<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>Integrating Railway Subdomain-Specific Data Standards into a common IFC-based Data Model</article-title>
      </title-group>
      <contrib-group>
        <aff id="aff0">
          <label>0</label>
          <institution>Sebastian Esser &amp; André Borrmann Technical University of Munich</institution>
          ,
          <country country="DE">Germany</country>
        </aff>
      </contrib-group>
      <abstract>
        <p>Like in many other domains in the construction industry, concepts and methods of Building Information Modeling (BIM) are becoming more and more important for the design of civil infrastructure assets. Although several data formats have been developed to capture domainspecific data, none of these schemas can be used for holistic communication and cross-domain data exchange. In this paper, we present approaches for mapping existing domain-specific data formats in a more generic schema like the Industry Foundation Classes (IFC). This offers the possibility to develop coordination models for seamless data exchange and improved communication between project stakeholders. Over the last years, the use of Building Information Modeling (BIM) related techniques has been established in the architecture, engineering and construction industry (AEC). A large number of software products is already available that helps engineers to create and validate digital models of buildings. Two critical aspects in BIM are the representation of geometric shapes and semantic attributes within a single data container as well as the seamless collaboration among several domains. Recent efforts have pushed forward the adoption of BIM methods also in the infrastructure sector. Particularly challenging, however, is the domain of railway design and operation (Bukhsh, Hartmann, &amp; Stipanovic, 2016). Many disciplines are involved in the design process of a railway asset, e.g., alignment design, overhead electrification, railway equipment, noise barrier design and structural asset design like bridges and tunnels. The traditional paper-based working procedures have shown to be highly error-prone and inefficient as they do not provide appropriate means for handling the complexity involving a multitude of subdomains. BIMmethods can help to overcome these issues. However, for enabling an integrated design, data exchange is required across all these sub-domains. In recent years, a couple of data formats were developed to store, and exchange data of specific sub-domains involved in a railway design process, but they are often limited to one specific use case. An example of such a format is the data model PlanPro developed by German Railways for representing information about railway equipment components such as signal systems or switch controls (Maschek, Klaus, Gerke, Uminski, &amp; Girke, 2012). For pre-design decisions as well as for maintenance and operation purposes, it is often not necessary to store an exact geometric representation of the railway equipment components. Thus, this exchange format was designed to transfer detailed semantic information and satisfies the specific needs of railway equipment design in Germany. However, there are use cases such as geometric clash detections or visualizations, which require at least a core geometric representation of each component in the three-dimensional space. From a BIM perspective, integrating and linking domain-limited data models is required to realize a higher quality of data exchange and reduce the error-prone and often manually re-</p>
      </abstract>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>1. Introduction</title>
      <p>
        input of already generated data. This causes the need for domain-neutral and also vendor-neutral
data formats such as the Industry Foundation Classes (IFC). This format is well-established for
the exchange of building models. IFC can be used to merge data into a coordinated central
model, which can then be used to perform data validation or clash detection between the
components of federated submodels
        <xref ref-type="bibr" rid="ref17">(Preidel &amp; Borrmann, 2015)</xref>
        .
      </p>
      <p>In this paper, we show how domain-specific data formats can be translated into a more holistic
and flexible IFC model. In detail, we compare the alignment description given in PlanPro with
the available IFC classes to represent alignment axes an IFC model. Furthermore, we describe
the opportunities how components can be placed in an IFC model. This paper also discusses
the flexibility of IFC with respect to dynamic extensions by means of proxy objects and
PropertySets without the need of an official schema extension for an individual domain.
The paper is structured as follows. Section 1.1 gives an overview of related work and related
approaches. We will focus on the PlanPro data model as one specific data model to describe a
possible mapping process of a domain-specific data schema to a more generic data framework.
Two possible approaches converting alignment information and related element placement are
presented in Section 2. Those approaches and their successful implementation show that it is
possible to utilize well-established domain-specific data-formats in a BIM process and hence
prepare the data for further analyses and simulations. We investigate some relevant classes of
the input data and explain approaches to transfer them into the IFC schema. However, the
proposed transformation from PlanPro to an IFC model is limited in some aspects. These issues
are discussed in Section 3 and should be addressed in future research.</p>
    </sec>
    <sec id="sec-2">
      <title>1.1 Related Work</title>
      <p>In the last years, researchers and domain experts have developed a broad range of data models
to carry information about civil infrastructure assets like roads and railways. These studies often
put their focus on carrying semantic information and element descriptions.</p>
      <p>
        Well-known data formats related to the field of railway infrastructure are PlanPro
        <xref ref-type="bibr" rid="ref15">(Maschek et
al., 2012)</xref>
        and RailML
        <xref ref-type="bibr" rid="ref5">(Bosschaart, Quaglietta, Janssen, &amp; Goverde, 2015)</xref>
        . PlanPro was
developed to exchange design data of train control systems, whereas RailML is more focused
on operation issues of rail-bounded vehicles. Both of these formats are based on the Extensible
Markup Language (XML), which is machine- as well as human-readable
        <xref ref-type="bibr" rid="ref6">(Bray et al., 2006)</xref>
        .
