,
dependency, redundancy
and impact attributes are
lacking.
5 Origin European Union The Netherlands /
Belgium
Global Singapore United Kingdom United States
6 Reference D2.8.III.6 INSPIRE Data
Specification on Utility
and Government Services
– Technical Guidelines
(INSPIRE, 2013)
IMKL 2015 –
Dataspecificatie
Utiliteitsnetten version 1.2
(Geonovum, 2017)
CityGML UtilityNetwork
ADE version 0.9.2
(CityGML, 2016)
Standard and
Specifications for Utility
Survey in Singapore
version 1.0 (Singapore
Land Authority, 2017)
PAS 128:2014
Specification for utility
detection, verification and
location (British Standards
Institution, 2014)
ASCE 38-02 Standard
Guideline for the
Collection and Depiction
of Existing Subsurface
Utility Data (American
Society of Civil Engineers,
2002)
Section 2 – The engineering cycle 24
4.3 Problem statement and opportunities
This section elaborates on the problem statement and opportunities, based on the aforementioned
stakeholder analysis and phenomena investigation. First, the problem statement is defined. Second, the
opportunities are explored, whilst also keeping possible challenges in mind.
4.3.1 Problem statement
The problem investigation and analysis, in essence, showed the following problems in the registration and
processing of utility information:
[1.] Plethora of utility modelling practices to model domain knowledge, all describing utility
concepts/object using different (1) attributes, formats, relations and (2) semantic terms.
[2.] Lack of coverage of operations and maintenance required utility information in existing standards.
The utility modelling practices of twelve utility owners observed in the qualitative case study showed a
plethora of ways to register and process utility information. This this end, I found differences in utility
information registration on (1) taxonomy and hierarchy, (2) vocabulary, terms and names, and (3)
semantics. Even though the differences are sometimes very subtle, they prevent proper integration and
system interoperability of utility data. Moreover, these differences can result in the aforementioned
confusion and miscommunication during the preparation and execution of utility works.
Concerning existing standardization efforts, analysis of six standards showed the absence of required
operations and maintenance utility information. Assessment of the standards on the infrastructure typology
of El-Diraby and Osman (2011) shows that none of the standards is capable of describing utility networks
and their domain knowledge in the consistent and complete manner required for operations and
maintenance – i.e. for life cycle management. Attributes related to material, performance and use are often
only partially covered, whereas attributes related to costs, surrounding soil, impact, and even state of
operations are sometimes completely lacking. Presumably, this is because none of the analysed standards
were explicitly developed for the purpose of operations and maintenance on utilities.
The absence of a digital modelling standard developed specifically for operations and maintenance
ultimately is the main reason to the plethora of ways utility information is registered and processed as seen
in practice. More specific, the findings within the qualitative case study showed the presence of
‘organizational’ standards being preferred by organizations over the existing international and national
standards. To summarize, I define the problem statement as follows:
For utilities’ operations and maintenance, the main concepts and relations are not covered within an
accepted and shared digital modelling standard. As a consequence, organizations create and use their own
organizational standards to model their comprehensive set of domain knowledge. Ultimately, this leads to
the presence of fragmented digital models in the utility sector, making the preparation and execution of
utility works prone to confusion and miscommunication of utility information.
4.3.2 Opportunities and challenges
The problem investigation and analysis – and, therefore, the problem statement – legitimates the project
goal as defined in section 2.1.1:
Development of a domain ontology that includes the relevant concepts and relations for the operations
and maintenance of utilities.
Section 2 – The engineering cycle 25
The lack of an operations and maintenance focused digital modelling standard for utilities provides room
for the development of one – i.e. the goal of this PDEng project. The contribution that the domain
ontology can deliver to the utility sector is:
[1.] to provide a domain concept overview that is useful to bridge the various organizational standards
within the domain. This may facilitate the transfer between formats and shape standardized object
descriptions;
[2.] the integration of utility domain information with other domains (such as BIM and GIS
information);
[3.] the support of smarter reasoning within utility models, and accordingly, digital maintenance
practices.
The place of a domain ontology within the existing utility sector is illustrated in
Figure 4. The figure shows the as-is situation on the left, and the future situation including the domain
ontology on the right. The figure starts at the bottom with the organizations involved within the utility
sector. Each of these organizations has a certain information use and need. Throughout the years these
information uses and needs have been translated into specific digital modelling standards, who can be of
an international, national or organizational kind. These distinctive standards are taken as a basis for utility
modelling, resulting in a certain ‘modelled reality’ applied in digital environments, such as asset
management software.
Figure 4 – The place of a domain ontology within the industry (left as-is, right with a domain ontology included)
The inclusion of the domain ontology, as illustrated on the right side of the figure, provides a consistent
and uniform layer between the existing standards and modelled realities. To this end, I propose the
domain ontology not as a replacement for existing standards, but either as a fundament on which these
standards can be based. This means organizations can still use their own organizational standards.
Modelled realities, however, now are all based upon the domain ontology, and therefore, interoperable
with one another.
One great challenge to be considered is the adoption of the domain ontology once developed. Drawing
upon Geels (2004), one must take the existing socio-technical system into account in which the domain
ontology will be introduced as a new innovation. A socio-technical system is a cluster of elements including
Information use / need
Standards
(International, national, organizational)
Domain ontology
Organization
(map requestor, network authority, public
domain authority)
Information use / need
Standards
(International, national, organizational)
Organization
(map requestor, network authority, public
domain authority)
Modelled
reality
Modelled
reality
Modelled
reality
Modelled
reality
Tooling Tooling Tooling Tooling
Modelled
reality
Modelled
reality
Modelled
reality
Modelled
reality
Tooling Tooling Tooling Tooling
used in used inused in used in used in used in used in used in
has has
translated intotranslated into
translated into
forms
forms
tran
slat
ed
into
Section 2 – The engineering cycle 26
technology, regulations, user practices, markets, cultural meaning, infrastructure, maintenance networks
and supply networks. Within a socio-technical system three elements play a role: (1) the socio-technical
system, (2) the social groups, and (3) the rules (also understood as regimes). Therefore, a socio-technical
system is a stable regime carried by the three elements mentioned.
Introducing a new innovation – in this case the domain ontology – in an existing socio-technical system
may induce resistance and hesitation through a certain level of stabilization of the established regime.
Geels (2004) mentions that niche innovations, which I consider the domain ontology, are rather difficult
to introduce in such existing regimes. Therefore, for niche innovations, new socio-technical regimes are
created, acting as ‘incubators’ for the innovation to nurture. The whole process of becoming the new socio-
technical regime, and thereby, the shared and accepted standard in the utility sector takes time and
requires momentum before it breaks through. Such momentum may be realized through the
establishment of a higher focus on collaboration, for example, in the utility sector. In addition, the greater
attention to lifecycle management may also be beneficiary for the adoption of the domain ontology as the
new regime.
The next chapter continues with the design of the ontology, phase two within the engineering cycle. The
chapter elaborates on:
the methodological steps for the design of the domain ontology.
the structural requirements for the design of the domain ontology.
the functional requirements for the design of the domain ontology.
the actual design steps taken to design the domain ontology.
Section 2 – The engineering cycle 27
5 Treatment design
This chapter elaborates on phase two of the engineering cycle, the treatment design. Within this chapter,
the project objectives 1, 2 and 3 are covered. To this end, this chapter starts with defining the
methodological steps of ontology development. Thereafter, both structural and functional requirements
are defined. Based on the methodological steps and requirements, this chapter concludes with the
ontology design.
5.1 Methodological steps
Based on a literature review and assessment of ontology development methodologies, I created a list of
methodological steps required for successful development of a domain ontology. Within this PDEng
project, I defined the following seven ontology development steps, still adhering to the engineering cycle
of Wieringa (2014):
[1.] Define competency questions
Discussed in section 5.2.1 ‘Competency questions’.
o Definition of informal competency questions.
[2.] Define requirements
Discussed in section 5.2.2 ‘List of requirements’.
o Definition of structural requirements.
o Definition of functional requirements.
[3.] Choose ontology development tool
Discussed in section 5.3.1 ‘Ontology development tool’.
[4.] Choose ontology language
Discussed in section 5.3.2 ‘Ontology language’.
[5.] Build ontology
Discussed in section 5.3.3 ‘Build ontology’.
o Definition of informal ontology description.
o Building of the formal domain ontology.
[6.] Validate and refine ontology
Discussed in chapter 6 ‘Treatment Validation’.
[7.] Implement ontology
Discussed in chapter 7 ‘Treatment Implementation’.
This chapter elaborates on the first five ontology development steps, with as a result the design of the
ontology. Both the definition of the competency questions (step 1) and requirements (step 2) are discussed
in the next section. The chapter concludes with the design of the ontology. Within this section, the
ontology development tool (step 3) and language (step 4) are chosen, followed by the build of the ontology
(step 5).
Section 2 – The engineering cycle 28
5.2 Requirements engineering
This section presents the list of requirements for design of the domain ontology. In order to elicit the
requirement, multiple methods and strategies were applied: (1) defining competency questions, (2)
translating stakeholder goals towards requirements, and (3) applying the principle of end-user engagement.
5.2.1 Competency questions
The definition of competency questions is the first step in the actual development process of the domain
ontology. As explained before in section 3.2, competency questions are questions that the ontology should
be able to answer (Obrst, Ceusters, Mani, Ray, & Smith, 2008). In other words, the ontology should be
able to provide the information required to answer the asked for questions. Therefore, the competency
questions directly relate to the main goal of the client in the PDEng project, the CFM, which is having
real-time insight in the as-built information of their utilities. To ensure the ontology could live up to this
goal, the informal competency questions were together with the CFM defined in multiple meetings.
Definition of the (informal) competency questions is closely related to the use cases of the ontology.
Therefore, during the meetings with the CFM, first the use cases were defined collaboratively. Within the
context of this PDEng project, a use case is seen as an intended utilization of the domain ontology. These
use cases supported the structuring and defining of the (informal) competency questions. In accordance
with the PDEng’s project goal, scope and objectives, in total seven use cases in the context of operations
and maintenance were defined. Altogether, these use cases are grounded as being a prerequisite of proper
life cycle management and operations and maintenance on utilities. Subsequently, competency questions
were defined and categorized within a particular use case. To this end, in collaboration with the CFM, I
specified the following list of use cases and their accompanying competency questions as a basis for the
domain ontology:
[1.] Assessment of the functioning and usage of a utility network and / or its individual components.
What kind of commodity does the utility network distribute?
What is the function of the utility network and / or its components?
What is the actual usage of the utility network and / or its components?
What is the state of operation of a utility network and / or its components?
[2.] Identification of a utility network and / or its individual components.
What is the location of a utility network and / or its components in the (1) x-, (2) y-, and
/ or (3) z-coordinate?
What is the spatial accuracy of the utility network and / or its components?
What are the interior and exterior dimensions of the utility component?
What is the shape of the utility component?
What is the (main) colour of the utility component?
What exterior, interior and filling material does the utility component exist of?
What is the unique ID of the utility network and / or its components?
[3.] Assessment of dependencies and redundancies between (1) utility networks, (2) a utility network
and its components, and (3) components.
How are utility networks related to each other?
How are utility networks and their components related to each other?
How are utility components within a single network related to each other?
How are utility components within distinctive networks related each other?
Section 2 – The engineering cycle 29
[4.] Mapping the impact of a utility network and / or its individual components.
What is the social impact of the utility network and / or its components?
What is the environmental impact of the utility network and / or its components?
What is the economic impact of the utility network and / or its components?
[5.] Mapping the risks involved within a network and its individual components.
What is the risk to damages due to excavation?
What is the risk to damages due to vegetation?
What is the risk to damages due to settlement of the subsurface?
What is the risk to function loss due to deterioration or any other reason?
[6.] Assessment of the performance of a utility network or its individual components.
Which performance requirements are set for the utility network and its components?
What is the nominal flow of a commodity through a distribution line?
What is the operational flow of a commodity through a distribution line?
What is the maximum flow of a commodity through a distribution line?
What is the age of the utility component?
For how long has the utility component been in use?
What is the expected lifespan of the utility component?
[7.] Planning of operations and maintenance related activities.
Who is the owner of the utility network and / or its components?
Who is the operator of the utility network and / or its components?
When was the last maintenance activity performed?
What has been done during the last maintenance activity?
How many utility components are enclosed within a particular protective component?
The above defined use cases and questions were taken as a first basis for the requirements of the ontology
and its initial design. Later on in the development process – i.e. validation of the domain ontology – the
use cases, competency questions, requirements and consequently the domain ontology were refined in an
iterative process.
5.2.2 List of requirements
Wieringa (2014) distinguishes both functional and non-functional requirements (in this report mentioned
as ‘structural requirements’). A functional requirement can be defined as a requirement for desired
functioning of the treatment. Examples are the function to map risks or assess the performance of a utility
network. A structural requirement on the other hand is any requirement with a non-functional property.
Examples are accuracy of output or interoperability with other systems (2014).
An ontology is only successful if it is capable of creating a shared set of conceptualization about the to-be
modelled domain. Therefore, to elicit the functional requirements, I particularly focused on end-user
engagement. This was realized through (1) the qualitative case study, (2) the expert panel sessions and (3)
regular contact with the CFM. The source of domain data retrieved from the qualitative case study allowed
direct insight in the information need of the end-user. The expert panel provided a multi-disciplinary
perspective of the utility sector. The CFM ultimately provided insight in the actual operations and
maintenance related activities and were the main contributor to the competency questions. These
competency questions defined the utility information to be covered by the ontology, thereby related to the
goal of the CFM to have real-time insight in the as-built information of their utilities. Altogether, these
sources of information were reflected upon – and subsequently coupled to – the stakeholder goals.
Section 2 – The engineering cycle 30
In addition to functional requirements, structural requirements were defined as well. These were elicited
from a literature review and document study. To this end, I studied five papers describing (structural)
principles for a good ontology (Gruber, 1995; Grüninger & Fox, 1995; Gómez-Pérez, 2004; Gangemi,
Catenacci, Ciaramita, & Lehmann, 2005; Obrst, Ceusters, Mani, Ray, & Smith, 2008). Eight criteria could
be defined being: (1) accuracy, (2) adaptability, (3) clarity, (4) completeness, (5) computational efficiency,
(6) conciseness, (7) consistency, and (8) organizational fitness.
The requirements were written down following the method proposed by INCOSE (2015). This method,
as explained before, distinguishes requirements from the five standpoints of (1) enterprise, (2) business
management, (3) business operations, (4) system and (5) system element. The complete list of
requirements for the domain ontology is presented in Table 6. In the second-last column of this table it is
mentioned whether I took responsibility for the requirement or not. A check mark illustrates the
requirement is taken care of within the PDEng project; a cross does mean the contrary. In the last column
of the table it is mentioned whether I could validate the requirement or not (discussed in chapter 6). Again,
a check mark illustrates the requirement is validated; a cross does mean the contrary.
5.3 Ontology design
This section elaborates on the chosen ontology development tool, chosen ontology development language
and rationale behind the ontology development process.
5.3.1 Ontology development tool
In order to develop an ontology, one requires an ontology development tool. Ontology development tools
are aimed at providing support for the development process of an ontology (Corcho, Fernández-López,
& Gómez-Pérez, 2002). Within this PDEng project I made use of a so-called ‘ontology editor’. These
tools help in organizing the overall structure of an ontology by modelling concepts, properties, relations
and constrains. In addition, these editors are often capable of reconciling syntactic, logical and semantic
inconsistencies among the elements of the ontology (Gasevic, Djuric, & Devedzic, 2009).
Many ontology editors, and moreover, ontology tools and environments for building ontologies, exist. The
broad range of tools may make it difficult to choose the right one for the ontology development process.
To make a grounded choice, I analysed multiple ontology development tools – including Protégé,
OntoloLingua, Enterprise Architect and WebODE – on four aspects:
[1.] Interoperability – Interoperability with ontology development languages.
[2.] Knowledge representation – Capability to model concepts, relations, classes, attributes and
instances.
[3.] Inference services – Services concerning constraints and consistency checking, automatic
classifications and exception handling.
[4.] Usability – Support of the user by a graphical taxonomy, graphical views, zooms and libraries.
Based on the analysis of the ontology development tools, I chose Enterprise Architect. This tool is
interoperable with many ontology development languages including UML, is capable of modelling a full
knowledge representation, is able to perform inference services, provides a graphical environment in
which the ontology is build, and moreover, is also used to model CityGML UtilityNetwork ADE. More
information about the assessment of ontology development tools is included in the report Designing an
ontology. A proposed framework. This report can be made available on request.
Section 2 – The engineering cycle 31
Table 6 – Goals, needs and requirements following the INCOSE method
# View Goals Needs Requirements Scope Validated
1.1 Enterprise Enterprise
strategy
Improved operations and maintenance of
utilities
Reduce costs Improved efficiency and effectiveness of operations and maintenance of
utilities.
1.2 Reduce errors and damages Reduction of work delays through real-time insight in status and
performance of utilities.
1.3 Reduction of deterioration and damages through real-time insight in
status and performance utilities.
2.1 Improved system interoperability of utility
data
Reduce miscommunication and
misinterpretation
Unified and uniform set of concepts and relations based on shared
conceptualizations of the end-user.
3.1 Business
Management
Business
Needs
Ontology is future-proof Modular Must allow new modules or elements to be added without changing its
already existing core.
3.2 Adaptable Must be easy to adapt (and extent) in case of changes in the information
need of users.
3.3 Must allow modelling of every kind of utility discipline.
4.1 Use of ontology is financially attractive Efficient Must lower life cycle costs of utilities through efficient operations and
maintenance.
4.2 Effective Must extend the lifespan of utilities through effective operations and
maintenance.
4.3 Affordable Must be open-source and free of use.
5.1 Business
Operations
Stakeholder
Needs
Real-time insight in the as-built information
of their utilities.
Accurate Must cover the expertise of its users in the correct fashion.
5.2 Complete Must cover all relevant concepts and relations within the domain of
operations and maintenance.
5.3 Concise Must not include redundant or irrelevant concepts and relations
regarding the domain of operations and maintenance.
5.4 Consistent Concepts and relations must be consistent without any contradictions.
6.1 Easy utilization of ontology Computational efficient Must allow an easy and successful reasoning process to the reasoners
(users)
6.2 Clear Must include a defined, explained and documented taxonomy and
hierarchy.
6.3 Must include a defined, explained and documented vocabulary of terms
and names.
6.4 Instructions Must provide instructions on how to apply the ontology which are easy
to read, learn and apply.
6.5 Organizational fitness Must comply with (spatial) asset management and GIS software.
Section 2 – The engineering cycle 32
6.6 Transparent Ontology and accompanying documentation must be made available
through an online repository.
6.7 User-friendly Components and relations must be easily traceable by the user.
7.1 System System
Needs
Supported by user and environment Support by (spatial) asset
management and GIS
Must comply with existing (spatial) asset management and GIS software
currently used in the utility sector.
7.2 Support by ontology modelling
languages
Must comply with existing ontology modelling languages (i.e. UML) to
be compatible with ontology modelling tools.
7.3 Support by digital modelling
standards
Must comply with utility data as modelled in existing digital modelling
standards (i.e. INSPIRE and CityGML UtilityNetwork ADE).
7.4
Supported by utility sector Must be supported by the utility sector through reaching consensus in
which concepts and relations to be covered in the ontology.
8.1 System
element
System
Element
Needs
Process utility data Data analysis Must allow analysis of data through a queries structure.
9.1 Support user Optionality Must allow partial use of its model in case that covers the information
use of its users.
9.2 Standardization Must acquire a certain level of standardization within the utility sector to
be effectively utilized
Section 2 – The engineering cycle 33
5.3.2 Ontology language
In order to allow the sharing, reuse and utilization of knowledge, ontologies are based upon ontology
languages. Ontology languages are defined as the formal languages that are used to encode an ontology
(Jakus, Milutinovic, Omerovic, & Tomazic, 2013). Ontology languages differ in the way they represent
knowledge (Corcho, Fernández-López, & Gómez-Pérez, 2002). Differences occur, for example, in the
way languages present attributes, taxonomies, relations, functions and axioms.
The Unified Modelling Language (UML) is the ontology language applied within this PDEng project. The
reason why this particular language was chosen, is because of its graphical notations for class modelling.
These graphical notations make the development of the ontology much more transparent and allows easier
communication of its content. It therefore, in literature (e.g. Cranefield, 2001), is often proposed as the
solution to bridge between software engineering and the semantic web and technologies.
A second reason to choose UML is the fact that is it also applied in similar digital utility modelling
standards. For example, both INSPIRE and CityGML UtilityNetwork ADE model their ontology in
UML. Developing the ontology within this PDEng project in UML, therefore, eases integration with and
/ or reuse of existing ontologies, i.e. digital modelling standards. More information about the assessment
of ontology languages is included in the report Designing an ontology. A proposed framework. This report
can be made available on request.
Figure 5 presents the UML notations as applied within the ontology development process following ISO
19103:2015. The figure presents the types of relations (association, aggregation, composition and
inheritance), the association cardinality (also called multiplicity) and its classes. A definition of these terms
can be found at the start of this report in ‘Terms and Definitions’, starting page V.
