Re-use of an ontology for modelling urban energy systems
Koen H. van Dam, Member, IEEE, and James Keirstead
Abstract— The use of ontologies for the interoperability of
software models is widespread, with many applications also in
the energy domain. By formulating a shared data structure
and a definition of concepts and their properties, a language is
created that can be used between modellers and—formalised
in an ontology—between model components. When modelling
energy systems, connections between different infrastructures
are critical, e.g. the interaction between the gas and electricity
markets or the need for various infrastructures including power,
heat, water and transport in cities. While a commonly shared
ontology of energy systems would be highly desirable, the fact
is that different existing models or applications already use
dedicated ontologies, and have been demonstrated to work well
using them. To benefit from linking data sources and connect-
ing models developed with different ontologies, a translation
between concepts can be made. In this paper a model of an
urban energy system built upon one ontology is initialised using
energy transformation technologies defined in another ontology,
thus illustrating how this common perspective might benefit
researchers in the energy domain.
I. INTRODUCTION
Ontologies, i.e. formalised conceptualisations [8], are a
proven tool for interdisciplinary modelling, providing an
indispensable shared formal language. They facilitate con-
sistent software design and interoperability between models,
even when they have been built by different modellers,
working with different techniques and in different domains.
In energy systems multiple infrastructures are interlinked and
modellers could therefore benefit from the interface provided
by a shared ontology in order to access disparate data sources
and connect models.
The energy modelling community has begun to recognize
this need and some initial work has been done. Keirstead
and van Dam [15] concluded that “we would be interested in
establishing an open community effort to build a standardised
modelling ontology” for energy systems, based on the lessons
learnt from a demonstration that an energy conversion tech-
nology from one ontology could successfully be described in
the other even though the exact properties and classes used
were different. Such a standard and shared ontology could
have great benefits to the modelling community as it would
enable stronger cooperation between groups and disciplines.
Similarly Catterson et al. [2] advocate the development of
a shared ontology for power systems, saying “[. . . ] the
K.H. van Dam is with the Faculty of Technology, Policy and Management
of the Delft University of Technology, 2600 GA Delft, The Netherlands
k.h.vandam@tudelft.nl
J. Keirstead is with the Department of Chemical Engineering of
Imperial College, South Kensington Campus, London, SW7 2AZ, UK
j.keirstead@imperial.ac.uk. He would like to acknowledge
the support of BP through the BP Urban Energy Systems project at Imperial
College London.
community of researchers working in this area must agree
on the following points: Standards for data exchange [. . . ]
and [. . . ] creation of an upper ontology for smart grid terms
and concepts, likely based on existing data standards [. . . ]”,
referring to the work of the IEEE Power Energy Society
trying to address these issues [16]. So far this work as
not been open—a main requirement for widespread use and
community involvement—but there are plans to release the
standards shortly.
There are however two pre-requisites for the development
of a shared ontology: that researchers have a shared view of
the world, and that they have comparable aims for which they
want to use the ontology. In the ontology definition phase, the
first fundamental rule is that “there is no one correct way to
model a domain—there are always viable alternatives” [19].
Guarino [9] stresses the intended meaning in his definition
of an ontology. If an ontology is to be used by people with
a different world view (e.g. a different valuation on what is
important and what does not need to be emphasized) then
the resulting ontology may end up being only generic and
without much expressive power for any application. When
these two conditions (shared view of the world and shared
aim) have been met, a joint effort to developing an ontology
may be fruitful.
Although it is our aim to design a high-level ontology for
modelling energy systems, it has to be acknowledged that
researchers already have existing tools and models which
incorporate ontologies that may be closely related, but not
the same. The challenge therefore is to develop an interface
which allows the re-use of elements from one ontology in
another, while providing a uniform representation of the
domain. The goal of this paper is to begin this process by
identifying the major concepts that might be part of such
an ontology and to explore some of the associated practical
issues.
The paper is structured as follows. First in Section II
the background of two existing ontologies considered in
this paper is briefly sketched, after which we discuss how
ontologies can be connected and how interoperability can
be provided (Section III). Using a case study presented in
Section IV, we demonstrate how the the use of a master
ontology enables elements from both ontologies to be used
in a single model of an urban energy system. Section V
concludes with a discussion on the usefulness of the approach
and we discuss how ontology couplings can be done more
easily in the future, particularly drawing attention to a new
initiative to develop a high-level energy systems modelling
ontology.