For simplicity, these data models are called Information Models (IM) in the following sections.
This term was introduced by
        <xref ref-type="bibr" rid="ref9">Bukhsh, Hartmann, and Stipanovic (2016</xref>
        ) and is used both for
RailML and for PlanPro in this paper. Additionally to the development of the aforementioned
formats, some initiatives investigated in ontologies and Linked-Data approaches
        <xref ref-type="bibr" rid="ref20">(Turck et al.,
2010)</xref>
        .
        <xref ref-type="bibr" rid="ref22">Wunsch et al. (2016)</xref>
        gave a comprehensive outlook on exchange requirements and
possible data formats for these use cases.
      </p>
      <p>
        Besides these separated approaches using IMs as exchange formats,
        <xref ref-type="bibr" rid="ref12">Luteberget and Johansen
(2018)</xref>
        took use of RailML to support and expand a CAD-based design application for railway
safety equipment. They developed a plugin to check the compliance of their design data against
national regulations at runtime. Their approach, however, is not applicable to perform
crossdomain rules, which can require information for assets that cannot be modeled within RailML.
In contrast, the vendor-neutral data format Industry Foundation Classes (IFC) offers the
opportunity to store detailed geometry as well as comprehensive semantic information. Initially
developed in the context of building models, IFC gets more and more extended to the needs of
civil infrastructure assets. Some initiatives investigated on refining the geometric descriptions
for large models, i.e., for exchanging tunnel or bridge models
        <xref ref-type="bibr" rid="ref21">(Vilgertshofer, Jubierre, &amp;
Borrmann, 2016)</xref>
        .
        <xref ref-type="bibr" rid="ref2">Augele (2017)</xref>
        worked out the similarities and differences between RailML
and IFC but did not take the recent IFC extensions for civil infrastructure assets into
consideration.
        <xref ref-type="bibr" rid="ref1">Amann et al. (2014)</xref>
        focused their research on the translation of alignment
models given in several data formats and proposed a common meta-model. Their model became
the base for the description in the IFC data schema extension describing alignments and was
added to the official schema with the version 4.1 published in summer 2016.
The contributions of the abovementioned studies are of high importance because their concepts
of structuring the relevant information in a specific sub-domain are derived from domain
experts’ knowledge. An international standard for exchanging digital models will never be
capable of all regional, local or project-specific features and requirements. Nevertheless,
additional approaches are required to realize cross-domain communication and data exchange
between several project stakeholders since domain-specific formats are often only supported
by a minimal number of specialized applications. Thus, connecting different Information
Models into a vendor-neutral and domain-neutral data format like IFC can help to overcome
this problem. IFC is common in the building sector; thus, there is already a broad range of
model viewer and project management applications supporting IFC import and/or export. These
programs can be used for project communication for building models as well as for civil
infrastructure models. Besides that, converting data from IMs into IFC makes it possible to
perform cross-domain simulations like geometric clash detections between components from
different domains. However, since there is no specific schema extension for railway assets in
the IFC schema yet, some problems occur during the mapping process, which will be discussed
afterward.
      </p>
    </sec>
    <sec id="sec-3">
      <title>1.2 Objectives &amp; Challenges</title>
      <p>
        As shown before, IMs provide a solid base of semantic information about a specific asset.
However, it is not possible to use them for cross-domain communication and simulations.
Therefore, another well-established concept from the building sector can be adapted: Federated
models are used to coordinate domain-specific data during the complete lifecycle of a building.
This approach is already standardized by the British Standard PAS1192-2 as well as by the ISO
standard 19650.
        <xref ref-type="bibr" rid="ref17">Preidel and Borrmann (2015)</xref>
        showed the necessity of working in federated
domain models and merging them into a coordinated model.
      </p>
      <p>
        Coordination and seamless data exchange are unavoidable between all project stakeholders to
optimize planning processes in the construction industry. Non-coordination of issues and design
data can lead to inconsistent data. Thus, a data structure should be chosen, which is capable of
all these aims. Model-based data structures for buildings are often led by a three-dimensional
geometric representation and a grid-based model orientation (axes and storeys in a building).