Figure 5 – UML notation (ISO 19103:2015 Geographic Information – Conceptual Schema Language)
Section 2 – The engineering cycle 34
UML knows multiple stereotypes. A stereotype can be defined as a metaclass that enables the modelling
of domain specific terminology or notation. In other words, stereotypes provides a categorization of UML
notations. Table 7 presents the stereotypes as applied within the domain ontology following the style used
in ISO 19103 and ISO 19109 for models of geographic information.
Table 7 – UML stereotypes applied in the domain ontology
Stereotype Model Element Explanation
ApplicationSchema Package Conceptual schema required by applications.
CodeList Class Value domain including a code for each permissible value.
DataType DataType Represent a set of properties without identity.
Enumeration Enumeration Data type whose numbers are enumeration literals.
FeatureType Class Represent a spatial object.
In addition to the stereotypes, attributes of the stereotypes are either described through other complex
types (i.e. other stereotypes) or through the values types as described in Table 8. The ‘date’ values follow
the Gregorian calendar with UTC as temporal reference system, following ISO 8601:2000. All units of
measure are expressed using the SI systems of unit, following ISO 1000:1992.
Table 8 – UML values types applied in the ontology
Value type Explanation
boolean Used for logical expression, consisting of the predefined values true and false.
date Describing the date in YYYY-MM-DD .
integer Integer values.
characterString Sequence of characters in set to display information.
measure Declared or measured quantity.
quantityExtent A range of integer values.
URI Uniform Resource Identifier used for identification.
To make all the above more explicit, an example of a (small) UML class model is provided in Figure 6.
This model is derived from the developed domain ontology. Four (FeatureType) classes, being ‘Network’,
RelatedParty’, ‘AbstractCommodity’, and ‘AbstractCommodityClassifier’ are shown, each at least
associated to one other class. For each association also its role is included, as well as its multiplicity. In this
example, a party is associated with a network, through the role ‘associatedNetwork’ and multiplicity of one
or many. This means a party is at least associated with one network, but may be associated to more. For
the ‘Network’ class also an aggregation is seen. In this example, an aggregation is used since a (sub)network
may exist of a network itself.
In addition to the relations, the stereotype and name are always shown at the top row of the class. The
middle row is used to describe the class’ attributes. Although not shown in the figure, a third row is
sometimes applied to model a constraint. The three colours as seen in Figure 6 are used for colour coding,
discussed in the next section.
Section 2 – The engineering cycle 35
Figure 6 – Example of a UML class model (from ‘Operations and Maintenance – Data specification’)
5.3.3 Build of the domain ontology
Even though the actual design of the domain ontology itself is not elaborated in this document, as
explained in section 1.4, I still would like to explain the rationale behind the choices made within the
ontology’s development process.
An ontology is a complex and multi-layered information source (Vrandecic, 2009). The following
components of an ontology exist:
Stereotypes (ApplicationSchema, Codelist, DataType, Enumeration, FeatureType)
Attributes
Attribute values (boolean, date, integer, characterString, measure, quantityExtent, URI)
Relations (association, aggregation, composition, inheritance)
Together, these components form the ontology’s (1) taxonomy and hierarchy, and (2) vocabulary, terms
and names (Gasevic, Djuric, & Devedzic, 2009). To prevent ambiguity and multi-interpretability of the
applied vocabulary, terms and names – i.e. semantic issues – definitions and explanations for all terms
and names were defined in a catalogue.
The build of the ontology revolves about the development of the three ontology elements within the
typology of Gasevic et al. (2009). To build the ontology, I distinguished its development process in two
phases: (1) development of the informal description and (2) development of the formal ontology. Both
these phases are explained below.
[1.] Informal description
The informal description of the ontology contains the model of the domain ontology in a textual and
descriptive manner. This means the ontology is not yet computer-interpretable. The informal description
forms the conceptual framework for the to-be formalized domain ontology and includes a textual
description of the outline of its taxonomy and vocabulary. In other words, the informal description of the
ontology describes the (information) domains covered within the ontology.
Section 2 – The engineering cycle 36
The before defined use cases and accompanying informal competency questions played a crucial role in
the development of the informal description. In order to answer these questions, I defined eight domains
that the ontology should cover. These domains were partially adapted from the CityGML UtilityNetwork
ADE digital modelling standard, whereas new domains were added to cover the domain of operations and
maintenance.
The informal description of the ontology includes the following eight domains and their main
characteristics:
[1.] Core and geometry
Defines the topographic model of the utility network
Defines the topological and functional model of the utility network
[2.] Functional characteristics
Defines the functional role of objects in the utility network
Defines the area the commodity is supplied to by the network
[3.] Network properties
Defines the commodity transported by the network as well as its characteristics
Defines the related organizations to the utility network
[4.] Network components
Defines the distribution components for transport and distribution of a commodity
Defines the functional components within the network
Defines the protective components relevant for the security of the network
[5.] Component properties
Defines the dimensional properties of utility network components
Defines the surrounding soil and groundwater properties of utility network components
Defines the material properties of utility network components
Defines the related organization to the utility network components
Provides extra information concerning identification of utility network components
[6.] Maintenance and operations properties
Defines the maintenance properties of utility network components
Defines the costs properties of utility network components
Defines the date properties of utility network components
[7.] Performance properties
Defines the performance properties of utility network components
Defines the environmental, social and economic impact of utility networks and their
components
[8.] Hollow space
Defines the free and occupied space of distribution and protective components
The eight domains do not include domains specifically oriented towards a single utility discipline.
Although debated, reason to not structure the domain ontology towards utility specific domains, is that
with the inclusion of the eight domains, still the discipline specific information can be modelled.
Therefore, more models would make the domain ontology unnecessarily complicated.
In addition to CityGML UtilityNetwork ADE conceptual schema, many new concepts and relations were
added within the context of operations and maintenance. Since the focus of the CityGML UtilityNetwork
Section 2 – The engineering cycle 37
ADE schema is different to that of the domain ontology in the PDEng report, it lacks information required
for operations and maintenance. For example, newly added concepts and relations to the CityGML
UtilityNetwork ADE are amongst others related to (1) related party, (2) performance, (3) dimensional, (4)
surrounding soil, (5) cost and (6) maintenance properties. Main addition to CityGML UtilityNetwork ADE
are the domains of ‘Maintenance and operations properties’ and ‘Performance properties’.
[2.] Formal description
The formal description of the domain ontology is the computer-interpretable translation of the informal
description. The domain ontology is developed in UML within the ontology development tool Enterprise
Architect. For each of the eight domains as defined within the informal description, a UML class model
was developed. These UML class models are included within the ‘Operations and Maintenance – Data
Specification’. In addition to the UML class models, this document does also contain an elaborate
explanation for each of the class models.
Reason to not model all eight domains within a single class model, is to keep the class model
understandable. Instead of having one big model, I found it to be more logic, understandable and clear to
have eight class models each covering a specific topic. This follows the approach of other digital modelling
standards such as INSPIRE, IMKL and the CityGML UtilityNetwork ADE. However, the eight class
models are still connected to each other. Therefore, it can be seen that within the ontology, the same class
is used within various of the eight models.
A design choice I made during the development of the ontology in UML, was to have abstract classes at
the top of the taxonomy, and detailed classes at the bottom. Since one of the requirements is to have a
modular and adaptable ontology, it should be relatively easy to alter the ontology in the future. Due to the
abstract classes at top of the taxonomy, the ontology allows removal and addition of classes without
changing the core of the class model.
Since the UML class models follow the eight domains as explained within the informal description, they
also are partially adapted from CityGML UtilityNetwork ADE. To make explicit what was copied from,
copied with alterations from or newly added to the CityGML UtilityNetwork ADE schema, I applied
colour coding as seen before in Figure 6 in the follow way:
Purple : copied from CityGML UtilityNetwork ADE version 0.9.2
Orange : copied with alterations from CityGML UtilityNetwork ADE version 0.9.2
White : newly added to CityGML UtilityNetwork ADE version 0.9.2
Alterations to the CityGML UtilityNetwork ADE classes include changes in the naming of the classes and
attributes, the addition of new attributes, and the removal of existing attributes. These alterations were
made to better fit the classes to the domain of operations and maintenance of utilities.
To illustrate both the use of the colour coding, as well as how the ontology looks like, the UML class
model for ‘Network Components’ is provided in Figure 7. Within this class model, it is, for example, seen
that the class ‘Bedding’ is copied, the class ‘AbstractDistributionComponent’ is copied with alterations,
and the class ‘WaterPipe’ is newly added. Figure 7 also shows the use of so called ‘Abstract’ classes. These
classes are used to provide some implementation, but do not allow instantiation. In other words, abstract
classes provide some basic implementation that is inherited by its subclasses. The abstract class is, thereby,
not considered as an instance, whereas its subclasses are.
Section 2 – The engineering cycle 38
Figure 7 – Network components UML class model (from ‘Operations and Maintenance – Data Specification’)
Section 2 – The engineering cycle 39
Figure 7 provides a good overview of how the ontology as a whole is modelled. In addition, the figure also
shows all stereotypes as applied within the ontology as a whole: data types, feature types, enumerations
and codelists. All types of relations can be recognized as well: association, aggregation, composition and
inheritance. For a complete overview of the domain ontology – including all eight class models and their
explanation – I refer to the ‘Operations and Maintenance – Data Specification’ document.
To explain all the terms and names as applied within the ontology and its UML class models, I developed
a feature catalogue. This catalogue contains the complete vocabulary of the ontology, providing a definition
and explanation for each term and name. More specific, I applied the reporting style for data types and
feature types as presented in Table 9. The applied reporting style for codelist and enumerations is
presented in Table 10.
Table 9 – Reporting style applied within Feature Catalogue for feature types and data types
Name
Name
Definition
Description
Stereotype
Name of spatial object type
Definition of spatial object type
Additional description of spatial object type
Stereotype of spatial object types (for example FeatureType)
Attribute:
Name
Value type
Definition
Multiplicity
Name of attribute
Value type of attribute (for example boolean)
Definition of attribute
Multiplicity of attribute (for example 0…1)
Association role:
Name
Class
Definition
Multiplicity
Name of association role
Associated class
Definition of association role
Multiplicity of association role (for example 0…1)
Table 10 – Reporting style applied within Feature Catalogue for codelist and enumerations
Name
Name
Definition
Description
Extensibility
Stereotype
Values
Name of codelist or enumeration
Definition of codelist or enumeration
Additional description for codelist or enumeration
Extensibility of codelist or enumeration by the user
<< Codelist >> or << Enumeration >>
Values within codelist or enumeration
The addition of the feature catalogue is necessary to cover the third element of an ontology: the semantics.
Without the feature catalogue, the linguistic meaning of the terms and names applied in the ontology can
be interpret differently by each individual user. The feature catalogues reduces this from happening, and
therefore, improves the system interoperability of utility data once modelled according to the domain
ontology. For a complete overview of the feature catalogue I refer to the ‘Operations and Maintenance –
Feature Catalogue’ document.
Section 2 – The engineering cycle 40
6 Treatment validation
This chapter elaborates on phase three of the engineering cycle, the treatment validation. Validation
assesses whether the ontology does contribute to the stakeholders goals, while meeting the requirements.
In other words, validation is done to assess whether the ontology is ready for implementation or still
requires refinement in its design (2014). Within this chapter, project objective 4 is covered.
To validate the domain ontology, I applied four validation techniques, being: (1) assessment against a
source of domain data, (2) check against class modelling rules, (3) assessment against expert panel input
and (4) assessment against competency questions. Within the following sections, I explain how each of
these techniques were applied in the PDEng project. By applying the four techniques, I was able to validate
all requirements within scope of the project, as shown before in Table 6 (page 31). This chapter concludes
with an overview of the gradual development of the ontology.
In addition to validation as described within this chapter, many (informal) meetings held during the entire
PDEng project supported the ontology development process. For an overview of these meetings as well
as their influence on the design of the ontology, I refer to Appendix II.
6.1 Validation techniques
This section elaborates on the validation techniques I applied within the PDEng project. For each of the
techniques, their purpose and outcome are explained. In addition, the number of iterations of, and
requirements treated by the technique are provided. I only refer to the requirements by their number,
following Table 6.
6.1.1 Assessment against source of domain data
Ontologies can be compared and evaluated with respect to other ontologies, knowledge bases and
classification models representing the same domain. Assessment of the domain ontology against a
comparable source of domain data can provide useful information about the relative quality of the
ontology (Obrst, Ceusters, Mani, Ray, & Smith, 2008). In the validation of the domain ontology, I
compared the domain ontology with (1) the domain data from practice (elicited from the qualitative case
study) and (2) comparable digital modelling standards.
Within the qualitative case study utility data was obtained from twelve utility owners, covering the
disciplines of water, gas, electricity, telecommunication and thermal. I received additional data for the
disciplines of water, gas and electricity – from the same Dutch engineering company as visited during the
qualitative case study – later on in the ontology development process. I compared the ontology’s concepts
and relations in a qualitative manner with those I recognized within the utility data from practice. As a
result, I was able to refine the ontology to the extent that it was capable of modelling the utility information
stored in the digital models from practice. The ontology was validated twice against the data from practice:
the first time against the original data from the case study, the second time against the newly acquired utility
data. Treated requirements mainly involved 3.3, 5.1 to 5.4, 7.1 and 7.4.
I also validated the ontology against comparable digital modelling standards, including INSPIRE,
CityGML UtilityNetwork ADE and IMKL. Since these standards are all modelled in UML, I could
compare the concepts and relations captured within those UML class models, with the ones of the domain
Section 2 – The engineering cycle 41
ontology. In addition, I analysed their vocabulary and accompanying glossary of terms and names. To this
end, it was validated whether the ontology could model utility data as modelled within comparable digital
modelling standards. The ontology was evaluated twice against the aforementioned standards: first time
for the first version of the ontology, the second time for the final version of the ontology. Treated
requirements mainly involved 3.3, 5.1 to 5.4, 7.3 and 7.4.
6.1.2 Check against class modelling rules
Class modelling rules are applied to perform a structural check on the domain ontology from a modelling
perspective. Within Enterprise Architect the domain ontology is evaluated on its hierarchy i.e. its
taxonomy, based on multiple UML class modelling rules. Enterprise Architect allows validation of the
UML class models based on the following four rules (Sparx Systems, 2017):
[1.] Well-formedness – Rules to check whether or not an element, relationship, feature or diagram is
well-formed, for example, whether the object is a valid UML item or whether a diagram contains
valid elements within it.
[2.] Element composition – Rules to check whether or not the UML element contains valid children,
whether it contains the right number of valid children, and whether or not the element is missing
any required children.
[3.] Property validity – Rules to check whether or not the element, relationship or feature has the
correct UML properties defined, and whether the properties contain incorrect of conflicting
values.
[4.] Custom properties – Rules to check an element, relationship or feature against any defined
constraint in OCL.
The ontology was validated after each new version against the class modelling rules, adding up to a total
of eight iterations. The final version of the ontology (version 4.1) did not show any errors when checked
on the UML class modelling rules. Treated requirement mainly involved 7.2.
6.1.3 Assessment against expert panel input
An expert panel was formed as a support group during the entire ontology development process. The
expert panel is a multi-disciplinary panel consisting of the CFM, information managers, the cadastre and
standardization establishments. In total two plenary expert panel sessions were held, complemented by
multiple individual sessions with experts. Both are explained in the subsequent paragraphs. Appendix V
provides the handouts developed for each of the plenary expert panel sessions.
Two plenary panel sessions were held. Whereas the first plenary session provided mainly input for
requirements engineering, purpose of the second session was explicitly to validate the ontology on its use
cases and accompanying competency questions. The session provided useful insights in which domain
knowledge – i.e. concepts and relations – was still missing within the domain ontology. In addition, the
representative from a standardization establishment gave useful advice regarding the use of association
roles and abstract classes in the UML class model. Based on the outcome of the second plenary session,
the ontology was refined accordingly. Treated requirements mainly involved 2.1, 3.1 to 3.3, 5.1 to 5.4,
6.2, 6.7 and 7.2 to 7.4.
Individual sessions were held with two standardization establishments (amongst others CityGML
UtilityNetwork ADE), an information manager and the client, the CFM. Aim of the individual expert
validation sessions was to validate the ontology more in-depth, by for example, validating a single UML
Section 2 – The engineering cycle 42
class model in detail. Excluding the sessions with the CFM, up to ten individual sessions were held, and
thereby, validation iterations. Outcomes of the individual sessions mainly involved small revisions on the
ontology’s concepts, relations, or vocabulary. To this end, the individual sessions provided a continuous
input during the entire ontology development process. Treated requirements mainly involved 2.1, 3.1 to
3.3, 5.1 to 5.4, 6.2, 6.3, 6.7, 7.1 to 7.4 and 9.1.
6.1.4 Assessment against competency questions
Competency questions are applied to validate whether the ontology can model the information that is
asked for. The ontology was validated three times against its competency questions. The first iteration took
place right after the establishment of the first version of the ontology. The second iteration was performed
during the second plenary expert panel session. Due to the fact that the ontology could not be validated
against all of its use cases and competency questions within that session, I validated the ontology once
more against the competency questions, i.e. the third iteration.
Outcomes of this validation measure mainly provided input on the coverage of the ontology’s concepts
and relations. In case a question could not be answered by the ontology, the ontology was refined
accordingly. Since competency questions were also used as a basis for the design, refinements as a result
of these validations were subtle. Treated requirements mainly involve 5.1 to 5.4.
Requirements missing in the aforementioned sections include number 4.3 (open-source and free of use),
6.5 (compliance with (spatial) asset management and GIS software) and 6.6 (visibility within online
repository). Concerning 4.3 and 6.6, the ontology has been made available an online repository (GitHub)
and is open source and free of use for everyone. Concerning 6.5, the next chapter shows a small proof of
concept of the ontology being modelled in GIS software.
6.2 Development of ontology
During the entire ontology development process, in total eight versions of the ontology were developed.
These versions were the result of the validation outcomes and subsequent refinements of the ontology.
To illustrate how the ontology evolved during its development process, I kept a changelog in which the
most important changes to the ontology were documented. Table 11 presents this changelog.
The changelog shows changes made within the ontology from the first version of the ontology, version 1.0,
to ultimately the final version of the ontology, version 4.1. All versions of the ontology can be made
available if requested. It should be acknowledged that this changelog only makes the most important
changes explicit. Subtle changes, such as rephrasing of a class name, are grouped together under one
header in the changelog.
Table 11 – Changelog of ontology development process
Version Change(s) made
4.1 - ‘MaterialValue’ values rephrased
- ‘Pipe’ and ‘Cable’ changed to ‘AbstractPipe’ and ‘AbstractCable added
- New subclasses for ‘AbstractPipe’ and ‘AbstractCable’
- Class names and attribute names rephrased
4.0 - ‘AbstractFunctionalComponent’ and its subclasses reframed
- Extra type values added for controller, storage, connection, measurement, terminal, other and complex
components
- ‘MaintenanceProperties’ reframed
Section 2 – The engineering cycle 43
- ‘ProtectiveShell’ and its subclasses reframed
- RelatedParty’ and its data types reframed
- Geometric primitives reframed
- Extra type values added for electrical, solid, liquid and gaseous commodities
- Class names and attribute names rephrased
3.2 - Attributes added to subclasses of ‘AbstractFunctionalComponent’
- ‘AbstractExtraInformation’ and its subclasses reframed
- Class names and attribute names rephrased
3.1 - Separate ‘Performance properties’ UML class model developed
- Separate ‘Maintenance and Operations’ UML class model developed
- Class names and attribute names rephrased
3.0 - ‘MaintenanceProperties’ reframed
- ‘NetworkClassValue’ values rephrased
- ‘StatusValue’ values rephrased
- New association roles added
- Class names and attribute names rephrased
2.1 - ‘PerformanceProperties’ and its subclasses reframed
- ‘MaintenanceProperties’ reframed
- Class names and attribute names rephrased
2.0 - ‘AbstractFunctionalComponent’ and its subclasses reframed
- ‘Depth’ class added
- ‘GroundWaterProperties’ class added
- Class names and attribute names rephrased
- Colour coding added
1.1 - Class names and attribute names rephrased
1.0 First version of the domain ontology
In addition to the changes made to the ontology itself, I also kept a changelog for the data specification
and feature catalogue as a whole. These changelogs can be found in the corresponding documents referred
to in section 1.4.
Section 2 – The engineering cycle 44
7 Treatment implementation
This chapter elaborates on phase four of the engineering cycle, the treatment implementation. Treatment
implementation is defined as the application of the treatment to the original problem context (Wieringa,
2014). Therefore, within the context of this project, implementation means the actual utilization of the
domain ontology in (spatial) asset management and GIS software. Within this chapter, project objective 5
is covered.
Within the PDEng project, implementation is limited to a small proof of concept. Full scale
implementation has not been part of the PDEng project, and therefore, has not been conducted. In the
first section, I provide a brief outline on how the domain ontology was implemented. In the second section,
I refer to the online repository where all documentation and files of the ontology can be found.