978-1-4244-8479-9/10/$26.00 c
2010 IEEE

II. BACKGROUND
Before going further, we should clarify what we mean
by an “energy systems model”. By “energy system” we
have in mind Jaccard’s [12] broad definition: “the combined
processes of acquiring and using energy in a given society
or economy” (p. 6). Models of course mean different things
to different people, but in this context we are primarily
concerned with quantitative models for the analysis, predic-
tion, exploration and study of different scenarios to support
decision making. This field therefore encompasses a wide
range of disciplines and modelling tools; some practitioners
may be using spreadsheets to examine aggregate national
consumption statistics, while others may use detailed simu-
lation software to assess the performance characteristics of a
specific energy technology. Yet all of these applications im-
ply an underlying conceptualization of the major elements of
an energy system and their associated attributes; ontologies
provide the tools to make these descriptions explicit.
In this paper two different energy ontologies, developed
independently from each other, are used. They are briefly
introduced below.
A. UES ontology
SynCity (short for “Synthetic City”) is a modelling system
for urban energy systems developed at Imperial College
London [13]. The goal of SynCity is to provide a platform
for the modelling of urban energy systems (UES) at multiple
scales. This requires the use of several different modelling
techniques, including mathematical programming and agent-
based modelling. Within this context, the UES ontology was
introduced to provide consistent class definitions between
the models and for the storage and management of system
components. The UES ontology consists of a number of
object classes that describe the main elements of an urban
energy system, as well as specific instances of these classes.
Within the context of this paper, two classes are highlighted:
Resources such as electricity or natural gas, are described
by a series of physical, economic, and model attributes.
These include mass and energy densities, unit prices, or
maximum stock values.
Processes are technologies that convert one set of resources
into another set. There are multiple subclasses to de-
scribe simple conversion technologies as well as more
complex transportation and storage processes.
The ontology also contains detailed classes for the defini-
tion of the physical infrastructure of a city and it features a
number of supporting classes, including the Unit class and
its instances, designed with the JScience library [3] in mind
to facilitate easy unit conversion.
B. STS Ontology
To support the development of agent-based models of
socio-technical systems (STS), the STS ontology has been
developed at the Delft University of Technology [21]. The
aim was to build an ontology not for one specific application
domain (e.g. an electricity infrastructure), but to find com-
monalities between applications and therefore to develop a
modelling framework that is able to deal with the reality
of socio-technical network systems that are interconnected
across sectors. Modellers use the ontology to formalize do-
main knowledge, as language in the definition of behavioural
rules and as communication protocols between agents. In this
way, parts of the model can be re-used (e.g. re-using the
model of a certain technology with a different agent, or re-
using behavioural rules of one agent in another one, or even
copying complete agents into another model) and models of
different infrastructures can be connected, even when they
are developed by different modellers. For the purpose of this
paper, the main classes are the following:
Technologies follow the input-output paradigm to define
energy or mass conversions. It can use different recipes
consisting of different input-output pairs to reflect dif-
ferent modes of operation. Properties define, for exam-
ple, the capacity of the technology to produce a certain
product, the maintenance and operational costs attached
to its operation, and so on. Technologies are not active
units but have to be operated by an agent, who makes
decisions about how to use the technology.
GoodNames describe the “products” that exist in the system
(e.g. crude oil or electricity).
Additionaly, the ontology features a rich set of classes to
describe agents, different types of contracts, and the physical
infrastructure as well as the actual flows in the system.
C. Comparison
Each ontology was designed with different initial goals
in mind, but as demonstrated in [15], there are significants
overlaps and compatible elements. The aim of the definition
of the Technology and Process concepts is the same:
to provide the inputs and outputs of (energy) conversion
technologies with several properties, including costs, for use
in models with which to assess different policies or config-
urations at the operational, tactical and strategic level. Both
ontologies have been defined in Prote´ge´ Frames [7]. There
are, however, also substantial differences and this creates
difficulties when trying to use objects from one ontology
within another modelling domain. While modellers know that
Technology and Process are the same and that their
properties, even though conceptualised slightly differently,
are comparable, this does not mean the software applications
“understand” this as well. To enable interoperability between
the two ontologies and applications built using them, they
need to be explicitly connected.
III. CONNECTING ONTOLOGIES
Haslhofer and Klas [11] present an extensive survey
of techniques for obtaining interoperability between data
formalised in different ways, highlighting the difference
between instance, schema and schema definition language
following the definition of meta levels in the Object Man-
agement Facility (OMF) specification [20]. We can map
this to ontology instances, ontology definition, and ontology
language, respectively. The ontology instances can be consid-
ered as level M0, the definition of the ontology as M1 and the