7.1 Proof of concept
This section provides an outline on how the domain ontology was implemented within the PDEng project.
First, I explain the application domain of the ontology. Second, I elaborate on the preparations I
performed to make the domain ontology interpretable by (spatial) asset management software and GIS.
Third, I explain how the ontology can be implemented by elaborating on a small proof of concept.
7.1.1 Application domain
The targeted use of the domain ontology lies in (spatial) asset management and GIS software. The domain
ontology is applicable on multiple utility disciplines, being: (1) electricity, (2) oil, (3) gas, (4) chemicals, (5)
sewage, (6) water, (7) thermal and (8) telecommunication. The use of the domain ontology has no
geographic boundary, but is based on utility networks of developed countries.
Asset management software and GIS can also be referred to as an application. Following ISO 19101-
1:2014, an application is defined as software used to manipulate and process data in support of user
requirements. Use and implementation of the domain ontology within such applications has the potential
to:
facilitate the transfer between format and shape standardized formats within utility models.
support smarter reasoning within utility models.
Support of smarter reasoning is especially beneficiary to the end-user, recognized as utility owners, utility
operators and contractors. Once accepted and shared, a domain ontology – i.e. a digital modelling
standard – allows optimization of an application that is fully compatible with this standard. To this end,
the ontology can enable digital maintenance practices, and even the creation of cyber-physical systems.
7.1.2 Preparation
The domain ontology is developed within the software Enterprise Architect as a UML class model.
However, a UML class model is not directly interpretable by (spatial) asset management and GIS software.
Therefore, I converted the class models to an XML (Extensive Markup Language) schema file. XML is a
widely adopted standard language for exchanging information (Bray, Paoli, Sperberg-McQueen, & Maler,
2008). In order to translate the UML class models of the domain ontology towards the XML schema file,
I executed the following steps:
Section 2 – The engineering cycle 45
[1.] Defining configuration file:
The XML schema file can be automatically generated from Enterprise Architect through
the software ShapeChange. However, in order to generate the XML file, ShapeChange
had to be configured first through a configuration file written in Java. This configuration
file was available from the CityGML UtilityNetwork ADE online repository (Kutzner,
2018). Through a few modifications, I was able to make the configuration file work with
the UML class models of the domain ontology.
[2.] Generation of XML file from UML of Enterprise Architect:
By running the configuration file, an XML schema file of the UML class models within
Enterprise Architect was generated. In addition, an XML file for each of the codelist of
the domain ontology was created. Having the codelists in human interpretable code,
allows manual adjustment of its values.
Whereas the XML schema file contains the domain ontology in code, it does not contain any utility data
yet. To store utility data in the domain ontology’s format, I mapped utility data to the XML schema file
within the software FME. This required the following step to be executed:
[1.] Mapping utility data to XML schema file:
In order to map utility data to the XML schema file, first the feature types of the domain
ontology were imported in the FME workspace. This allows mapping of the utility data
towards the feature types as depicted in the domain ontology. The feature types were
imported via an XML definition file I developed, including all feature types of the domain
ontology. I used an FME workspace from the CityGML UtilityNetwork ADE (den Duijn,
2018) as a basis, and adjusted it accordingly to make it work with the XML schema file.
7.1.3 Implementation and use
The created workspace in FME allows mapping of utility data according to the domain ontology’s format.
Once the mapping in FME was done, I ran the workspace to write a set of sewage data towards a
Geography Markup Language (GML) file. The GML format is introduced by the OGC and capable of
representing spatial information. In addition, GML is widely supported by GIS software.
To this end, I read the GML file in the open source GIS software QGIS. Figure 8 shows a snapshot of
QGIS in which the modelled sewer pipes are highlighted. On the right side of the figure, QGIS provides
a table including the attributes of the object at hand, i.e. the sewer pipe. In the figure, the attributes
‘cathodic protection’ and ‘pressure grade’ are seen, which are also attributes of a sewer pipe within the
domain ontology.
Ultimately, this small test shows that the domain ontology can be incorporated in (spatial) asset
management and GIS software. Therefore, the created GML file acts as a proof of concept. It shows that
the code derived from the domain ontology – the XML schema file – is structurally sound.
Section 2 – The engineering cycle 46
Figure 8 – Utility data modelled according to the domain ontology’s format in QGIS
However, further implementation of the domain ontology requires the mapping of the remaining feature
types in the FME workspace. The feature types can be imported through the feature type definition file,
available at the online repository of the domain ontology (as shown in next section). Once all feature types
are mapped, one can map utility data towards a GML file following the domain ontology’s format.
7.2 Online repository
For the developed domain ontology to be successful, it has to be accepted and shared within the utility
sector. A first step to acquire this is to make the domain ontology available. To this end, an online GitHub
repository is created to make all the documentation and data of the domain ontology available:
https://github.com/RamonTerHuurne/UtilityNetwork-OperationsAndMaintenance
The documentation includes the data specification and feature Catalogue of the domain ontology. In
addition to the documentation, the repository includes the following:
[1.] The UML class model of the domain ontology created within Enterprise Architect:
The .eap file (Enterprise Architect file) containing the UML class model.
The UML class models in in PDF.
[2.] The XML schema file of the domain ontology directly derived from the UML class model using
the software ShapeChange:
The file containing the XML schema file.
Codelist dictionary in .xml accessible for alterations.
[3.] The ShapeChange configuration file to derive the XML schema file and code lists dictionaries:
For further use of the software it is referred to http://shapechange.net/.
[4.] The FME feature types definition file.
Section 3 – Concluding words 47
8 Discussion
The utility sector is characterized by its widespread in utility modelling practices. Utility owners apply
standards from the international, national and organizational level to model their utility data for operations
and maintenance related activities. As a consequence, the modelled domain knowledge differs on its
regional coverage and its ontological aspects taxonomy, vocabulary and semantics. Experiencing the lack
of a standard for operations and maintenance related activities herself, the client within the PDEng project,
the CFM, opted for the development of a shared and accepted digital modelling standard. To this end, I
developed a domain ontology for the domain of operations and maintenance.
An ontology is a complex and multi-layered information source making its development process rather
difficult. Within the project, I acknowledged that perceptions of human beings – including the modellers
of the ontology – are incomplete and bounded by rationality, resulting in various viewpoints (Olde
Scholtenhuis & Hartmann, 2015). As a consequence, an ontology is always at best a partial understanding
of the world (Turk, 2001). Therefore, for the ontology to be successful it is crucial to create a shared and
accepted set of conceptualization about the domain of utilities’ operations and maintenance. Acquiring
this shared and accepted set of conceptualizations was accomplished through active end-user engagement.
Whereas in the development of comparable standards often a top-down approach is applied to formulate
the concepts and relations, within the PDEng project a mixed approach of both a bottom-up and a top-
down approach was applied. I perceive active end-user engagement as a key driver for the establishment
of a shared and accepted set of conceptualizations about the domain.
One of the means applied to incorporate this principle of end-user engagement, involved the qualitative
case study on utility modelling practices. However, this qualitative case study also had its limitations. The
source of domain data acquired was relatively small and only focused on the Dutch utility sector.
Therefore, the data may not be fully representative for the full utilities’ industry world-wide, although it
still managed to cover utility data from a very high level of detail. Detailed information from water, gas,
electricity, telecommunication and thermal utility networks was acquired and in-depth and tacit knowledge
about the domain of operations and maintenance was retrieved.
Reflecting upon the engineering cycle, the domain ontology has not been fully implemented within the
application domain. Reason is mainly that implementation is a very time-consuming process and did not
fit within the time available for the PDEng project. As a consequence, the last part of the engineering
cycle - treatment evaluation – has not been assessed within the PDEng project. The use of formal
competency questions as a means of validation was, therefore, also not applied within the project. For
proper validation of the domain ontology’s structural soundness, validation based on these formal
measures is important to conduct within future studies.
Continuing on validation, various requirements could not be measured and, therefore, validated during
the PDEng project. These requirements include the required improved effectivity and efficiency of
operations and maintenance, reduced deterioration and damages to utilities, reduced life-cycle costs of
utilities, computational efficiency, ability to run queries, and effective utilization of the ontology throughout
the utility sector. Reason is that these requirements mainly require the ontology to be implemented,
whereas, as explained, full-scale implementation of the ontology in its application domain did not take
place within the PDEng project. Further studies should elaborate on the treatment validation of the
ontology and its assessment against aforementioned requirements.
A final point of discussion concerns the actual adoption of the ontology within the utility sector.
Momentarily the domain ontology is developed in a rather small environment. Drawing upon
Timmermans and Epstein (2010), a standard is only successful once it is shared and accepted within its
community. Otherwise, a standard may end up in the already existing ‘world of standards, but not a
standard world’. However, adoption of the domain ontology may be a rather difficult process. Hesitation
Section 3 – Concluding words 48
and resistance to use the model may occur. Utility owners and other end-user are presumable not willing
to alter their complete routine of utility modelling practices. Drawing upon Geels (2004), the existing socio-
technical regime prevents the domain ontology as an innovation to be directly introduced.
To break this existing regime, time is required. Because of its low initial adoption, the ontology will not
directly be capable of prevailing over existing digital modelling standards. A gradual development of its
social network, in the form of end-users, software developers and possibly other standardization
establishments, is important in its adoption process. This adoption process could be sped up in case a
particular window of opportunity appears to break the socio-technical regime. Such a window of
opportunity may come from new mandatory legislation, but also from new ideas on the assessment of
lifecycle management on subsurface utilities.
Section 3 – Concluding words 49
9 Conclusions and recommendations
This chapter elaborates on the conclusions and recommendations. In the conclusions, I also refer back
to the PDEng assessment criteria. Within the recommendations, I specify recommendations for both
further development and further adoption of the domain ontology.
9.1 Conclusions
As a utility owner and operator, the CFM – client within the PDEng project – is responsible for the
installation, operations and maintenance of her utilities. Consequently, the CFM wants to be able to rely
on uniform and consistent digital information about these utilities. This is also recognized within the utility
sector as a whole. Due to the sector shifting towards a more lifecycle oriented management approach,
utility owners increasingly want to have a comprehensive set of digital information about their utilities.
Problem however, and thereby the starting point of the PDEng project, is that the utility sector currently
lacks a digital modelling standard with the emphasis on lifecycle management, i.e. operations and
maintenance. Therefore, it was found that within the utility sector a plethora of utility modelling practices
to model domain knowledge exists. The modelling practices describe utility concepts using different (1)
attributes, formats, relations and (2) semantic terms. Altogether, the lack of a digital modelling standard –
i.e. domain ontology – leads to confusion and misunderstandings when exchanging data during the
preparation and execution of utility works.
Main goal during the two-year course of the PDEng project has been the development of a domain
ontology that does include those concepts and relations required within operations and maintenance
related activities. This domain ontology offers the CFM – and utility owners, utility operators and
contractors in general – a structure alongside they can store their utility asset information to utilize it for
installation, operations and maintenance purposes. The ontology can enable digital maintenance practices,
and possibly even the creation of cyber-physical systems. The ontology has been documented in three
documents:
[1.] Operations and Maintenance – Data Specification
[2.] Operations and Maintenance – Feature Catalogue
[3.] Operations and Maintenance – Overview of class models in UML
The data specification provides the UML class models of the domain ontology, together with a detailed
description of the class models and domain ontology as a whole. The feature catalogue provides the
complete overview of all the terms and names included in the vocabulary, with for each of these a definition
and explanation. This overcomes semantic issues such as ambiguity and multi-interpretability of terms and
names. The overview of class models in UML is for those only requiring the sole class models of the
domain ontology.
Within this conclusion I would also like to refer back to the criteria of interest to the PDEng design
product and design process, starting with the design product:
Construction: Concerning the structure of the ontology itself as well as the structure of the documentation,
it follows the structure as applied in comparable digital modelling standards, in specific CityGML
Section 3 – Concluding words 50
UtilityNetwork ADE. Therefore, the structure is easily recognizable by it users. In addition, the ontology
is split up in eight models with clear and explained interfaces between them. The decision to have separate
documents for both the product and the process also supports the clarity and clear structure of the work.
The inventivity of the ontology in the utility sector is relatively high. No other digital modelling standard
within the domain of operations and maintenance of utilities exists. This also justifies the development of
the domain ontology within the PDEng project. Concerning convincingness, it was proven in the validation
of the ontology that it is capable of fulfilling to the information need of its users – in specific, the CFM –
by, for example, answering to an extensive list of competency questions.
Realisability: Although the ontology may at this point in time be relatively hard to model into asset
management software, I was able to do a small proof of concept. In addition, it is explained in this
document how the ontology can be implemented. For future adoption, it is recommended that software
developers build the ontology as a data layer within existing (spatial) asset management and GIS software.
However, before this to happen, the utility sector should start adopting the model. The more organizations
that use the domain ontology, the more likely that software developers will consider integration of the
ontology in their software. In section 9.2, I provide recommendations to facilitate the uptake of the
ontology in the utility sector. The economic realisability, on the other hand, is very high, since the ontology
is open source and, therefore, economic feasible to everyone interested.
Functionality: The domain ontology is able to model the information need of its end-users – in specific
the CFM –, satisfying the ontology’s main requirement. Through active end-user engagement, the ontology
was developed based on a shared and accepted set of conceptualizations about the domain of operations
and maintenance. However, eight requirements could not be met, since these required full-scale
implementation of the ontology, being out of the project scope. Working routines in asset management
and GIS related software can still remain the same, since the ontology can be modelled as the layer existing
standards are linked to. Concerning its reusability, the ontology is applicable to utility networks of
developed countries and does allow modelling of extensive sets of utility data. In addition, the ontology is
based on a modular design, thereby allowing future developments and adaptations – such as the emerge
of a new utility infrastructure type – without having to change its core structure.
Impact: The ontology does support lifecycle management of utilities. Due to having a more uniform and
consistent overview of their utilities, users may more effectively and efficiently operate and maintain their
utilities. Risks to the user by using the ontology do not exist, although the conversion from existing and
applied utility modelling standards to the use of the domain ontology may require manual labour. This
may make this conversion process prone to faults and errors.
Presentation: In total four reports were written, including this PDEng report, the data specification, the
feature catalogue, and the sole overview of the eight class models of the domain ontology. The correctness
of the information as described within these documents is validated by various measures as reported in
this document.
The following conclusions can be made regarding the criteria concerning the design process:
Organization and planning: For the qualifier of the PDEng a proposal was made including a detailed
planning, containing of milestones, specifications and activities during the project. Due to changes and
new insight within the project environments, the planning was slightly altered during the course of the
PDEng project. Especially since development of an ontology is necessarily and iterative process, it was
Section 3 – Concluding words 51
decided to have a deadline for its final version. In addition, meetings were planned in a weekly to monthly
fashion with the CFM and supervisors of the University of Twente. The meetings were, above all,
conducted when necessary. During the meetings, I always took care of the preparations and agenda.
Problem analysis and solution: The design strategy I applied within the PDEng project is the engineering
cycle of Wieringa (2014). Following the engineering cycle, I executed the phases of problem investigation,
treatment design, treatment validation and treatment implementation. I investigated the problem through
the means of a stakeholder analysis, a qualitative case study, a literature review, a document study and the
inclusion of an expert panel. In order to design the ontology, I first established an ontology design method
through an in-depth study of the ontology development process. Various ontology development methods,
tools and languages were considered. Ultimately, the ontology development method was coupled to the
engineering cycle of Wieringa to make the entire design process clear and consistent. Principles applied
to develop the ontology include active end-user engagement in combination with a mix of a bottom-up
and top-down approach. Eventually, the domain ontology was validated. Thereby, it was shown that the
ontology met its requirements (except for those related to implementation). Even more, a small proof of
concept of the ontology within its application domain was performed. In terms of impact of the solution,
the impact of the ontology, especially in socio-technical terms, is high. Within the utility sector, utility
owners, operators and contracts are digitizing utility data according to their own utility modelling practices
(organizational standards). A change in the utility modelling regime may be seen as a radical innovation,
though in the end will reduce interoperability issues. In addition, smarter reasoning in utility models
through analysis software is made possible. Even more, all IT used in the field can be based upon the
domain ontology, improving interoperability even on a higher scale.
Communication and social skills: In total four reports were written, being (1) the PDEng report, (2) the
data specification of the domain ontology, (3) the feature catalogue for the domain ontology, (4) and the
overview of UML class models. The documents are split up to have publishable documents for the
ontology. The data specification, feature catalogue and overview of class models specifically focus on the
product. The PDEng report specifically focuses on the process. In addition to the reports written, I also
actively communicated with the CFM, but also the experts involved, to keep them updated but also to
discuss about the progress within the project. All meetings were written down in a logbook, enclosed to
this report in Appendix II. To extent the outreach of the project, various presentations were given, for
example at the ARCOM conference and the ZoARG symposium.
Structure and attitude: Concerning structure, multiple documents were developed, as explained before,
Whereas the publishable documents – the data specification, feature catalogue, and overview of UML
class models – are for external use, the PDEng report itself is for internal use. Concerning my own attitude,
which is possibly rather subjective, I have been rather critical and self-reflective at my work, but also other’s.
With other work, for example, comparable digital modelling standards but also feedback from experts is
meant. I tried to keep my personal opinion throughout the whole development process, though I also
accepted that experts are generally more knowledgeable in the field. Therefore, a healthy balance between
one the one hand accepting other’s opinions and feedback and on the other hand my own opinion was
maintained.
I have to acknowledge that I perceived these criteria from my point of view. Other opinions may exist.
Section 3 – Concluding words 52
9.2 Recommendations
Several recommendations can be made according to the developed domain ontology. Recommendations
are, thereby, divided into recommendations for further (1) development of the domain ontology and (2)
adoption of the domain ontology:
[1.] Further development:
Implementation (evaluation): The ontology has not been fully implemented in (spatial) asset management
and GIS software. The proof of concept should be extended towards a complete workspace in FME,
capable of mapping utility data to the whole domain ontology. Implementation also allows evaluation of
the ontology through formal competency questions (queries). These queries can perform sophisticated
geospatial analysis of the utility networks to validate the ontology against its use cases.
[2.] Further adoption:
Sector-wide coverage: The domain ontology may also be applied to IT used in the field, such as the ground
penetrating radar (GPR) and sensors. Once all IT stores and registers utility data according to the same
format, one can, for example, start to automate digital maintenance practices This could even lead to the
creation of cyber-physical systems. Future studies should elaborate on how IT can be directly integrated
with the domain ontology. Overall, this will improve system interoperability of utility data on a sector-wide
scale.
Integration within applications: Integration of the domain ontology within applications allows for easier
utilization of the ontology. This is similar to the use of IFC in applications such as AutoDesk Revit.
Possibly, the domain ontology could then be integrated as a foundational layer, which is linked to the
utility owner’s data set through the principle of linked data. However, as mentioned within the conclusions,
first the utility sector should start adopting the domain ontology. The more organizations that use the
domain ontology, the more likely that software developers will consider integration of the ontology in their
software.
Speed up adoption: Adoption requires time. To speed up adoption, extending the outreach of the domain
ontology is recommended. This can be realized through pitching the ontology’s potential by, for example,
distributing flyers, presenting at congresses, symposia, or conferences, and writing an article in magazine.
In addition, one can directly inform the end-user, such as utility owners and operators, by simply
contacting them. Especially once the domain ontology is successfully implemented in (spatial) asset
management and GIS software (referring to the recommendation about further development), it is
recommended to show by a demo how the ontology works and what its capabilities are.
In addition to the aforementioned recommendations, I also wrote two brief research assignments for either
a Bachelor or Master thesis. These research assignments touch upon the recommendations for further
development and adoption and can be found in Appendix VI.
Lastly, continuing on adoption of the ontology, I was invited by the developers of CityGML UtilityNetwork
ADE to present the domain ontology at their workshop. Unfortunately, no date has been set yet for this
workshop to take place. However, presenting the ontology at such an international stage may possibly
facilitate the uptake of the ontology.
PDEng – Final Report 53
10 Bibliography
Alexander, I. F. (2005). A Taxonomy of Stakeholders. International Journal of Technology and Human
Interaction, 1(1), 23-59.
American Society of Civil Engineers. (2002). Standard guideline for the collection and depiction of
existing subsurface utility data (38-02). American Society of Civil Engineers.
Blanchard, B., & Fabrycky, W. (2014). Systems Engineering and Analysis (5th ed.). Harlow (UK):
Pearson Education Limited.
Bowker, G., & Star, S. (1999). Sorting Things Out. Cambridge: MA: MIT Press.
Bradley, A., Li, H., Lark, R., & Dunn, S. (2016). BIM for infrastructure: An overall review and
constructor perspective. Automation in Construction, 71, 139-152.
Bray, T., Paoli, J., Sperberg-McQueen, C., & Maler, E. (2008). Extensible Markup Language (XML) 1.0
(Fifth ed.). W3C Recommendation. Retrieved from https://www.w3.org/TR/REC-xml/
British Standards Institution. (2014). PAS 128:2014 Specification for underground utility detection,
verification and location. BSI Standards Limited.
CityGML. (2016). CityGML UtilityNetwork ADE. Retrieved from
http://www.citygmlwiki.org/index.php/CityGML_UtilityNetworkADE
Corcho, O., Fernández-López, M., & Gómez-Pérez, A. (2002). Methodologies, tools and languages for
building ontologies. Where is their meeting point? Data and Knowledge Engineering, 46(1), 41-
64.
den Duijn, X. (2018). CityGML-UtiltiyNetwork-ADE. Retrieved from GitHub:
https://github.com/XanderdenDuijn/CityGML-Utility-Network-ADE
El-Diraby, T. E., & Osman, H. (2011). A domain ontology for construction concepts in urban
infrastructure products. Automation in Construction, 20(2011), 1120-1132.
European Commission. (s.d.). INSPIRE DIRECTIVE. Retrieved May 20, 2017, from INSPIRE:
http://inspire.ec.europa.eu/webarchive/
Fernández-López, M., Gómez-Pérez, A., Sierra, J., & Sierra, A. (1999). Building a chemical ontology
using Methondology and the Ontology Design Environment. IEEE Intelligent Systems, 14(1),
37-46.
Gangemi, A., Catenacci, C., Ciaramita, M., & Lehmann, J. (2005). Ontology evaluation and validation:
An integrated formal model for the quality diagnostic task. Rome, Italy: Laboratory of Applied
Ontologies - CNR.
Gasevic, D., Djuric, D., & Devedzic, V. (2009). Model Driven Engineering and Ontology Development.
New York: Springer.
PDEng – Final Report 54
Geels, F. W. (2004). From sectoral systems of innovation to socio-technical systems: insights about
dynamics and change from sociology and institutional theory. Research Policy, 33(6), 897-920.
Geonovum. (2017). IMKL2015 - Dataspecificatie Utiliteitsnetten version 1.2.
Gómez-Pérez, A. (2004). In S. Staab, & R. Studer, Handbook on Ontologies in Information Systems,
First Edition (pp. 251-274). Berlin, Germany: Springer.
Gruber, T. R. (1995). Towards principles for the design of ontologies used for knowledge sharing.
International Journal of Human-Computer Studies, 43(5), 907-928.
Grüninger, M., & Fox, M. S. (1995). Methodology for the design and evaluation of ontologies. Montreal.
Howard, R., & Björk, B. (2008). Building Information Modelling - Experts views on standardisation and
industry deployment. Advanced Engineering Informatics, 22(2), 271-280.
INCOSE. (2015). Guide for writing requirements: summary sheet. San Diego (CA), USA.
INSPIRE. (2013). D2.8.III.6 Data Specification on Utility and Government Services - Technical
Guidelines. European Commission Joint Research Centre.
Jakus, G., Milutinovic, V., Omerovic, S., & Tomazic, S. (2013). Concepts, Ontologies, and Knowledge
Representation. New York: Springer.
Kutzner, T. (2018). CityGML-UtilityNetwork-ADE. Retrieved from GitHub:
https://github.com/TatjanaKutzner/CityGML-UtilityNetwork-ADE
Noy, N., & McGuinnes, D. (2001). Ontology Development 101: A Guide to Creating Your First
Ontology. Stanford University, Knowledge Systems Laboratory, Stanford, CA.
Obrst, L., Ceusters, W., Mani, I., Ray, S., & Smith, B. (2008). The evaluation of ontologies. In C. J.
Baker, & K. Cheung, Revolutionizing Knowledge Discovery in the Life Sciences (pp. 139-158).
Berlin: Springer.
Olde Scholtenhuis, L. L., & Hartmann, T. (2015). Fieldwork-Based Method for End-User Engagement
in Domain Ontology Development. In R. Issa, & I. Mutis (Eds.), Ontology in the AEC industry:
Decade of Research and Development in Architecture, Engineering and Construction. Reston,
Virginia, USA: American Society of Civil Engineers (ASCE). doi:ISBN: 978-0-7844-1390-6
Pauwels, P., Zhang, S., & Lee, Y. (2017). Semantic web technologies in AEC industry: A literature
overview. Automation in Construction, 73(2017), 145-165.
Pinto, H., Tempich, C., & Staab, S. (2009). Ontology Engineering and Evolution in a Distributed World
Using DILIGENT. In S. Staab, & R. Studer, Handbook on Ontologies (pp. 153-176). Berlin:
Springer.
Saldaña, J. (2015). The Coding Manual for Qualitative Researchers. London: Sage.
Singapore Land Authority. (2017). Standard and Speficiations for Utility Survey in Singapore. Singapore
Land Authority.
Sparx Systems. (2017). Model Validation.
PDEng – Final Report 55
Staab, S., & Studer, R. (2009). Handbook on Ontologies (2nd edition ed.). Berlin: Springer.
Studer, R., Benjamins, R., & Fensel, D. (1998). Knowledge Engineering: Principles and methods. Data
& Knowledge Engineering, 25(1), 161-197.
Sure, Y., & Studer, R. (2002). On-To-Knowledge Methodology - Final version. Institue AIFB.
Karlsruhe: University of Karlsruhe.
Sure, Y., Staab, S., & Studer, R. (2009). Ontology Engineering Methodology. In S. Staab, & R. Studer,
Handbook on Ontologies (pp. 135-152). Berlin: Springer.
Tempich, C. (2006). Ontology Engineering and Routing in Distributed Knowledge Management
Applications. Karlsruhe: Karlsruhe University.
ter Huurne, R., & olde Scholtenhuis, L. (2018). Digitization for integration: fragmented realities in the
utility sector. ARCOM, (pp. 93-100). Belfast, UK.
Timmermans, S., & Epstein, S. (2010). A world of standards, but not a standard world: Toward a
sociology of standards and standardization. Annual Review of Sociology, 36, 69-89.
Turk, Z. (2001). Phenomenologial foundations of conceptual product modelling in architecture,
engineering and construction. Artificial Intelligence in Engineering, 2(15), 83-92.
Vrandecic, D. (2009). Ontology Evaluation. In S. Staab, & R. Studer, Handbook on Ontologies (pp.
293-313). Berlin: Springer.
Wieringa, R. J. (2014). Design Science Methodology. Springer.
Yin, R. K. (1989). Case study research, design and methods (Vol. 5). SAGE Publications.
ZoARG. (2018). ZoARG | ReDUCE. Retrieved from ZoARG: http://www.zoarg.com/zoarg-reduce-eng/
PDEng – Final Report 56
Appendix I – Overview of educational activities
Professional development courses
Course type Course name Course code ECTS Period/date followed Finished Grade
Professional development
course
Creative Thinking - 0,5 18-05-2017 Yes V
Professional development
course
Professional Effectiveness - 2 01-10-2018 – 11-12-2018 Yes V
Professional development
course
Storytelling and Media
Training
- 0,5 01-12-2017 Yes V
Professional development
course
Technical Writing and
Editing
- 2 08-05-2017 – 12-05-2017 Yes V
Professional development
course
Write to Publicize - 1 14-11-2016 – 19-12-2016 Yes V
Total 6
Compulsory courses
Course type Course name Course code ECTS Period/date followed Finished Grade
Compulsory course Building Information
Modelling and 5D
Planning
201400012 7,5 Quartile 2A 2016-2017 Yes 9
Compulsory course CS Construction
Management and
Engineering
200900128 7,5 Quartile 1A 2018-2019 Yes 8
Compulsory course Design Methodology 201400412 7,5 Quartile 2A 2017-2018 Yes 8
Compulsory course System Engineering for
PDEng
201400313 7,5 Quartile 1A 2017-2018 Yes 8
Total 30
PDEng – Final Report 57
Elective courses
Course type Course name Course code ECTS Period/date followed Finished Grade
Elective course Brand Management 201700019 5 Quartile 1A 2017-2018 Yes 8
Elective course Design for Maintenance
Operations
201500235 5 Quartile 2A 2017-2018 Yes 8
Elective course Product Life Cycle
Management
192850750 5 Quartile 2A 2016-2017 Yes 9
Total 15
Additional courses
Course type Course name Course code ECTS Period/date followed Finished Grade
Additional course Conference 201600191 5 Quartile 1A 2018-2019 Yes 8
PDEng – Final Report 58
Appendix II – Overview of meetings
Person/organization Function Date/period Activity Goal Result Influence on design
Campus and Facility
Management University
of Twente
Ray Klumpert
Alexander van
Duursen
André de Brouwer
All three are involved in
the management of the
utilities on the campus of
the university.
Oct 2016 – now Progress meeting. Align expectations from
the Campus and Facility
Management with project
progress.
Wishes from and
possibilities for the
Campus and Facility
Management have
become clearer.
Contiguous
improvements on the
design. Great
contribution to the
functional requirements.
University of Twente
Léon Olde Scholtenhuis
André Dorée
Supervisors from the
University of Twente.
Oct 2016 – now Progress meeting. Align expectations from
the University of Twente
with project progress.
Alignment to project
goals as expected by the
University of Twente.
Continuous
improvements on the
design. Ensured
alignment to the PDEng
design process and
design product criteria.
Nationaal Kabel- en
Leidingcongres
Congress to discuss the
latest developments
within the excavation
scene.
2 November 2016 Attending congress. Getting familiar with the
latest developments in
the excavation scene, and
see how these can help
me in my own PDEng
project.
New insights in especially
rules and legislation was
obtained. Both the new
IMKL2015 and
CROW500 were
presented.
Familiarity with the
IMKL2015 and
concepts, relations and
attributes of interest to
the domain ontology.
Cadaster
Fuat Akdeniz
Product and process
manager, frontrunner
KLIC-WIN.
20 January 2017 Progress meeting. Becoming familiar with
new IMKL, identifying
importance of
IMKL2015 for the
project.
Access granted to the
IMKL2015 data
repository, Cadaster
acquainted with the
project.
Familiarity with the
IMKL2015 and
concepts, relations and
attributes of interest to
the domain ontology.
Recognize
Tom Knol
Daniel Khiati
Software developers. Jan – March 2017 Developing shapefiles
for the app ‘Spying The
Underground’ for the
University of Twente.
Have a Spying the
Underground version for
the University of Twente.
A beta version of the
Spying the Underground
app is currently available.
No influence on the
design. Project was not
connected with the
PDEng project.
PDEng – Final Report 59
Siers Infraconsult
Harry Moek (director)
Jochen Tulp
Multiple employees
Involved in the
management and
registration of utility data
of multiple utility owners.
March – April 2017 Observation of the utility
data registration in
practice.
Identify how utility data
currently is registered in
practice.
A report has been written
in which the observations
and their findings are
provided, providing
direct input for the
project.
High influence on the
design. Provided insight
in the information need
from the end-user.
Directly influence on the
concepts, relations and
attributes captured in the
ontology. Uses as a basis
for the taxonomy and
vocabulary.
Delft University of
Technology
Xander den Duijn
Master graduate at the
Delft University of
Technology involved in
utility data modelling for
the municipality of
Rotterdam.
25 April 2017 First meeting. Identify how each other’s
project could contribute
to the other.
An overlap in utility
modelling was seen.
Eventually only little
influence on the design.
Work what directly
based on CityGML
UtilityNetwork ADE and
did not add anything
new.
Technical University of
Munich
Tatjana Kürzer
Member of the group
developing the CityGML
UtilityNetworkADE,
specialized in modelling
within FME.
22 May – 24 May 2017 Workshop followed on
how to model CityGML
UtilityNetworkADE in
the program FME.
Become familiar with
CityGML
UtilityNetworkADE and
the software FME.
Being able to model
utility data according to
the CityGML
UtilityNetworkADE
framework.
Moderate influence on
the design. Workshop
gave insights in how to
model XML schema files
and shapefiles through
FME.
Delft University of
Technology
Sisi Zlatanova
Associate professor at the
Delft University of
Technology, specialized
in 3D Geoinformation.
11 September 2017 First meeting. Discuss the aim of the
comparison between
state-of-the-art standards
from literature and data
structures used in
practice.
Goal and structure of the
report that is worked on
became more clear.
Possible interfaces with
other projects/persons
were identified.
High influence.
Assessment of utility
modelling practices and
standards helped in
getting a clear view on
how the domain ontology
should be structured. In
addition, it was made
explicit wat was still
missing in existing
practices and
standardization.
PDEng – Final Report 60
Siers Infraconsult
Harry Moek (director)
Jochen Tulp
Multiple employees
Involved in the
management and
registration of utility data
of multiple utility owners.
28 September 2017 Presentation of result
obtained during the study
performed at Siers
Infraconsultancy.
Inform about the
conclusions that were
made based on the data
obtained during the study
performed at Siers
Infraconsultancy.
Discuss possible future
steps to be made.
Contacts obtained from
several utility owners, to
extent the dataset.
Meeting only moderate
influence, but the
research itself had a high
influence as explained
before.
Technical University of
Munich
Tatjana Kürzer
Member of the group
developing the CityGML
UtilityNetworkADE,
specialized in modelling
within FME.
4 October 2017 Follow up from
workshop followed in
Munich.
Discuss the possible
collaboration between
my project and the work
of the CityGML Group.
My project can help
CityGML Utility
Network ADE by
providing data from the
end-user. Even more, it
can extent the Utility
Network ADE. A first
draft idea/model will be
discussed in a next
meeting.
High influence on
design. First proposal to
connect the domain
ontology to CityGML
UtilityNetwork ADE and
how they could enforce
each other.
ZoARG Symposium ZoARG (or ReDUCE) is
an initiative that aims to
improve design,
construction, installation,
and maintenance of
buried infrastructure in
shallow subsurface.
19 October 2017 Symposium organized by
the ZoARG group of the
University of Twente.
Inform the attendees
about the current project
within the ZoARG
group, as well as giving a
more general awareness
of all that has to do with
the utilities.
Several people interested
in following the project.
A mail will be send
around to ask people if
they want to participate in
the expert panel.
Moderate influence. A
small workshop was
organized in which the
participants wrote down
what must be included in
the ontology in terms of
information. Types of
information collected
was very widespread
though and fairly generic.
Landmeetdienst
Krinkels
Landmeetdienst
Campus and Facility
Management
All three parties are
involved in the
digitization of the
campus of the University
of Twente, both
underground as above
ground.
9 November 2017 First discussion about
how to align and what to
include concerning the
subsurface and surface in
a digital environment.
Clarity about mainly what
to include in the digital
environment.
New insights in how the
campus should be
digitized.
Moderate to low
influence. No direct
influence on the domain
ontology itself, but it
showed how the domain
ontology could be
integrated with a digital
PDEng – Final Report 61
environment of the
surface elements.
Cadaster
Laris Noordegraaf
Fuat Akdeniz
Caroline Groot
All three involved in the
new IMKL 2015. The
transition to this model
by the industry is a core
aspect of their current
work.
22 November 2017 First meeting. Discussion about the
possibilities to visualize
the risks by making use
of the enriched and
recently updated IMKL
(IMKL 2015). Outcome
should be a showcase to
illustrate the possibilities
of such.
The question whether or
not the University of
Twente has the capacity
at the moment to assess
this problem. Follow up
required.
Low influence. The
project was a side-project
and not part of the
PDEng project.
Technical University of
Munich
Tatjana Kürzer
Member of the group
developing the CityGML
UtilityNetworkADE,
specialized in modelling
within FME.
28 November 2017 First discussion about the
ontology and relation of
the ontology with
CityGML Utility
Network ADE.
Gaps in CityGML Utility
Network ADE and how
the domain ontology may
fill these.
Positive attitude towards
contribution of the
domain ontology within
CityGML Utility
Network ADE.
High influence.
Advantages of modelling
the domain ontology
according to CityGML
UtilityNetwork ADE
were discussed. It was
agreed upon that my
PDEng project could
support the CityGML
project and vice versa.
Bouwend Nederland
Scholingsdag
Educational training day
at Bouwend Nederland
for vocational education.
2 February 2018 Presenting the latest
developments within the
ZoARG programme.
Creating affinity between
vocational education with
our the latest
developments within our
ZoARG programme,
without scaring them
technology might
possible replace them in
the future.
Most attendants were
quite positive about our
technologies, and did see
the potential of these in
their work. Overall
people were enthusiastic
and enjoyed the
presentation.
Low influence.
Presentation was fairly
general and not
specifically for my
PDEng project only.
Presentation was not
held to discuss the
technologies in detail.
Landmeetdienst
BAM
Landmeetdienst
Campus and Facility
Management
All three parties are
involved in the
digitization of the
campus of the University
of Twente, both
15 February 2018 Second discussion about
progress made in the
digitization of the
campus.
Proper alignment of
subsurface and surface,
and clarity about the
implementation.
Possible implementation
solutions were discussed,
and one is now further
developed.
Moderate to low
influence. No direct
influence on the domain
ontology itself, but it
showed how the domain
PDEng – Final Report 62
underground as above
ground.
ontology could be
integrated with a digital
environment of the
surface elements.
Technical University of
Munich
Tatjana Kürzer
Member of the group
developing the CityGML
UtilityNetworkADE,
specialized in modelling
within FME.
23 April 2018 Meeting to discuss the
progress of CityGML
Utility Network ADE
and the domain
ontology.
Learning from new
insights retrieved from
the CityGML Utility
Network ADE model.
Alignment of the domain
ontology.
New information on
CityGML Utility
Network ADE
development. Invitation
to present work at
workshop in September /
October.
High influence. New
developments within the
CityGML UtilityNetwork
ADE were applied to the
domain ontology to keep
these aligned to each
other. Concerned mainly
the network components
class model.
Expert panel
University of Twente
Campus and Facility
Management
Kadaster
Siers
CityGML (not
present)
Group of experts from
various disciplines with
knowhow on utility data
processing and
structuring.
3 May 2018 First expert meeting. Refine the ontology’s
alignment with practice.
A list of use cases was
individually defined and
collaboratively discussed.
Feedback was received
on the main functions
and use cases of the
ontology.
High influence. Existing
list of use cases was
complemented with the
newly added ones. New
competency questions
were, thereby, developed
as well. This led to newly
added concepts, relations
and attributes in the
domain ontology.
Cadastre
Caroline Groot
Fuat Akdeniz
Bart Maessen
Magdalena Gus
All involved in the
development and
adoption of the new
IMKL2.1.
May 2018 – September
2018
External project Development of
showcases for the new
IMKL2.1 model of the
Cadastre assessing risk
analysis and visualization
during utility works.
Published report
explaining the
possibilities within the
new IMKL2.1 to analyze
and visualize risks,
supported by four
showcases.
Low influence. The
project was a side-project
and not part of the
PDEng project.
Technical University of
Munich
Tatjana Kürzer
Member of the group
developing the CityGML
UtilityNetworkADE,
specialized in modelling
within FME.
22 June 2018 Meeting to discuss the
progress of CityGML
Utility Network ADE
and the domain
ontology.
Validating the newly
added ideas in addition
to the CityGML
UtilityNetwork ADE.
New insights in how the
domain ontology can be
better aligned to
CityGML UtilityNetwork
ADE and vice versa.
High influence. New
developments within the
CityGML UtilityNetwork
ADE were applied to the
domain ontology to keep
PDEng – Final Report 63
these aligned to each
other. Concerned mainly
the network components
class model.
CROW Infradagen Congress where road
managers, contractors
and consultancies share
state-of-the-art
infrastructure
developments, both
technical and governance
related.
27 June 2018 Attending congress and
presenting paper.
Increasing outreach of
the PDEng project by
presenting the paper and
possibly making new
contacts.
Utilities were relatively
underexposed at the
congress. Still several
people showed interest in
the research.
Moderate to low
influence. Congress was a
good place to find
exposure for the domain
ontology. However, no
real new input was
gathered afterwards to be
included within the
domain ontology.
Expert panel
University of Twente
Campus and Facility
Management
Kadaster
Siers
CityGML (not
present)
Group of experts from
various disciplines with
knowhow on utility data
processing and
structuring.
28 June 2018 Second expert meeting. Assessing the
applicability of the
ontology by validating
whether the previous
established use cases can
be performed based on
data provided by the
ontology.
Missing information for
specific uses cases was
obtained. General
remarks and feedback on
the modelling of the
ontology.
High influence. Use
cases defined within the
first expert panel sessions
were validated against the
domain ontology.
Missing concepts,
relations and attributes
were collaboratively
defined. In addition,
general comments about
the class modelling of the
ontology were given.
Delft University of
Technology
Sisi Zlatanova
Associate professor at the
Delft University of
Technology, specialized
in 3D geo-information.
29 August 2018 Meeting to discuss the
utility modelling report.
Discuss the utility
modelling report written
as part of a capita selecta.
Positive feedback as well
as extra information on
standardization.
Meeting only moderate
influence, but the
research on existing
utility modelling
standards and practices
had a high influence as
explained before.
ARCOM Conference Conference by the
Association of Research
in Construction
3 Sept – 5 Sept 2018 Attending conference
and presenting paper.
Gaining feedback on my
paper, and making new
contacts within the
The paper was relatively
well received and some
insightful questions were
Moderate influence. No
new information was
provided that could be
PDEng – Final Report 64
Management, bringing
together all those
interested in construction
management research.
construction
management domain.
asked and comments
were given.
applied within the
PDEng project.
However, some useful
advices were given on
how to make the paper in
general stronger.
Technical University of
Munich
Tatjana Kürzer
Member of the group
developing the CityGML
UtilityNetworkADE,
specialized in modelling
within FME.
31 October 2018 Feedback meeting to
discuss the prototype of
the Operations and
Maintenance domain
ontology.
Receiving feedback on
the prototype of the
Operations and
Maintenance domain
ontology, to work
towards a final version.
Detailed feedback on the
data specification and the
feature catalogue.
High influence. Had a
thorough discussion on
each of the eight models
of the domain ontology.
Within each model
minor adjustments were
made, for example on
the 3D modelling of
utilities within the core
model. In addition,
network components
were changed once again
and consequently
adapted in the domain
ontology.
Technical University of
Munich
Tatjana Kürzer
Member of the group
developing the CityGML
UtilityNetworkADE,
specialized in modelling
within FME.
7 November 2018 Follow-up feedback
meeting to discuss the
prototype of the
Operations and
Maintenance domain
ontology.
Receiving feedback on
the prototype of the
Operations and
Maintenance domain
ontology, to work
towards a final version.
Detailed feedback on the
data specification and the
feature catalogue.
High influence. In
addition to the meeting
before, within this
meeting the final models
were discussed. Small
adjustments were made
in terms and definitions,
as well as in the
explanation of the hollow
space model.
Geonovum
Paul Janssen
8 November 2018 Feedback meeting to
discuss the prototype of
the Operations and
Receiving feedback on
the prototype of the
Operations and
Maintenance domain
Detailed feedback on
especially the data
specification. Feature
High influence. Data
specification was
especially reviewed from
a data modelling
PDEng – Final Report 65
Maintenance domain
ontology.
ontology, to work
towards a final version.
catalogue was studied in
less detail.
perspective. Several
associations had to be
corrected and association
roles were added to the
feature catalogue. Other
remarks concerned
definitions to be not
specific enough yet and
the fact that abstract
classes had to be in italic.
PDEng – Final Report 66
Appendix III – Collected empirical data
Discipline n System level
objects
Subsystem level
objects
Component level objects Attributes
Gas 3 Gas distribution
system
Gas low pressure
network
Pipe
Station
Connection
Shut off valve
Blowhole
Saddle
Flange
Transition
T-piece
Bend
Mantle tube
Location, status, sub net, date of installation, pressure level, material, implementation,
coating, connection type, cathodic protection, tensile resistance, length, date of
rehabilitation, accuracy, owner, operator, financial characteristic, controlled drilling
Location, sub net, name, identifier no., status, address (street, house no., postal code,
place, municipal), owner, operator, financial characteristic, date in use, date of
expiration, date of installation, type of construction, type of foundation, type of lock,
internal attributes, etc.
Location, sub net, type of appendage, date of installation, installation company,
pressure level, implementation, material, coating, connection type, status, EAN code,
manufacturer, owner, operator
Location, sub net, type of appendage, date of installation, installation company, status,
material, pressure level, implementation, diameter, brand, coating, function, no.,
preferred mode, direction of rotation, number or rotations, station no., owner, operator
Location, sub net, type of appendage, date of installation, installation company,
pressure level, diameter, implementation, construction type
Location, sub net, type of appendage, date of installation, installation company,
pressure level
Location, sub net, type of appendage, date of installation, installation company,
pressure level, implementation
Location, sub net, type of appendage, date of installation, installation company,
pressure level, material, implementation, manufacturer, coating, no. of transitions
Location, sub net, type of appendage, date of installation, installation company,
pressure level, implementation
Location, sub net, type of appendage, date of installation, installation company,
pressure level, degree
Location, sub net, type of appendage, date of installation, pressure level,
implementation
PDEng – Final Report 67
Gas high pressure
network
Duct
Pipe
Station
Connection
Shut off valve
Blowhole
Saddle
Cabinet
Flange
Transition
T-piece
Bend
Mantle tube
Insulation
piece
Location, sub net, type of appendage, date of installation, installation company,
pressure level, owner, operator, cathodic protection, status, implementation,
connection type, accuracy
Location, status, sub net, date of installation, pressure level, material, implementation,
coating, connection type, cathodic protection, tensile resistance, length, date of
rehabilitation, accuracy, owner, operator, financial characteristic, controlled drilling
Location, sub net, name, identifier no., status, address (street, house no., postal code,
place, municipal), owner, operator, financial characteristic, date in use, date of
expiration, date of installation, type of construction, type of foundation, type of lock,
internal attributes, etc.
Location, sub net, type of appendage, date of installation, installation company,
pressure level, implementation, material, coating, connection type, status, EAN code,
manufacturer, owner, operator
Location, sub net, type of appendage, date of installation, installation company, status,
material, pressure level, implementation, diameter, brand, coating, function, no.,
preferred mode, direction of rotation, number or rotations, station no., owner, operator
Location, sub net, type of appendage, date of installation, installation company,
pressure level, diameter, implementation, construction type
Location, sub net, type of appendage, date of installation, installation company,
pressure level
-
Location, sub net, type of appendage, date of installation, installation company,
pressure level, implementation
Location, sub net, type of appendage, date of installation, installation company,
pressure level, material, implementation, manufacturer, coating, no. of transitions
Location, sub net, type of appendage, date of installation, installation company,
pressure level, implementation
Location, sub net, type of appendage, date of installation, installation company,
pressure level, degree
Location, sub net, type of appendage, date of installation, installation company,
pressure level, implementation
-
PDEng – Final Report 68
Water 3 Water distribution
system
Water line Pipe
Flange
Fire hydrant
Bend
Shut off valve
Transition
Mantle tube
Location, status, sub net, function, material, external diameter, internal diameter,
nominal diameter, date of installation, status, length, water type, interior protection,
exterior protection, decommissioning, measurement type, pressure level, location
quality, owner
Location, sub net, function, status
Location, identifier no., municipality, sub net, implementation, function, rotational
direction, date of installation, status, secured, type of measurement, owner
Location, sub net, function, status, implementation, material
Location, sub net, function, implementation, identifier no., municipality, material pipe,
material, diameter, rotational direction, date of installation, status, preferred mode,
actual mode, date since actual mode, operator, measurement type, owner
Location, sub net, function, status
Location, material, diameter, length
Electricity 3 Electricity
distribution system
Low voltage line
system
Cable
Station
Connection
Street
furniture
Earthing
Charging pole
Mantle tube
Transition
Location, status, sub net, length, identifier no., status, nominal voltage, material core,
material isolation, implementation, date of manufacturing, date of installation,
installation company, accuracy, phase type, financial characteristic, owner, operator,
manufacturer
Location, name, code, area, measurements, status, date of installation, installation
company, function, implementation, owner, operator, circuit diagram, voltage
Location, address, status, function, pass through value, connection type, measurement
type, public priority, earthing checked, date of installation, installation company, level
of building, region, EAN code, owner, phase, earthing type, safety type, safety value,
safety characteristic, take off point, earthing remarks, no. of voltage errors
measurement, impedance measurement, general remarks, main cable, length of cable,
main cable type, cable sub group
Location, address, status, earthing, switch, EAN code, owner, function, date of
installation, installation company, mast no., identifier no., main cable, cable sub net
Location, sub net, type of appendage, date of installation, installation company
implementation
Location, sub net, type of appendage, date of installation, installation company
implementation
Location, sub net, type of appendage, date of installation, installation company
implementation
Location, sub net, type of appendage, date of installation, installation company
implementation, function, status, phase
PDEng – Final Report 69
Middle voltage line
system
PL cabinet
Cable
Station
Connection
Earthing
Mantle tube
Cross bonding
cabinet
TC coil
TC cabinet
Location, sub net, type of appendage, date of installation, installation company
implementation
Location, status, sub net, length, cable number, status, nominal voltage, material core,
material isolation, implementation, date of manufacturing, date of installation,
installation company, accuracy, phase type, financial characteristic, owner, operator,
manufacturer
Location, name, code, area, measurements, status, date of installation, installation
company, function, implementation, owner, operator, circuit diagram, voltage
Location, address, status, function, pass through value, connection type, measurement
type, public priority, earthing checked, date of installation, installation company, level
of building, region, EAN code, owner, phase, earthing type, safety type, safety value,
safety characteristic, take off point, earthing remarks, no. of voltage errors
measurement, impedance measurement, general remarks, main cable, length of cable,
main cable type, cable sub group
Location, sub net, type of appendage, date of installation, installation company
implementation
Location, sub net, type of appendage, date of installation, installation company
implementation
-
-
-
District
heating
3 District heating
distribution system
District heating
line
Pipe (supply
Pipe (return)
Mantle tube
Shut off valve
Transition
Data cable
T-piece
Weld
Reducer
Location, material, status, diameter, length, coordinates
Location, material, status, diameter, length, coordinates
-
-
-
-
-
-
-
PDEng – Final Report 70
Condensator
Measuring
cabinet
-
-
Telecom 3 Telecom
distribution system
Fiber line system
Coax line system
Cable
Connection
Station/center
Patch box
Handhole
PoP
Tube
Transition
Cable
Connection
Station/center
Mantle tube
Amplifier
Splitter
Location, type, status, system link, length, owner, accuracy, date of installation, date of
manufacturing, manufacturer, operator
Location, function, status, owner, date of installation, installation company
Location, function, status, address, identifier no., connector type, date of installation
Location, identifier no., function, status, date of installation, owner, operator
Location, area, function, status, data of installation, owner, operator
Location, area, function, status, data of installation, owner, operator
Location, system link, cable link, status, accuracy, owner, operator, type of cable, date
of installation, date of manufacturing, manufacturer, operator, depth
Location, function, type, date of installation
Location, type, status, system link, length, owner, accuracy, date of installation, date of
manufacturing, manufacturer, operator
Location, function, status, owner, date of installation, installation company
Location, function, status, address, identifier type, connector type, date of installation
Location, function, status, sleeve type, date of installation, supply point
-
-
PDEng – Final Report 71
Appendix IV – Overview of existing standards
# Standard Domain of application Domain Explanation
1 ArcGIS Utility
Networks
Global Utility sector Specific component users of ArcGIS use when managing utility and telecom networks.
Covers electric, gas, water, storm water, wastewater, and telecommunication
Designed to model all components that make up the network, such as wires, pipes, valves,
zones, devices, circuits
Real world behavior in a model
2 ASCE 38-02 United States Utility sector Primarily focuses on classifying the quality of existing utility data.
Collection and depiction of existing subsurface utility data
Classifying quality of existing subsurface utility data
Applies to the disciplines of electricity, gas, oil, crude, water, thermal, telecommunication,
sewage or any other similar commodity.
Provides project owner, engineer and construction information to develop risk reduction
strategies
3 AS 5488 Australia Utility sector Standard providing a framework for the classification of subsurface utility location and attributes information
in terms of specified quality levels. The standard does not apply to above ground infrastructure, such as
overhead wires.
Involves the utilities and associated subsurface features that facilitate location and identification
of the subsurface utility infrastructure.
4 CityGML
UtilityNetworkADE
Global Utility sector Develops a homogenized 3D network model for multi-utility failure simulation including the relevant
thematic attribution.
Application independent geospatial information model
Comprises different thematic areas such as buildings, vegetation, water, terrain, tunnels and
bridges.
Extension of CityGML providing the possibility to represent supply and disposal networks in
3D city models
5 IEC CIM Global Power utilities Global standard for power utilities, covering substation, transmission and distribution.
Extended to gas and water
Defines system architecture, API, data models, security protocols, meter reading and functional
safety
6 IMKL The Netherlands /
Belgium
Utility sector A standardized data specification on a national level intended to communicate utility data in case of
excavation works. Used through a central platform.
Provides a taxonomy and vocabulary for utility data
Is coupled with the INSPIRE model
7 INSPIRE EU Including utilities Developing a common information model for spatial data throughout Europe.
Covers electric, oil, gas, chemicals, sewage, telecommunications and water
PDEng – Final Report 72
Has a specific Utility Network application schema
8 LandInfra/InfraGML Global Land and infrastructure OGC approved standard (OGC 15-111r1) that captures the land and the infrastructure facilities build upon
the land.
Its encoding standard, InfraGML marks a significant, pragmatic realization in the often idealized
integration of GIS/CAD/BIM
Builds upon and extends GML3.2 and GML3.3
Provides framework for addressing the full facility life cycle
9 PAS128 UK Utility sector Primarily focuses on classifying the quality of existing utility data. Set out clear and unambiguous provisions
to those engaged in the detection, verification and location of active, abandoned, redundant or unknown
utilities.
Not object-based
Applies to disciplines of drainage, gas, water, electricity and telecommunication.
10 PAS256 UK Utility sector Developed to improve the quality, accuracy and reliability of asset owner’s records.
Drives towards improved accuracy of data and facilitates electronic sharing
Linkage between critical infrastructure assets and other initiatives for digital built Britain (e.g.
Smart Cities, BIM)
Includes rules for information exchange, locational accuracy, number of days to make
information reliable, unidentified buried objects, wrongly marked objects, and safety and
protection
11 Pipeline ML Global Oil and gas (midstream)
pipelines
An open industry data standard designed to help midstream pipeline operators to share information about
their assets and the various construction and maintenance activities performed on them.
Ongoing OGC standards development initiative
The need for Traceable, Verifiable and Complete (TVC) asset management data.
Open and vendor-neutral data interchange standard to enable the interoperable interchange of
pipeline data between parties, disparate systems and software applications
Limit loss of accuracy, density or data resolution; without ambiguity, without need for
conversion
12 SEDRIS Global Environment Offers a data representation model to articulate environmental data. Put simply, it is an infrastructure
technology providing the foundation for IT applications to express, understand, share and reuse
environmental data.
Represents environmental data
Facilitates the exchange of environmental data sets
Data representation aspects is about capturing and communication meaning and semantics
Representation aspect much like a language to unambiguously describe the environment
13 Utility Survey Standard Singapore Utility sector Standard providing guidelines on data capturing to improve reliability of utility data.
Aim to create a common understanding of the acquisition and production process of utility data
Provides a survey template
PDEng – Final Report 73
Appendix V – Expert panel sessions handouts (Dutch)
Handout expert meeting – May
Doel van vandaag
Vandaag staat in het teken van het evalueren van
versie 1.0.
De huidige versie van de ontologie is ontwikkeld
samen met het Facilitair bedrijf van de
Universiteit van Twente.
Doel van vandaag is om deze versie een breder
draagvlak te geven, door overeenstemming te
bereiken over (1) de inhoud en (2) de structuur
van de ontologie.
De evaluatie vindt vandaag plaats in drie rondes.
Elke van deze rondes adresseert een specifiek
thema, hierna nader besproken.
Agenda
13:00
16:30
15:30
14:20
14:35
15:45
Opening
Afsluiting
Voorstelronde
Tweede evaluatieronde
Presentatie
Derde evaluatieronde
Doornemen agenda
Resumé
13:05
13:10
13:15
Pauze
Pauze
16:15
Eerste evaluatieronde13:35
PDEng – Final Report 74
Eerste evaluatieronde – Aansluiting op de praktijk
Overzicht eerste evaluatieronde
Omschrijving
Doel Het bepalen van de use cases en hoofdfuncties van de ontologie.
Activiteit 1 Brainstormen over de use cases.
Activiteit 2 Brainstormen over de functies.
Resultaat (1) Een lijst van use cases, en (2) een set van (hoofd)functies die de ontologie moet omvatten.
Tweede evaluatieronde –Structuur en het level van detail
Overzicht derde evaluatieronde
Omschrijving
Doel Het bepalen van het de structuur en level van detail van de ontologie.
Activiteit 1 Brainstormen over de structuur.
Activiteit 2 Brainstormen over het level van detail.
Resultaat Set van eisen aan de ontologie betreffende (1) de structuur en (2) het level van detail.
Derde evaluatieronde – Integratie met bestaande modellen
Overzicht tweede evaluatieronde
Omschrijving
Doel Het afstemmen van de structuur van de ontologie met die van bestaande modellen, zijnde het IMKL en
INSPIRE.
Activiteit 1 Brainstormen over de integratie met het IMKL.
Activiteit 2 Brainstormen over integratie met INSPIRE.
Resultaat Set van eisen aan de ontologie betreffende integratie met (1) het IMKL en (2) INSPIRE.
Resultaat van vandaag
Lijst van use cases voor de ontologie;
Set van hoofdfuncties die de ontologie moet omvatten;
Set van eisen aan de structuur van de ontologie;
Set van eisen aan het level van detail van de ontologie;
Set van eisen aan de ontologie betreffende integratie met het IMKL;
Set van eisen aan de ontologie betreffende integratie met INSPIRE.
Volgende expert meetings
Overzicht aankomende expert meetings
Datum Tijd Locatie
Meeting 2 Juni 28, 2018 13:00 – 16:30 Universiteit Twente
Paviljoen, kamer 9
Meeting 3 September 13, 2018 13:00 – 16:30 Universiteit Twente
Paviljoen, kamer 9
PDEng – Final Report 75
Handout expert meeting – June
Doel van vandaag
Vandaag staat in het teken van het evalueren van
de domein ontologie.
De huidige versie is verder ontwikkeld aan de
hand van de verkregen feedback uit de eerste
expert meeting (25 mei jl.), in samenwerking met
het Facilitair Bedrijf van de Universiteit van
Twente.
Doel van vandaag is allereerst om te bespreken
in hoeverre de ontologie aan de hand van de
feedback van de vorige meeting is aangepast.
Vervolgens zal in twee evaluatierondes, hierna
nader besproken, de content en structuur van de
ontologie geëvalueerd worden, aan de hand van
enkele use cases.
Agenda
13:00
16:30
15:00
13:45
14:00
15:15
Opening
Afsluiting
Voorstelronde
Eerste evaluatieronde
Terugkoppeling
Tweede evaluatieronde
Doornemen agenda
Resumé
13:05
13:10
13:15
Pauze
Pauze
16:15
jan
feb
mrt
apr
mei
jun
jul
aug
sept
okt
PDEng – Final Report 76
Eerste evaluatieronde – Objecten en attributen
Tabel 1 – Overzicht eerste evaluatieronde
Omschrijving
Doel Het bepalen van de nodige objecten en attributen die de ontologie moet omvatten.
Activiteit 1 Het opstellen van een eigen ontologie.
Activiteit 2 Analyseren van type objecten en attributen genoemd binnen de ontologie.
Resultaat Een groslijst van (1) objecten, met daaraan gekoppeld (2) een set van attributen.
Tweede evaluatieronde –Toepasbaarheid
Tabel 2 – Overzicht derde evaluatieronde
Omschrijving
Doel Het bepalen van de toepasbaarheid van de ontologie aan de hand van use cases.
Activiteit 1 Lijst van aansprekende use cases, waaraan de ontologie getoetst wordt, opstellen.
Activiteit 2 Analyseren van de ontologie op haar toepasbaarheid, met inachtneming van de volgende aspecten:
Content – objecten en attributen
Structuur – logica
Resultaat Lijst van tips / aandachtspunten betreffende (1) de content en structuur, en (2) de toepasbaarheid.
Resultaat van vandaag
Lijst van algemene objecten die de ontologie moet omvatten;
Lijst van algemene attributen die de ontologie, voor de betreffende objecten, moet omvatten;
Lijst van tips / aandachtspunten aan de ontologie betreffende content en structuur;
Lijst van tips / aandachtspunten aan de ontologie betreffende toepasbaarheid.
Volgende expert meeting
Tabel 3 – Overzicht aankomende expert meeting
Datum Tijd Locatie
Meeting 3 September 13, 2018 13:00 – 16:30 Universiteit Twente
Paviljoen, kamer 9
PDEng – Final Report 77
Handout expert meeting – September (plenary session cancelled)
Doel van vandaag
Vandaag staat in het teken van het evalueren van
versie 3.0. Aan de hand van de verkregen
feedback uit deze meeting zal de definitieve
versie worden opgesteld.
De huidige versie is verder ontwikkeld aan de
hand van de verkregen feedback uit de tweede
expert meeting, in samenwerking met het
Facilitair Bedrijf van de Universiteit van Twente.
Doel van vandaag is allereerst om te evalueren of
de aangebracht wijzigingen voldoen. Vervolgens
vind een evaluatieronde plaats, hierna nader
uitgelegd.
Tot slot wordt de adoptie en implementatie van
de ontologie kort besproken.
Agenda
13:00
16:30
15:15
14:00
14:15
15:30
Opening
Afsluiting
Voorstelronde
Evaluatie
Terugkoppeling
Implementatie en adoptie
Doornemen agenda
Resumé
13:05
13:10
13:15
Pauze
Pauze
16:15
jan
feb
mrt
apr
mei
jun
jul
aug
sept
okt
PDEng – Final Report 78
Eerste evaluatieronde – Knelpunten
Tabel 2 – Overzicht eerste evaluatieronde
Omschrijving
Doel Het bepalen van de nog aanwezige knelpunten in de ontologie, zowel in de content als in de structuur.
Activiteit 1 Brainstormen over de knelpunten in de content.
Activiteit 2 Brainstormen over de knelpunten in de structuur.
Resultaat Set van knelpunten aan de ontologie betreffende (1) content en (2) structuur.
Tweede evaluatieronde –Implementatie en adoptie
Tabel 2 – Overzicht derde evaluatieronde
Omschrijving
Doel Het bepalen van de nodige vervolgstappen om implementatie en adoptie van de ontologie te stimuleren.
Activiteit 1 Brainstormen over de mogelijke bottlenecks omwille implementatie.
Activiteit 2 Brainstormen over de mogelijke bottlenecks omwille adoptie.
Resultaat Set van eisen aan de ontologie betreffende (1) implementatie en (2) adoptie.
Resultaat van vandaag
Set van knelpunten aan de ontologie betreffende content;
Set van knelpunten aan de ontologie betreffende structuur;
Set van eisen aan de ontologie betreffende implementatie;
Set van eisen aan de ontologie betreffende adoptie.
Vooruitblik
Aan de hand van de feedback verkregen uit deze derde en laatste expert meeting, zal de definitieve
ontologie gevormd worden. Na de vorming van de ontologie, staan er twee belangrijke aspecten te
wachten, zijnde implementatie en adoptie.
Implementatie
De definitieve ontologie wordt geïmplementeerd bij het Facilitair Bedrijf van de Universiteit van Twente.
Er zal getest worden met het inladen van verscheidene datasets voor een tal van disciplines. Tevens zal er
getest worden met het importeren en exporteren van de data.
Adoptie
Het creëren van draagvlak is essentieel voor het voortbestaan van de ontologie. Aansluiting van de
ontologie met de datamodellen binnen het IMKL en INSPIRE draagt daar aan bij. Verder vergroot de
connectie met CityGML UtilityNetwork ADE, maar ook met ZoARG, de bekendheid van het model.
PDEng – Final Report 79
Handout expert meeting – September (template individual sessions)
Introductie
Dit document dient als hand-out voor de derde expert sessie ten behoeve van de evaluatie van de domein
ontologie voor ondergrondse infrastructuur. Deze ontologie wordt opgesteld vanuit een initiatief aan de
Universiteit van Twente. In onderstaande paragrafen wordt de hand-out van deze sessie uiteengezet.
Uitleg bij documenten
Voor deze derde expert sessie zijn in totaal drie documenten van belang. Elk van deze wordt hieronder
kort toegelicht.
[1] Hand-out expert sessie
De hand-out voor de expert sessie ligt voor u. Dit document dient als doel het evalueren van de domein
ontologie voor ondergrondse infrastructuur. In hoofdstuk 2 van dit document wordt uiteengezet op welke
criteria geëvalueerd wordt. De documenten die geëvalueerd worden betreffen de volgende twee.
[2] Operations and Maintenance ADE – Data Specification
De ‘Operations and Maintenance ADE – Data Specification’ is het hoofd-document en beschrijft en
presenteert de ontologie aan de hand van verscheidene class-modelling (UML) diagrammen. Deze
diagrammen zijn ook in de voorgaande expert sessie met u besproken. Op basis van de toen verkregen
feedback zijn de UML diagrammen aangepast en is het aangevuld met een rapportage, resulterende in de
data specificatie. De ontologie wordt middels een achttal modellen en de bijbehorende UML diagrammen
beschreven in de data specificatie. Voor elk van de modellen is een uitleg aanwezig in de data specificatie.
[3] Operations and Maintenance ADE – Feature Catalogue
De ‘Operations and Maintenance ADE – Feature Catalogue’ is de catalogus behorende bij de data
specificatie. In deze catalogus zijn alle concepten en relaties opgenomen die deel uitmaken van de UML
diagrammen. De catalogus biedt daarmee ondersteuning in het geval van onduidelijkheid van gebruikte
termen en definities in de UML diagrammen. De catalogus volgt de volgorde van de acht modellen en
bijbehorende UML diagrammen beschreven in de data specificatie.
Doel
Het doel van deze derde en laatste expert sessie is het evalueren van de ontologie op basis van criteria
beschreven in hoofdstuk 2. De evaluatie binnen deze sessie is met name gericht op de functionaliteit en
bruikbaarheid van de ontologie. Evaluatie richt zicht daarbij met name op de vocabulaire (gebruikte
termen en definities binnen de ontologie) en de taxonomie (toegepaste structuur en opbouw binnen de
ontologie) van de ontologie. De ontologie is reeds geëvalueerd op logica (aangaande de UML modelleer
taal) binnen het softwarepakket waarin hij is opgesteld.
PDEng – Final Report 80
Beoogd resultaat
Op basis van de verkregen feedback uit deze sessie zal de definitieve versie van de ontologie opgesteld
worden. Het beoogde resultaat is daarmee het verkrijgen van feedback op de beide documenten van de
‘Operations and Maintenance ADE’. Hierbij draait het met name om feedback op de acht UML
diagrammen en bijbehorende catalogus. Feedback op de rapportage in het algemeen wordt gewaardeerd
en meegenomen, maar dient niet het hoofddoel van deze expert sessie.
Vooruitblik
De definitieve versie die volgt uit de feedback uit deze derde expert sessie zal (deels) worden
geïmplementeerd bij het Facilitaire Bedrijf van de Universiteit van Twente. Er zal getest worden met het
inladen van bestaande netinformatie in een GIS georiënteerde omgeving. Mogelijke fouten in de ontologie
die daarbij aan het licht komen worden nog gecorrigeerd, alvorens de ontologie gepubliceerd kan worden.
Verdere details over publicatie zijn nog niet bekend.
PDEng – Final Report 81
Evaluatie
Binnen deze expert sessie wordt de ontologie geëvalueerd aan de hand van een zestal criteria:
[1.] Aanpassingsvermogen:
Is de conceptuele basis van de ontologie voldoende voor het scala aan verwachte taken die zij moet kunnen? Kan de ontologie daar waar nodig uitgebreid
en gespecialiseerd worden door haar gebruikers?
[2.] Beknoptheid:
Bevat de ontologie irrelevante concepten of relaties betreffende het domein van beheer en onderhoud van ondergrondse infrastructuur? Zijn er overbodige
concepten en beschrijft de ontologie puur de essentiële zaken van het domein?
[3.] Consistentie:
Zijn er tegenstrijdigheden in de ontologie? Is de beschrijving van de concepten en relaties binnen de ontologie consistent?
[4.] Compleetheid:
Bevat de ontologie alle relevante concepten en relaties voor het domein van beheer en onderhoud van ondergrondse infrastructuur?
[5.] Helderheid:
Wordt de bedoelde betekenis van de gebruikte termen en definities in de ontologie effectief gecommuniceerd? Zijn de gebruikte termen en definities
helder beschreven en is daarmee de ontologie begrijpelijk?
[6.] Nauwkeurigheid:
Legt en representeert de ontologie de correcte aspecten van de ‘echte’ wereld vast? Voldoet de ontologie aan de deskundigheid van haar gebruikers?
Doel is om de acht UML diagrammen en bijbehorende catalogus langs te gaan en te evalueren hoe aan elk van de criteria wordt voldaan. Mochten er zich
onduidelijkheden of fouten voordoen in het model, kan dat beschreven worden in de tabel die voor elk van acht de modellen is opgesteld op de volgende pagina’s.
Als voorbeeld is onderstaande gegeven voor criteria ‘Compleetheid’. Voor elk model is tevens ruimte gelaten op de volgende pagina’s voor overige opmerkingen
op de stippellijnen onder de tabel. De volgende pagina’s volgen de volgorde van de UML diagrammen zoals in de data specificatie en catalogus.
# Criteria Opmerking
1 Compleetheid Voor het uitvoeren van activiteit ‘X’ mist het object ‘Y’ met attribuut ‘Z’ in het UML diagram.
PDEng – Final Report 82
Appendix VI – Possible BSc / MSc research subjects
[1] Treatment evaluation of the Operations and Maintenance conceptual schema
The Operations and Maintenance conceptual schema is not fully implemented yet in (spatial) asset
management and GIS software. In order to do so, the student continues with the implementation of the
conceptual schema at the Campus and Facility Management of the
University of Twente. Implementation is done in QGIS by making use
of the FME software. In addition to implementation only, the student
evaluates the conceptual schema through querying (by for example using
the software pgRouting) against its use cases. Outcome of the student’s
work is one utility discipline fully mapped to the conceptual schema’s
format. Also, the student shows whether the conceptual schema is
capable of delivering the information required for operations and
maintenance. In general, this research includes the following activities:
Mapping utility data of one utility discipline according to the Operations and Maintenance conceptual
schema through FME.
Validating the Operations and Maintenance conceptual schema through computer-interpretable
queries, for example, by making use of the pgRouting software.
Academic level BSc / MSc
Required background Civil engineering, geo-information science
Host organization Campus and Facility Management University of Twente
[2] Visualization of utility information from a socio-technical perspective
Mapping utility data through the Operations and Maintenance conceptual schema does result in a GIS
database structure. However, questions remain on how to visualize the utility data, but foremost, on which
utility data to be visualized. Therefore, the student focuses on the following two elements: (1) analysis on
how to visualize spatial data in geospatial software, and (2) analysis of which spatial utility data to be
visualized. Whereas the former is a relatively technical questions, the latter touches on the socio-technical
perspective. The visualization needs of utility data from the
end-user perspective receive only little attention in literature.
In this context, an end-user may be an utility owner, utility
operator or contractor. The student, thereby, contributes to
literature by providing new insights in visualization of
geospatial information, in specific, utilities. In general, this
research includes the following activities:
Analysis of which utility data to be visualized from the socio-technical, and thereby, end-user
perspective.
Analysis on how to visualize utility data in geospatial software, including its technical implications.
Academic level MSc
Required background Civil engineering, geo-information science
Host organization University, utility owner, utility operator
PDEng – Final Report 83
Appendix VII – Publications
Title Digitization for Integration: Fragmented Realities in the Utility Sector
Author Ramon ter Huurne, Léon Olde Scholtenhuis
Type Conference paper
Conference ARCOM 2018 Conference, Belfast, UK, 3-5 September 2018
Published Yes
Presented Yes
Language English
DIGITIZATION FOR INTEGRATION: FRAGMENTED
REALITIES IN THE UTILITY SECTOR
The construction industry and its reform agendas commonly assume that digitization of a
construction asset's life cycle also integrates its stakeholders. Behind this lies the premise that
stakeholders reduce ambiguity and create consistency by using software that operates on the
basis of shared and uniform knowledge. To explore this premise, this study identified the
knowledge bases - data standards and modelling protocols for engineering software - that
distinctive underground infrastructure owners use. To this end, we analysed a utility
engineering consultancy that registers and processes asset data of twelve major utility owners.
We observed their utility information managers and studied their asset management guidelines.
We used two utility taxonomies from literature to compare identified digital modelling
standards. Subsequently, we used literature about modelling standards in digital practices to
argue how selected examples of divergent digital models hamper uniformity. We conclude that
digital reality models may also differ and thus confuse, fragment, and ultimately delimit
collaborative digital practices. This insight stresses the relevance of defining shared domain
understanding to facilitate the uptake of software for collaborative engineering practices. It
stimulates construction improvement agents to consider this important notion of shared digital
realities in their debates about achieving integration by ‘going digital'.
Keywords: digitization, fragmentation, integration, standardization, utility sector.
INTRODUCTION
Building Information Modelling (BIM) advancements drive state-of-the-art engineering and
problem-solving in the construction industry. Its implementation is a much-discussed topic in
literature and policy documentation (e.g. Bradley et al. 2016; Lu et al. 2015; Pauwels et al.
2016). One of the main steps in converting to this 'BIM paradigm' is through digitization,
which involves the development of 'digital twins' of construction assets. A general belief in
the construction industry - for example, visible in industry reform agendas - is that
PDEng – Final Report 84
digitization of the information relevant to the construction asset's life cycle further integrates
stakeholders. The argument follows the implicit logic that ambiguity will decrease and
consistency between exchanged information will increase when stakeholders all accept and
implement one uniform knowledge base for their (BIM) software. Taking this for granted,
however, may cause 'digitization hubris'.
The problem with this ideal perception that digitization stimulates integration of the
fragmented supply chains in the construction industry is that it ignores that digital practices
themselves are also fragmented. Software interoperability and information integration issues
(Lu et al. 2015) complicate this integration of fragmented practices. Turk (2001) describes a
typical way in which construction-IT developers capture practitioners' realities. He argues that
standardized data formats and structures help to achieve integration under the condition that
these are accepted and represent practitioners' shared perceptions of realities. In addition,
Gustavsson et al. (2012) and Samuelson and Björk (2011) indicate that a lack of consistency
in and acceptance of the adoption of standards creates a barrier to the realization of
integration.
The lack of consistency in and acceptance of the adoption of standards has led to the
phenomenon of, drawing on Timmermans and Epstein (2010), "a world of standards, but not a
standard world". This phenomenon is also typical for the architecture, engineering and
construction (AEC) industry, where nations, organizations and even individuals have been
developing digital (BIM) standards in a rather fragmented and self-centred manner (Azhar
2011).
Although existing literature argues that consistency in and acceptance of standards is a
prerequisite in achieving integration through digitization, this suggests that a professional
paradigm is able to develop a unified and accepted standard. This study verifies whether such
a unified and accepted standard is developed in practice. In specific, it explores the
digitization practices in the utility sector to understand the extent to which digitization leads
to uniform digital practices. We identified and assessed digital standardization practices in a
utility engineering consultancy and show that the industry currently uses various standards to
model the same utility asset information. Consequently, we argue that implementation of
digital practices may lead to digital information but that this does not necessarily result in a
common and accepted set of uniform practices. We show that the existence of different digital
realities hampers the integration of stakeholders.
The remainder of this paper is organized as follows: we first describe related literature about
digitization and standardization. Second, we present how we used a case study to compare
standards in practice. Third, we present findings by elaborating on selected examples of
differing standard descriptions of an asset object. We compare the findings to literature before
we draw our conclusions.
RELATED LITERATURE
The use of technology in construction projects can improve the inter-organizational
communication, cooperation and coordination (Adriaanse et al. 2010; Peansupap and Walker
2005). In the process of using such technologies, the construction industry digitizes
construction assets by defining concepts, attributes, and their relations (El-Diraby and Osman
2011). Digitizing asset data is realized through applications like BIM (Bradley et al. 2016).
Methodologies should be developed to facilitate exchange and communication between such
applications in order to streamline work flows. Currently, this happens through the
development of semantic web ontologies (Pauwels et al. 2016) and asset data modelling
PDEng – Final Report 85
standards such as the Industry Foundation Classes (IFC) for BIM software (Turk 2001) and
CityGML for geospatial software (Open Geospatial Consortium).
Standardization is critical to facilitate communication between different stakeholders in a
fragmented industry - like the architecture, engineering and construction (AEC) industry
(Howard and Björk 2008). Standards for representing construction data, such as IFC and
CityGML are gradually accepted and used as the predominant way to exchange data between
engineering software. However, common data formats for major types of infrastructure
projects, such as transport, utilities or environmental projects (Bradley et al. 2016) are less
developed.
The aim of standardization, in general, is to capture realities and construct uniformities across
cultures, time and geography on the basis of agreed-upon rules (Timmermans and Epstein
2010). These agreed-upon rules are thereby captured in standards, which, on their turn,
emerge from many fields (Bowker and Star 1999; Timmermans and Epstein 2010). The types
of standard relevant to this study, are digital modelling standards. Such standards are also
referred to as 'ontologies' (El-Diraby and Osman 2011). In computer science, ontologies are
used in the process of knowledge elicitation - i.e. the process and output of reality modelling.
In this process, knowledge is modelled in artefacts such as domain ontologies. Ontologies
describe the world as seen by a group of people at a certain time according to a school of
thought that is based on a set of fundamental propositions or world views (El-Diraby and
Osman 2011). Once adopted and shared amongst practitioners, they represent domain
knowledge in a unified, simplified, and consistent way.
Capturing realities through digital modelling standards - i.e. ontologies - for the fields in
which digitization emerges requires thought about phenomenology - a branch of philosophy
that deals with how to take things for what they are and what it means 'to be' - and
hermeneutics - a branch of philosophy focussing on interpretation. Intention and interpretation
are relevant when capturing realities, because their meanings can be shaped both by the
authors and users of standards. Once explicated in a textual form realities can, therefore, have
several plausible interpretations (Turk 2001). Likewise, Lampland and Star (2009) argue that
it is not surprising that standards have many possible antonyms, given the range of possible
meanings packed in a term.
When digital modelling standards are not adopted in digital practices, this may hamper
construction IT adoption. In their adapted model of IT adoption - based on the unified theory
of acceptance and use of technology, the theory of planned behavior, and the technology
acceptance model - Adriaanse et al. (2010), have identified that differences in working
practices, resources and objectives hamper successful adoption of IT. The existence of
differing work practices also impacts digital modelling practices. Co-existence of multiple
practices, may, for example also result in situations where practitioners use distinctive digital
modelling standards to model an asset. As a result, adopted IT may use distinctive knowledge
bases, and therefore show little uniformity in knowledge representation. Therefore, we
hypothesize that using IT, and developing standardizations, does not automatically enable
adoption and integration of IT users. The different digital realities can explain why adoption
and use of IT in AEC industry are not always as effective and efficient as they could
potentially be (Hjelt and Björk 2006; Sulankivi 2004).
To date, literature seems to have focussed on proposing a standard for a certain domain (e.g.
El-Diraby and Osman 2011) or argue that standards should be developed to achieve
integration (e.g. Gustavsson et al. 2012; Samuelson and Björk 2011; Turk 2001). It does not,
however, focus on the multiple standards - and as a consequence realities - that co-exist as a
result of varying work practices. Limited attention is given to the impact of different realities
PDEng – Final Report 86
on the successful adoption and integration of digital modelling practices. In light of shit
research gap, this study observes the use of standards and IT in the AEC industry and seeks to
provide the first evidence for our hypothesis.
RESEARCH METHOD
Different standards of reality, as a result of varying work practices, have received limited
attention in literature. Therefore, we considered an exploratory research approach most
appropriate. We conducted a qualitative case study to collect data and gain insights in the
topic investigated (Yin 2014). In this case study, we identified and compared the different
standards used in the construction domain. More specifically, we decided to focus on a
domain in which BIM technologies and standards are not adopted on large scale. The utility
sector is a good example of such a domain and was therefore selected. We studied how
standards and data protocols have been used in a Dutch utility engineering consultancy
company. This company registers and processes utility data of twelve major utility owners,
being our unit of analysis. The twelve utility owners cover the disciplines of water, gas,
electric, telecom and district heating, where several utility owners represented multiple
disciplines. As such, each utility discipline was covered by three utility owners.
The data collection approaches used in our case study include observations of work practices
and study of asset management guidelines. To this end, we observed seven information
managers whose daily task was to model underground networks. During the observations, the
information managers showed and explained how utility owners store their realities in digital
environments. While keeping notes on this, we had dialogues in which they explained their
asset management work routines.
We also collected asset management guidelines from six utility owners that were used by the
information managers as a guideline to verify data that surveyors and engineers modelled. We
found international, national, and organizational standards. Asset management guidelines
from the other utility owners were either not available or made accessible due to privacy
reasons.
The elicited realities from practice were analysed in a two-stage process: (1) qualitative
coding to abstract and compare the standards and guidelines, and (2) identifying differences
or similarities between the elicited realities.
To abstract and compare the standards and guidelines, we used qualitative coding (Saldaña
2015). El-Diraby and Osman's (2011) typology of infrastructure objects was used to compare
the various concepts and attributes in the standards that we found. This typology distinguishes
between component level, subsystem level, and system level objects and captures, for
example, spatial, dimensional, and material attributes. To this end, we identified how various
asset owners of underground infrastructures describe and store their ‘realities’ in data
standards and protocols.
After assigning objects to these various levels and attributes, our next step was to identify
differences or similarities between modelled realities. For this, we used a typology from
Gasevic et al. (2009) that essentially describes three elements of an ontology: (1) taxonomy
and hierarchy, (2) vocabulary, terms and names, and (3) semantics: the linguistic meaning of
the representation.
After identifying similarities and differences, we selected examples to explain how standards,
albeit digitized, are still fragmented. This serves as first evidence for our hypothesis that using
digital systems, and developing standardizations, does not automatically enable adoption and
integration of IT users.
PDEng – Final Report 87
FINDINGS
Our findings show that modelled realities used by information modelers were represented in:
(1) international, (2) national and (3) organizational standards. The case study, both through
observations of the work practices and study of asset management guidelines, reveals that, for
example, each utility owner prescribed her own asset data registration standard, based on their
own work practices. According to the modelers, the organizational standards were used more
frequently than the international and national standards. Since most examples follow from
these organizational standards, these will be elaborated on below.
The modelled realities represented in the organizational standards describe a comprehensive
set of utility asset data. To this end, we found around twenty to forty infrastructure objects
within a single reality, dependent on the type of utility discipline. Captured realities within the
telecom discipline, for example, covered around twenty objects, whereas realities from the gas
discipline covered up to forty objects. Most of the infrastructure objects found belong to the
component level of El-Diraby and Osman's (2011) typology of infrastructure objects. We also
found extensive lists of attribute information for the infrastructure objects at hand. Within
elicited realities from the electricity domain, for example, up to fifteen attributes for the
component level object 'kabel' (cable) were found. Within this example, spatial and
dimensional attributes were, amongst others, included. A comparable number of attributes
was found for other infrastructure objects.
When comparing the abstracted realities, we found that in general similar objects and
attributes were captured in the standards. Standards include comparable component,
subsystem and system level objects. This finding also applies to the type of attributes
captured. However, a more detailed comparison between these standards - i.e. realities -
shows differences. Following the typology of Gasevic et al. (2009), three illustrative
differences are presented in more detail, while briefly noting others:
First, one example of a taxonomical difference is visible when comparing three standards of
the telecom discipline. These standards use subsystem level and system level objects to model
their telecom network in the following two ways. First, two standards use subsystem level
objects to model their coaxial and fiber network as individual subsystems. Their main telecom
network, including both the coaxial and telecom network, is modelled on top of the taxonomy
at the system level. Second, one standard models both the coaxial and fiber network on the
subsystem level as one main telecom network. The particular network type here is specified
as an attribute of the subsystem object, i.e. the main network.
Another taxonomical difference, for example, is visible within the modelling of the object
'station' (station) when comparing three standards of the gas discipline. Where two standards
model a station as one single component level object and define its type (internal, external) at
the attribute level, one other standard defines the type of station already at the component
level object. In total, we found around six taxonomical differences.
Second, an example of the way in which realities are modelled with a different vocabulary is
the similar use of the words 'meting' (measurement) and 'nauwkeurigheid' (accuracy) when
comparing two standards of the electricity discipline. Both standards use these words to
describe the type of measurement of coordinates (either analogous or digital) in order to
estimate how accurate the coordinates of component level objects are.
Another notable difference in use of vocabulary, for example, is the similar use of the words
'ligging' (area) and 'locatie' (location) to describe the x and y coordinates of a component level
object. Differences in use of vocabulary were found to be most frequent, compared to the
PDEng – Final Report 88
other types of differences in ontological representation. In total, we found around fifteen
differences of the way in which realities are modelled with a different vocabulary.
Third, a semantic difference in the elicited realities is the ambiguous meaning of the word
'uitvoering' (implementation). The word is used to describe various characteristics of objects
on mainly the component level. Six standards across the telecom, gas and electricity
discipline, show the use of the word to describe divergent characteristics such as diameter,
manufacturer, material type, and installation type. This difference was even found in
individual realities of the gas discipline. In addition, in three standards of the electricity
discipline, the word is used to describe multiple characteristics at once, including
manufacturer, material and diameter.
Other examples of semantic differences include the ambiguous meaning of both the words
'sort' (soort) and 'type' (type). Similar to 'uitvoering' (implementation), these words are used to
describe various characteristics of objects, being mainly the installation type and material
type. In total, we found around eight examples of semantic differences.
Findings show that assets were modelled in standards for international, national and
organizational communities. There seems to be a favored use of organizational standards
defined by utility owners, rather than international and national standards. Moreover, the
results show differences in ontological representations, since the elicited realities differ in
their structure, use of terms and linguistic meaning of their representation.
DISCUSSION
Many organizations nowadays digitize their assets (e.g. Bradley et al. 2016; Lu et al. 2015;
Pauwels et al. 2016) through applications like BIM. Consequently, lots of different digital
knowledge bases for software systems are created. Whereas it is believed that digitization of
the information relevant to the construction asset's life cycle further integrates stakeholders,
the existence of different realities - as represented in standards - has implications on the
successful adoption of digital practices.
Realities in utility companies show differences in regional coverage and the ontological
aspects taxonomy, vocabulary, and semantics. The realities, therefore, reveal different
standards. This stresses the importance of the perspective of philosophy on integration. The
utility owners studied create their own knowledge base, from their own perception on how to
model the reality. Every utility owner, therefore, followed her own organizational standard.
Moreover, these various standards reveal different realities and work practices. Literature
argues that work practice differences and lacking acceptance of standards provide barriers to
successful IT adoption (Adriaanse et al. 2010), but we add to this that, even with usage of
locally accepted standards, it is not likely that an integrated digital practice emerges. This
eventually limits the adoption of a sector-wide uniform digital modelling practice.
Differences in standardized realities cause confusion in a digital reality. One reason for this
observation may be the fragmented nature of the AEC industry. According to Dubois and
Gadde (2002), the AEC industry as a whole can be featured as a loosely coupled system,
which stimulates the generation of variation. Because of this loose coupling, and therefore,
fragmentation, it is not a great surprise that also fragmented realities are seen in digitization
practices. A second reason may be the existing assumption in technology acceptance models
on the effect of standardization. The study showed that without a shared and accepted
standard, fragmented realities are likely to emerge. Whereas standardization is believed to be
a driver of adoption and integration of digital practices, fragmented standards mainly cause
confusion and thereby delimit collaborative digital practices.
PDEng – Final Report 89
This study contributes to literature as follows. First, this study adds better understanding to
the dynamics underlying the studies that identify 'factors' for the adoption of technology and
concludes that effectivity and efficiency of inter-organizational technologies are sometimes
below average (Hjelt and Björk 2006; Sulankivi 2004). By going beyond assuming that
digitization, via standardization, leads to integration, we provide evidence for the hypothesis
that technology adoption in an emerging field goes in line with the definition of many
standards. Based on this, we argue that such digital standards need to be unified and shared
before being able to achieve collaboration benefits of information systems. This notion of
'varied standardization' should receive greater attention in technology acceptance models.
The limitation of this study is that it focussed only on the modelling practices in the utility
domain. Although this is not representative for the full construction industry, it provides
valuable lessons for other domains in construction where digital systems are not yet adopted
by the larger community - e.g. in lower-tiers of supply chains and infrastructure domains.
Therefore, we believe that the differences in realities found in this study are likely to learn the
utility industry and the wider construction community about the importance of considering
shared domain understanding in research as well as in industry digitization initiatives. Future
studies can confirm these findings by extending our work with observation in other parts of
the construction sector. Moreover, this study has only assessed digitization of construction
asset data from an ontological perspective. Future studies should elaborate on the effect of
other types of standards on the integration of stakeholders through digitization.
CONCLUSION
This study postulates the hypothesis that digitization of the construction asset life cycle does
not automatically lead to integration of stakeholders and more collaborative work practices.
Our findings show that modelled realities are represented in international, national, and
organizational standards. Although similar objects and attributes were captured in the various
digital modelling standards, we show that these are 'standardized' in distinctive ways. A
selection of examples from digitization practices in the utility sector illustrates differences in
how domain knowledge is represented. The examples show how elicited realities differ in use
of taxonomy, vocabulary, and semantics. This provides evidence for the existence of
diverging realities. Such differences in modelled digital realities are likely to create ambiguity
and may, therefore, fragment and ultimately delimit integration of digital practices.
The existence of the distinctive standard realities imply that the utility sector currently lacks a
uniform digital modelling practice. This, in turn, limits the possibilities for IT developers to
align information systems that exchange network data in a uniform way between network
operators and contractors. These observed diverging realities seem less present in the facility
and building industry. This part of the construction industry already established and adopted
the object-modelling standard IFC.
This study confirms that various standards can co-exist in the utility sector. One reason for
this may be that this domain is currently making the transition toward a digital practice for
planning and managing their assets. This demonstrates that, while a sector moves toward
implementing digital practices, this does not immediately lead to unification of practices and
communication flows.
The example shows that initiatives that are aimed at stimulating 'digital collaboration', should
be cautious in assuming that digitization immediately supports integration. We, therefore,
urge practice in developing standards that capture shared ontological understandings. Such
shared perception is a precondition for achieving integration though digitization in the
fragmented construction sector.
PDEng – Final Report 90
ACKNOWLEDGEMENTS
We thank Siers Infra Consultancy for providing the authors access to their workplace for data
collection.
REFERENCES
Adriaanse, A, Voordijk, H and Dewulf, G (2010) Adoption and Use of Interorganizational
ICT in a Construction Project. Journal of Construction Engineering and
Management, 136(9), 1003-1014.
Azhar, S (2011) Building Information Modeling (BIM): Trend, Benefits, Risks, and
Challenges for the AEC Industry. Leadership and Management in Engineering,
11(3), 241-252.
Bradley, A, Haijiang, L, Lark, R and Dunn, S (2016) BIM for infrastructure: An overall
review and constructor perspective. Automation in Construction, 71, 139-152.
Bowker, G and Star, S L (1999) Sorting Things Out. Cambridge, MA: MIT Press.
Dubois, A and Gadde, L (2002) The construction industry as a loosely coupled system:
implications for productivity and innovation. Construction Management and
Economics, 20(7), 621-631.
El-Diraby, T E and Osman, H (2011) A domain ontology for construction concepts in urban
infrastructure projects. Automation in Construction, 20(8), 1120-1132.
Gasevic, D, Djuric, D and Devedzic, V (2009) Model Driven Engineering and Ontology
Development. New York, NY: Springer.
Gustavsson, T E, Samuelson, O and Wikforss, Ö (2012) Organising IT in Construction:
Present State and Future Challenges in Sweden. Journal of Information Technology in
Construction, 17, 520-533.
Hjelt, M and Björk, B C (2006) Experiences of EDM usage in construction projects.
Journal of Information Technology in Construction, 11, 113-125.
Howard, R and Björk, B C (2008) Building Information Modelling - Experts views on
standardisation and industry deployment. Advanced Engineering Informatics,
22(2), 271-280.
Lampland, M and Star, S L (2009) Standards and Their Stories: How Quantifying,
Classifying, and Formalizing Practices Shape Everyday Life. Ithaca, NY: Cornell
Univ. Press.
Lu, Y, Li, Y, Skibniewski, M, Wu, Z, Wang, R and Le, Y (2015) Information and
communication technology applications in architecture, engineering, and construction
organizations: a 15-year review. Journal of Management in Engineering, 31(1),
4014010.
Pauwels, P, Zhang, S and Lee, Y (2016) Semantic web technologies in the AEC industry: A
literature review. Automation in Construction, 73, 145-165.
Peansupap, V and Walker, D H T (2005) Factors enabling information and
communication technology diffusion and actual implementation in construction
projects. Construction Management and Economics, 24(3), 321 332.
Saldaña, J (2015) The Coding Manual for Qualitative Researchers. London: Sage
PDEng – Final Report 91
Samuelson, O and Björk, B C (2011) Adoption Processes for EDM, EDI and BIM in the
Construction Industry. Journal of Civil Engineering and Management, 19, 172
187.
Sulankivi, K (2004) Benefits of centralized digital information management in multipartner
projects. Journal of Information Technology in Construction, 9, 35-63.
Timmermans, S and Epstein, S (2010) A world of standards but not a standard world:
Toward a sociology of standards and standardization. Annual Review of
Sociology, 36, 69-89
Turk, Z (2001) Phenomenologial foundations of conceptual product modelling in
architecture, engineering and construction. Artificial Intelligence in Engineering,
2(15), 83-92
Yin, R K (2014) Case Study Research Design and Methods. 5ed. London: Sage
Title Introductie van een uniform objectmodel voor het beheer en onderhoud van
ondergrondse infrastructuur
Author Ramon ter Huurne
Type Congress
Congress CROW Infradagen 2018, Arnhem, the Netherlands, 27-28 June 2018
Published Yes
Presented Yes
Language Dutch
Introductie van een uniform objectmodel voor het beheer en onderhoud van ondergrondse infrastructuur
Ir. R.B.A. ter Huurne
Universiteit Twente
Samenvatting Het digitaliseren van bouwactiva is een veelbesproken onderwerp in de
bouwsector. Om de interoperabiliteit en uitwisselbaarheid van digitale modellen te garanderen,
kent de sector datastandaarden zoals IFC en CityGML. Echter, voor de ondergrondse
PDEng – Final Report 92
infrastructuur bestaan dergelijke datastandaarden slechts in beperkte mate. Bovenal voldoen
deze standaarden niet aan de toenemende informatiebehoefte van organisaties binnen de
nutssector om hun assets te beheren en onderhouden. Op basis van een observerende studie van
de digitale werkpraktijken van twaalf netbeheerders werd duidelijk dat om aan de
informatiebehoefte te kunnen voldoen, netbeheerders hun eigen objectmodellen en
datastandaarden configureren. Een vergelijk tussen deze organisatorische standaarden toont
verschillen in hoe domeinkennis wordt gemodelleerd. De verschillende wijzen waarop
‘realiteiten’ worden gemodelleerd, verminderen de interoperabiliteit en uitwisselbaarheid van
informatie in de sector. Een uniform datamodel voor het beheer en onderhoud van ondergrondse
infrastructuur zou uitwisselbaarheid van informatie en gebruik van digitale beheersystemen in
BIM en GIS kunnen vergroten. Deze paper licht dit standpunt toe en bespreekt de standaard die
momenteel aan de Universiteit Twente wordt ontwikkeld.
Steekwoorden Ondergrondse infrastructuur, beheer en onderhoud, objectmodel, ontologie,
datastandaard
1. Introductie
Building Information Modelling (BIM) is momenteel een veelbesproken onderwerp in
literatuur en beleidstukken (Bradley et al. 2016; Lu et al. 2015; Pauwels et al. 2016). BIM kan
gedefinieerd worden als een gedeelde digitale representatie van de fysieke en functionele
karakteristieken van een bouwobject (ISO 29481 2010) . Kenmerkend voor BIM modellen is
dat zij parametrisch en object-georiënteerd zijn. Implementatie van BIM heeft het ontstaan van
zogenoemde digitale ‘tweelingen’ als resultaat. Deze digitale tweelingen weerspiegelen het
werkelijke object middels een virtuele reconstructie, bijvoorbeeld in een ‘BIM systeem’. Om
de communicatie tussen dergelijke BIM systemen te faciliteren zijn datastandaarden
(objectmodellen) als de Industry Foundation Classes (IFC) (buildingSMART) geïntroduceerd
(Turk 2001).
Daar waar BIM met name wordt toegepast in de bouwsector, worden voor het modelleren van
ruimtelijke gegevens vooral Computer Aided Design (CAD) en Geographic Information
Systems (GIS) toegepast. Voor ruimtelijke gegevens bestaat de datastandaard CityGML (Open
Geospatial Consortium). Zowel IFC als CityGML worden geleidelijk aan geaccepteerd en
gebruikt als de dominante manier om domeinkennis te modelleren en gegevens uit te wisselen
tussen digitale systemen.
Echter, in tegenstelling tot de bouwsector zijn gemeenschappelijke objectmodellen voor andere
grote typen van infrastructuur, zoals transport en ondergrondse infrastructuur (Bradley et al.
2016) slechts in beperkte mate ontwikkeld. Zonder gemeenschappelijke objectmodellen is de
interoperabiliteit en uitwisselbaarheid van gegevens beperkt, omdat gebruikers geen afspraken
gemaakt hebben over hoe zij bepaalde objecten noemen, beschrijven en visualiseren. Dit kan
leiden tot verwarring en miscommunicatie indien gegevens toch uitgewisseld worden.
Vanuit deze gedachte werkt de Universiteit Twente aan een uniform objectmodel voor
ondergrondse infrastructuur, specifiek voor het domein van beheer en onderhoud. Daar waar
voor ondergrondse infrastructuur wel enkele objectmodellen bestaan, zoals het
Informatiemodel Kabels en Leidingen (IMKL) (Geonovum 2017) zijn deze vrij summier in het
beschrijven van beheer en onderhoud gerelateerde informatie. Echter, door informatie over
kabels en leidingen op een consistentie en complete manier vast te leggen, worden digitale
PDEng – Final Report 93
modellen beter bruikbaar voor beheer en onderhoud. Dit kan leiden tot efficiëntere
beheerprocessen.
Deze paper beschrijft een selectie van bestaande objectmodellen voor ondergrondse
infrastructuur en legt uit waarom één uniform objectmodel voor de uitwisseling van kabels- en
leidingdata voor het domein van beheer en onderhoud nodig is. Daartoe wordt allereerst de
invloed van digitalisering op het modelleren van asset data in het algemeen besproken. Daarna
worden bestaande initiatieven tot datastandaardisatie voor ondergrondse infrastructuur
uiteengezet op internationaal, nationaal en organisatorisch niveau. Dit leidt tot het bespreken
van de behoefte aan een uniform objectmodel. De paper eindigt met conclusies en
aanbevelingen voor vervolgonderzoek.
2. Een digitaal tijdperk
Sinds de opkomst van het digitale tijdperk, zijn digitale modellen bijna niet meer weg te denken
in de bouwsector. CAD, GIS en BIM systemen worden door tal van partijen binnen de
bouwsector inmiddels ingepast ter realisatie van de digitale ‘tweeling’.
2.1 Ontwikkeling: digitale ‘tweelingen’ van bouwobjecten
Het toepassen van technologie in bouwprojecten kan helpen in het verbeteren van de inter-
organisatorische communicatie, coöperatie en coördinatie (Adriaanse et al. 2010; Peansupap en
Walker 2005). Het digitaliseren van bouwactiva – en daarbij een digitale ‘tweeling’ – geschiedt
middels het proces van het definiëren van de concepten, attributen en hun relaties van een
betreffend bouwobject. Voor dit digitalisering-proces wordt gebruik gemaakt van met name
GIS en BIM systemen en in beperkte mate CAD systemen.
CAD systemen worden gebruikt voor het ontwerpen en visualiseren van, onder andere,
bouwactiva, in zowel 2D als 3D. CAD is sterk gefocust op het tekenen van gedetailleerde
modellen, bestaande uit punten, lijnen en polygonen. Kenmerkend voor CAD modellen is dat
zij geen topologie hebben en niet gekoppeld zijn aan objectbibliotheken en databases. Daarmee
is een CAD systeem minder geschikt voor het modelleren van digitale ‘tweelingen’.
GIS systemen zijn informatiesystemen die worden toegepast om (ruimtelijke) informatie over
geografische objecten op te slaan en daarbij te beheren, bewerken, analyseren en presenteren.
Ook bevatten GIS systemen, in tegenstelling tot CAD systemen, over een topologisch model
van punten, lijnen en polygonen, daarbij mogelijk object-georiënteerd en gelinkt aan een
objectbibliotheek en database. Hierdoor ‘begrijpt’ een GIS model de onderlinge relaties tussen
de gemodelleerde concepten en relaties en worden analyses op data mogelijk.
BIM systemen zijn informatiesystemen die worden toegepast voor het modelleren van alle
fysieke en functionele kenmerken van een gebouw. Kenmerkend voor een BIM model is, is dat
zij een gedeelde kennisbron is voor de informatie van het betreffende gebouw gedurende de
gehele levenscyclus. Daarnaast zijn BIM modellen object-georiënteerd, middels de koppeling
met een objectbibliotheek, op basis van een van een topologisch model. Dit betekent dat
verschillende partijen die in het BIM model werken, hun informatie aan het gelijke object
kunnen koppelen, wat niet alleen geometrische informatie behelst, maar ook non-geometrische
informatie zoals kosten en planning gerelateerd informatie. BIM modellen genereren daardoor
een rijke set aan asset data.
Hoewel BIM en GIS systemen een ander doel dienen, zijn beiden uitermate geschikt voor het
realiseren van een digitale ‘tweeling’ voor respectievelijk bouwactiva en geografische objecten.
PDEng – Final Report 94
De rijke set aan informatie die deze systemen kunnen genereren, biedt de mogelijkheid het
digitale model zo nauwkeurig mogelijk te beschrijven en aan te laten sluiten op de realiteit.
2.2 Gevolg: explosie van data
Een herkenbare trend in de bouwsector is de groeiende aandacht voor de
levenscyclusbenadering van assets (Bradly et al. 2016). Waar men voorheen alleen
objectlocaties digitaliseerde, vereist men tegenwoordig al meer digitale informatie voor het
beheer en onderhoud. Zo worden tegenwoordig bijvoorbeeld ook materiële, dimensionele,
kosten en planning gerelateerde informatie opgeslagen. Men refereert ook wel naar deze
ontwikkeling als ‘de explosie van data’. Om deze data op te slaan wordt gebruik gemaakt van
bouwobject- of geografische informatiesystemen.
De verschuiving naar een levenscyclusbenadering van assets gaat idealiter uit van
domeinkennis die bestaat uit een consistente en complete set aan concepten en relaties. Het
faciliteren van domeinkennis op een
dergelijke consistente en complete
manier vraagt om standaardisatie.
Voor BIM gerelateerde systemen is
dat IFC voor gebouwen. Voor GIS
gerelateerde systemen bestaat onder
andere CityGML voor stedelijke
informatie. Deze standaarden
schrijven een uniforme structuur en
naamgeving voor gerelateerde
domeinkennis voor. In het geval van
IFC wordt bijvoorbeeld een ‘deur’
(IfcDoor) gemodelleerd als gepresenteerd in figuur 1 (buildingSMART 2016). In figuur 1 is
‘IfcDoor’ onderdeel van bovengelegen klassen, waaronder ‘IfcBuildingElement’. Daarnaast,
niet gevisualiseerd in de figuur, heeft elk van de klassen een lijst aan attributen, en wordt binnen
IFC elk van de klassen en attributen middels een eenduidige definitie beschreven.
Een gemeenschappelijk objectmodel als IFC helpt in het systematisch vastleggen van
domeinkennis in een consistente en complete manier. IFC verbetert daarmee de
interoperabiliteit en uitwisselbaarheid van gebouw-gerelateerde digitale informatie. Echter,
voor ondergrondse infrastructuur zijn dergelijke gemeenteschappelijke objectmodellen in
slechts beperkte mate ontwikkeld.
3. Modelleren van ondergrondse infrastructuur
Het modelleren van ondergrondse infrastructuur heeft zich lange tijd beperkt tot het gebruik
van CAD omgevingen. Gezien de beperkte toepasbaarheid van CAD voor het opslaan van
informatie, omvatten deze CAD modellen vaak niet meer dan geometrische informatie.
Vanwege de toenemende complexiteit en omvang van netinformatie – gestimuleerd door de
groeiende aandacht voor de levenscyclusbenadering – wordt ondergrondse infrastructuur
tegenwoordig steeds vaker gemodelleerd in GIS georiënteerde omgevingen. Echter, zoals
Bradley et al. (2016) aangeven, zijn gemeenschappelijke dataformaten voor ondergrondse
infrastructuur slechts in beperkte mate ontwikkeld. Standaardisatie initiatieven die al wél
bestaan, worden in de volgende paragrafen op nationaal, internationaal en organisatorisch
niveau beschreven.
Figuur 1 – Modelleren van een ‘deur’ volgens IFC
IFC
PDEng – Final Report 95
3.1 Nationale en internationale initiatieven
Op basis van verkregen informatie uit een literatuurstudie is vergeleken welke initiatieven tot
standaardisatie van ondergrondse infrastructuur momenteel bestaan. Deze paper presenteert een
selectie van enkele bekende initiatieven, op zowel nationaal als internationaal niveau. Een
overzicht van deze geselecteerde initiatieven is gepresenteerd in tabel 1. Meerdere initiatieven
bestaan, maar worden in deze paper niet toegelicht.
Tabel 1 – Overzicht van een selectie van nationale en internationale initiatieven Naam Toepassingsgebied Doel
INSPIRE Europa Vergroten van interoperabiliteit en uitwisselbaarheid
van ruimtelijke gegevens tussen Europese lidstaten.
CityGML
UtilityNetwork ADE
Wereld Modelleren van ondergrondse infrastructuur in 3D
stadsmodellen.
IMKL Nederland Vergroten van interoperabiliteit en uitwisselbaarheid
van ruimtelijke gegevens tijdens
graafwerkzaamheden.
PAS 256 Verenigd Koninkrijk Verbeteren van de kwaliteit, nauwkeurigheid en
betrouwbaarheid van bestaande netinformatie.
INSPIRE (Infrastructure for Spatial Information in Europe) (European Commission 2013) is
een Europese richtlijn gericht op het creëren van een Europese infrastructuur waarover lidstaten
ruimtelijke gegevens uitwisselen. Doel hierbij is het vergroten van de interoperabiliteit en
uitwisselbaarheid van ruimtelijke gegevens. INSPIRE omvat dataspecificaties voor een breed
spectrum aan ruimtelijke gegevens, waarvan ondergrondse infrastructuur er één is. De INSPIRE
richtlijn beschrijft hoe ondergrondse infrastructuur middels concepten en relaties beschreven
dient te worden. CityGML UtilityNetwork ADE (Application Domain Extension) is een
extensie op CityGML (Open Geospatial Consortium 2012) en is gericht op het modelleren van
ondergrondse infrastructuur in 3D stadsmodellen. De standaard kan toegepast worden over de
gehele wereld. Als extensie op CityGML beschrijft de UtilityNetwork ADE de concepten en
relaties benodigd om ondergrondse infrastructuur te beschrijven binnen CityGML. Het IMKL
(Informatiemodel Kabels en Leidingen) (Geonovum 2017) is een Nederlands initiatief en
beschrijft concepten en relaties benodigd voor het vastleggen van netinformatie, om daarmee
de interoperabiliteit en uitwisselbaarheid van netinformatie tussen netbeheerders te vergroten.
Middels een centraal platform bij het Kadaster voorziet het IMKL netbeheerders van een
uniforme manier om netinformatie uit te wisselen in geval van graafwerkzaamheden. De PAS
(Publically Available Specification) 256 (British Standards Institution 2017) is een richtlijn
afkomstig uit het Verenigd Koninkrijk die helpt in het vastleggen, opnemen, onderhouden en
delen van locatiegegevens en gegevens van ondergrondse infrastructuur. De richtlijn heeft als
doel de kwaliteit, nauwkeurigheid en betrouwbaarheid van bestaande netinformatie te
verbeteren bij uitwisseling van dergelijke informatie.
Wanneer deze standaarden vergeleken worden met de informatiebehoefte vanuit het domein
van beheer en onderhoud, wordt duidelijk dat bestaande standaarden deze informatiebehoefte
niet kunnen beschrijven. Gezien het doel van de standaarden – welke niet het beheer en
onderhoud van ondergrondse infrastructuur betreft – is dit ook niet verwonderlijk. In het
algemeen beschrijven deze initiatieven wel de locatie, het thema, de dimensies en het materiaal,
maar missen zij specifieke gegevens zoals bijvoorbeeld het type van bepaalde objecten, kosten
en planning gerelateerde attributen alsook onderhoudsgegevens. Het gevolg hiervan is dat men
organisatorische initiatieven gaat ontwikkelen.
3.2 Organisatorische initiatieven
PDEng – Final Report 96
Gezien nationale en internationale standaarden niet aan de informatiebehoefte voldoen die
tegenwoordig in de nutssector vereist wordt, beslaat een overgroot deel van de manier waarop
de ondergrondse infrastructuur wordt gemodelleerd organisatorische initiatieven. Middels een
observatie van de werkpraktijken van twaalf netbeheerders is gebleken dat netinformatie wordt
vastgelegd volgens (1) software standaarden / configuraties of (2) standaarden / configuraties
van netbeheerders zelf. Binnen deze standaarden worden de concepten en relaties die relevant
geacht voor het beheer en onderhoud van ondergrondse infrastructuur opgenomen.
Bij een vergelijk van de geobserveerde organisatorische standaarden valt op dat concepten en
relaties beschreven worden middels verschillende (1) attributen, formats en relaties, en (2)
semantiek. Netbeheerders schrijven hun eigen objectmodel voor, gebaseerd op hun eigen
werkpraktijken. Een verschil in hoe netbeheerders de structuur van hun netten modelleren, is
bijvoorbeeld het specificeren van het type net als attribuut of als een eigen klasse. Meer
specifiek, in het geval van glasvezel kan er bijvoorbeeld gekozen worden om dit als attribuut
bij een telecom net te modelleren, of het te modelleren als een eigen klasse en daarmee
glasvezelnet. Een simpel voorbeeld voor een verschil in semantiek is bijvoorbeeld het
beschrijven van de x- en y-coördinaat van een object middels de term ‘ligging’ en de term
‘locatie’. In beide gevallen is de intrinsieke betekenis gelijk, maar gebruiken netbeheerders een
verschillende term om dit te beschrijven.
Tezamen kunnen dergelijke verschillen leiden tot verwarring en miscommunicatie indien de
netinformatie wordt uitgewisseld met andere partijen. Interoperabiliteit van netinformatie blijft
daarmee een groot probleem. Ploeger et al. (2005) formuleert de grote verscheidenheid in
datastandaardisatie van ondergrondse infrastructuur dan ook wel als “de chaos in de bodem”.
4. Behoefte aan een uniform objectmodel
Daar waar de bouwsector digitale gestandaardiseerde objectmodellen kent – zoals voor
gebouwen in de vorm van IFC – ontbreekt een dergelijke standaard voor de ondergrondse
infrastructuur. Vele initiatieven, voornamelijk op organisatorisch niveau, vergroten de
verscheidenheid in datastandaardisatie van ondergrondse infrastructuur en creëren daarmee een
“wereld van standaarden, maar niet een gestandaardiseerde wereld” (Timmermans en Epstein
2010). Om de interoperabiliteit en uitwisselbaarheid van asset data in de nutssector te
verbeteren, is het zaak af te spreken hoe de domein specifieke kennis van de nutssector voor
het beheer en onderhoud opgeslagen moet worden. Er is behoefte aan één uniform objectmodel.
4.1 Het standaardiseren van domeinkennis
Het doel van standaardisatie, in het algemeen, is het vastleggen van de ‘realiteit’ en daarmee
het creëren van uniformiteit tussen culturen, tijd en geografie. Het vastleggen van deze realiteit
gebeurt middels afgesproken regels, ook wel aangeduid als een standaard (Timmermans en
Epstein 2010). Standaarden komen daarbij in vele vormen voor (Bowker en Star 1999),
waaronder datastandaarden die relevant zijn binnen de context van deze paper.
De vraag is echter hoe een dergelijke realiteit gevangen kan worden in een datastandaard voor
het beheer en onderhoud van ondergrondse infrastructuur. De grote verscheidenheid in
organisatorische objectmodellen zoals eerder beschreven toont aan dat er vele verschillende
percepties op de realiteit bestaan. De intentie en interpretatie van domeinkennis zijn daarom
PDEng – Final Report 97
van grote relevantie wanneer een realiteit in een standaard gevangen moet worden, omdat de
betekenis zowel door de auteurs als de gebruikers van de standaarden gevormd kan worden.
Eenmaal geëxpliciteerd in een tekstuele vorm kunnen realiteiten daarom verschillende
plausibele interpretaties hebben (Turk 2001).
Naar een datastandaard kan ook gerefereerd worden als een ‘ontologie’. Een ontologie
beschrijft de wereld als gezien door een groep personen op een bepaalde tijd, volgens een school
van denken die gebaseerd is op een reeks fundamentele proposities of wereldbeelden (El-Diraby
en Osman 2011). Simpel gezegd is een ontologie een model of structuur van kennis. Eenmaal
geadopteerd, representeert een ontologie domeinkennis – en daarmee de ‘realiteit’ – in een
verenigde, vereenvoudigde en consistente manier. Binnen een ontologie wordt de taxonomie
en vocabulaire van de domeinkennis vastgelegd. De taxonomie beschrijft de hiërarchische
categorisatie / classificatie van de concepten. De vocabulaire beschrijft de terminologie en
naamgeving van de concepten, om semantische diversiteit te voorkomen (Gasevic et al. 2009).
Terugkijkend naar figuur 1 kan een uniforme taxonomie en vocabulaire ook in IFC herkend
worden.
Het gebruik van een ontologie is zodanig een uitstekend middel om de kennis benodigd voor
het domein van beheer en onderhoud voor ondergrondse infrastructuur vast te leggen in een
uniforme realiteit. Met andere woorden, aan de basis van een uniforme realiteit benodigd voor
een uniforme standaard, staat een ontologie waarin de concepten en relaties voor het domein
van beheer en onderhoud van ondergrondse infrastructuur zijn opgenomen.
Voor het verkrijgen van de domeinkennis is het van groot belang dat de meest representatieve
kennis vergaard wordt. Binnen het project aan de Universiteit Twente – waarin een ontologie
voor het domein van beheer en onderhoud voor ondergrondse infrastructuur wordt ontwikkeld
– wordt dit bewerkstelligd door eindgebruikers van de ontologie nauw te betrekken bij de
ontwikkeling. Hiermee wordt getracht de uniforme realiteit binnen de ontologie zo goed
mogelijk te laten aansluiten bij de verschillende percepties op de realiteit die de nutssector nu
kenmerkt. Dit vergemakkelijkt adoptie en implementatie van de ontologie.
Ter illustratie toont figuur 2 een
snapshot van de ontologie die
momenteel ontwikkeld wordt aan
de Universiteit Twente. De
ontologie is gebaseerd op het
CityGML UtilityNetwork ADE
objectmodel en is daar waar
nodig aangepast dan wel
uitgebreid met de relevante
concepten en relaties voor het
domein van beheer en
onderhoud. De ontologie is
gevisualiseerd in de Unified
Modelling Language (UML)
welke ook toegepast wordt in de
UtilityNetwork ADE, maar
tevens in bijvoorbeeld INSPIRE
en het IMKL. Figuur 2 toont hoe
een ‘Cable’ (kabel) en ‘Pipe’
(leiding) gemodelleerd kunnen
worden en toont de
verscheidene attributen die beide objecten bevatten zoals ‘materialType’
(materiaaleigenschappen) en ‘class’ (type kabel of leiding). Een kabel of leiding is binnen de
Figuur 2 – Snapshot van de ontologie ontwikkeld door de
Universiteit Twente
PDEng – Final Report 98
ontologie een ‘AbstractDistribution’ (distributie) element. Deze snapshot toont slechts een klein
gedeelte van de ontologie. Daarnaast is de ontologie in ontwikkeling wat betekent dat het model
nog aan mogelijke wijzigingen onderhevig is.
4.2 Voordelen: meer dan interoperabiliteit alleen
Het opstellen van één uniforme standaard – de ontologie – biedt de nutssector meerdere
voordelen. Allereerst helpt het in het overbruggen van de kloof tussen alle verschillende
organisatorische standaarden die de nutssector nu kenmerkt. Eén uniforme datastandaard helpt
om deze realiteiten op elkaar aan te laten sluiten. Dit vergroot de interoperabiliteit en
uitwisselbaarheid van data binnen de sector en helpt verwarring en miscommunicatie in de
werken van de sector voorkomen.
Ten tweede helpt een uniforme standaard in het versnellen van de ontwikkelingen van software
ontwikkelaars. Gekeken naar IFC en de toepassing hiervan in verschillende BIM gerelateerde
software, zoals Autodesk Revit, is het voor software ontwikkelaars gemakkelijker om hun
software te optimaliseren indien deze gebaseerd is op één uniforme datastandaard. Software
kan op deze manier ‘slimmer’ worden gemaakt door bijvoorbeeld specifieke simulaties aan te
bieden die analyses maken van de onderhoudsdata die wordt opgeslagen. Nu gebeurt het
configureren van software (bijna) geheel binnenshuis bij de gebruikers zelf. Een integrale
aanpak op basis van één uniforme standaard kan het optimaliseren van software versnellen.
Ten derde draagt een standaard voor de nutssector mogelijk ook bij aan de versnelde adoptie
en integratie van BIM en GIS systemen voor de ondergrond. In de literatuur (Bormann et al.
2014; Cheng et al. 2013; Hor et al. 2016) wordt de integratie van GIS en BIM als een grote
uitdaging gezien, vanwege syntactische en semantische verschillen tussen beiden domeinen.
Syntactisch refereert naar de opbouw en structuur van de data. Semantisch refereert naar de
betekenis van de data. Een uniforme standaard voor de nutssector draagt met name bij aan de
syntactische integratie. Om semantische verschillen te overbruggen wordt de laatste jaren veel
onderzoek gedaan naar ‘semantic web’ technologie (Pauwels et al. 2016).
5. Conclusies en vervolgonderzoek
Deze paper biedt een kijk in de spreekwoordelijke keuken van het modelleren van ondergrondse
infrastructuur. Het digitale tijdwerk waarin we ons bevinden, tezamen met een groeiende
aandacht voor de levenscyclusbenadering, heeft geleid tot een explosie van data binnen het
beheer en onderhoud van ondergrondse infrastructuur. Daar waar de verschuiving naar een
levenscyclusbenadering van assets idealiter uit gaat van domeinkennis bestaande uit een
consistente en complete set aan concepten en relaties, toont de praktijk grote verschillen.
De aanwezigheid van organisatorische – en daarbij uiteenlopende – standaarden impliceert dat
de nutssector op het moment een uniforme digitale modelleer standaard mist. Netbeheerders
modelleren netinformatie volgens hun eigen ‘realiteit’. Dit hindert de interoperabiliteit en
uitwisselbaarheid van netinformatie in de nutssector. Bovendien belemmert het software
ontwikkelaars om informatiesystemen gebruikt binnen het beheer en onderhoud van
ondergrondse infrastructuur op elkaar te laten aansluiten.
Deze paper zet aan tot het ontwikkelen van één uniforme datastandaard voor het domein van
beheer en onderhoud van ondergrondse infrastructuur. Als basis voor een dergelijke standaard
wordt het gebruik van een ontologie voorgesteld. Een ontologie representeert domeinkennis –
en daarmee de ‘realiteit’ – in een verenigde, vereenvoudigde en consistente manier. Een
ontologie helpt daarmee in het overbruggen van de kloof tussen organisatorische standaarden,
PDEng – Final Report 99
in het versnellen van software optimalisatie en mogelijk in de integratie van BIM en GIS
gerelateerde domeinen.
Adoptie van een uniforme datastandaard door de nutssector zal mogelijk in eerste instantie met
enige aarzeling verlopen. Tenslotte, het doorbreken van huidige routines en werkpraktijken kan
een kostbaar en tijdrovend proces zijn. Echter, de voordelen van één uniforme datastandaard
zullen op de lange termijn de gehele nutssector naar een hoger plan tillen. Deze paper tracht de
bewustwording van de huidige problematiek te vergroten en de sector aan te zetten tot het
investeren in uniformering van netinformatie.
Het ontwikkelen van een datastandaard voor de ondergrondse infrastructuur binnen het project
van de Universiteit Twente zal eind 2018 tot een einde komen. De paper roept daarom op tot
vereniging van netbeheerders om verdere ontwikkeling en adoptie van één uniforme standaard
te stimuleren.
6. Referenties
Adriaanse, A, Voordijk, H en Dewulf, G (2010) Adoption and Use of Interorganizational
ICT in a Construction Project. Journal of Construction Engineering and
Management, 136(9), 1003-1014.
Bormann, A, Kolbe, T H, Donaubauer, A, Steuer, H, Jubierre, J R en Flurl, M (2015) Multi
Scale Geometric-Semantic Modeling of Shield Tunnels for GIS and BIM
Applications. Computer-Aided Civil and Infrastructure Engineering, 30, 263-281.
Bower, G en Star, S L (1999) Sorting Things Out. Cambridge, MA: MIT Press.
Bradley, A, Haijiang, L, Lark, R en Dunn, S (2016) BIM for infrastructure: An overall
review and constructor perspective. Automation in Construction, 71, 139-152.
British Standards Institution (2017) PAS 256:2017. BSI Standards Limited 2017.
buildingSMART (2016) Industry Foundation Classes Version 4 – Addendum 2. Opgehaald
van http://www.buildingsmart-tech.org/ifc/IFC4/Add2/html/.
Cheng, J C P, Deng, Y en Du, Q (2013) Mapping Between BIM Models and 3D GIS City
Models of Different Levels of Detail. 13th International Conference of Construction
Applications of Virtual Reality, 30-31.
El-Diraby, T E en Osman, H (2011) A domain ontology for construction concepts in urban
infrastructure projects. Automation in Construction, 20(8), 1120-1132.
European Commission (2013) D2.8.III.6 INSPIRE Data Specification on Utility and
Government Services – Technical Guidelines.
Gasevic, D, Djuric, D en Devedzic, V (2009) Model Driven Engineering and Ontology
Development. New York, NY: Springer.
Geonovum (2017) IMKL2015 – Dataspecificatie Utiliteitsnetten.
Hor, A-H, Jadidi, A en Sohn, G (2016) BIM-GIS Integrated Geospatial Information Model
Using Semantic Web and RDF Graphs. ISPRS Annals of the Photogrammetry,
Remote Sensing and Spatial Information Sciences, 3(4), 73-79.
ISO 29481 (2010) ISO 29481-1-201(E) Building Information Modeling – Information
Delivery Manual – Part 1.
Lu, Y, Li, Y, Skibniewski, M, Wu, Z, Wang, R en Le, Y (2015) Information and
communication technology applications in architecture, engineering, and construction
organizations: a 15-year review. Journal of Management in Engineering, 31(1),
4014010.
Pauwels, P, Zhang, S en Lee, Y (2016) Semantic web technologies in the AEC industry: A
literature review. Automation in Construction, 73, 145-165.
PDEng – Final Report 100
Peansupap, V en Walker, D H T (2005) Factors enabling information and communication
technology diffusion and actual implementation in construction projects. Construction
Management and Economics, 24(3), 321-332.
Ploeger, H, Loenen, B, van, Kap, A, P en Stoter, J, E (2005) Kabels en leidingen: de chaos in
de bodem. Nederlands Juristenblad, 23, 1186-1191.
Timmermans, S en Epstein, S (2010) A world of standards but not s standard world:
Toward a sociology of standards and standardization. Annual Review of Sociology,
36, 69-89.
Turk, Z (2001) Phenomenologial foundations of conceptual product modelling in
architecture, engineering and construction. Artificial Intelligence in Engineering,
2(15), 83-92.