Contrary to buildings, civil infrastructure models are often line-oriented. The representation of
these alignment curves is typically split into a horizontal description in the  -plane and its
corresponding gradient described by a parametric coordinate  and the height
        <xref ref-type="bibr" rid="ref14">(Markic,
Schlenger, &amp; Bratoev, 2018)</xref>
        . Besides that, there are many cases in a design process of a civil
infrastructure asset, in which the engineer must deal with a large number of semantic values but
can only provide a rough estimation of the product shape. Thus, a data format for railway
coordination models should be chosen, which matches the following requirements:
      </p>
      <p>A line-oriented model orientation and product placements referencing relative to a given
alignment.</p>
      <p>Three-dimensional geometric component representations in any granularity.</p>
      <p>The capacity to carry semantic information assigned to components.</p>
      <p>
        IFC is capable of all these requirements. Therefore, we introduce mapping processes from pure
IMs into the IFC format to remedy the disadvantages of pure IMs. Especially, the strong linkage
between geometric and semantic information in an IFC model enables the user to formulate
complex queries against the generated model. Cross-domain problems benefit a lot from the
direct integration, e.g. by accessing semantic attributes of elements in a particular area or
volume
        <xref ref-type="bibr" rid="ref10">(Daum &amp; Borrmann, 2014)</xref>
        . Design decisions often require a detailed examination of a
complex geometric situation, e.g., mounting several components of different domains on one
pole. Hence, linking the requirements and design options of all sub-domains should be already
investigated in early design phases.
      </p>
      <p>We developed a software prototype to examine the possibilities and limitations of mapping an
IM into an IFC model. In the case study conducted, PlanPro was selected as origin format. The
prototype supports the alignment description and the placement of signals as point-based
objects along a track. Although the validation of the proposed concept is based solely on
PlanPro and provides only the mapping of alignment data and signals, the strategies discussed
in the next sections can also be applied for other components like switches, axle counters, or
platforms.</p>
    </sec>
    <sec id="sec-4">
      <title>2. Approaches</title>
      <p>As there is an increasing number of applications supporting IFC, several options exist to
implement the proposed converter from IMs into IFC models. Currently, several IFC versions
are in use. Thus, a short overview regarding IFC is provided.</p>
    </sec>
    <sec id="sec-5">
      <title>2.1 Industry Foundation Classes (IFC)</title>
      <p>The Industry Foundation Classes (IFC) are an open, vendor-neutral data format, which is
intended to transfer information among different stakeholders across the lifecycle of a built
facility. IFC is mainly focused on the building sector and has many classes to describe elements
from different domains like architecture, structural analysis or HVAC. Although IFC is often
related to data exchange in the planning and execution phase of a project, IFC models can also
be used for operation and maintenance.</p>
      <p>
        With version 4.1, the classes for describing alignments were added to the official schema. These
classes offer the possibility to store information about three-dimensional axes, which are the
base for line-based infrastructure assets e.g., roads, rails and their corresponding structures such
as bridges, tunnels or platforms. Extension projects are currently conducted by buildingSMART
to develop the necessary extensions of the IFC data model. This includes IfcBridge, IfcRoad,
and IfcRail
        <xref ref-type="bibr" rid="ref8">(BuildingSMART international, 2018)</xref>
        . A small number of authoring tools already
supports the import and export of IfcAlignment into or from an IFC file. However, the number
of applications capable to interpret IfcAlignment entities is low compared to a massive range of
tools supporting IFC for building models.
      </p>
      <p>
        Although there is yet no standard defined for infrastructure assets in the IFC schema, many use
cases in an infrastructure planning process are already supported by IFC. The IFC data model
provides flexible extension mechanisms including proxy elements and user-defined property
sets, which enable the user to perform analyses and simulations on these models and allow data
exchange among different stakeholders. It is possible to use the generic
IfcBuildingElementProxy class to model components without a specific definition in the data
schema. IfcBuildingElementProxy is derived from the abstract class IfcProduct. Therefore, it is
capable of carrying a geometric representation as well as every other class derived from
IfcProduct. Classes derived from IfcProduct can be placed inside the model using subclasses
from the abstract class IfcObjectPlacement. A subclass of this class is the IfcLinearPlacement
class, which places components (derived from IfcProduct) relatively to a given curve
        <xref ref-type="bibr" rid="ref7">(BuildingSMART international, 2017)</xref>
        .
      </p>
      <p>Since IFC version 4.1, specialized classes are available to describe typical alignment curves
used for civil infrastructure assets. The IfcAlignment class is derived from IfcProduct and
IfcCurve and is used to place elements along the described curve. Figure 4 shows the relevant
IFC classes used in the implementation approaches.</p>
      <p>An international standard for exchanging digital models will never be capable of all regional,
local or project-specific features and requirements. To mitigate this problem, IFC provides the
concept of property sets. There are many predefined attributes (e.g., for walls, windows or
beams). Additionally, it is possible to create user-defined property sets and append them to
components. Thus, it is possible to extend the IFC data schema for individual requirements,
while at the time the model remains interpretable by any software product that has implemented
IFC reading capababilities.</p>
      <p>In both presented approaches, an additional property set was assigned to every positioned
component in the IFC model, which contains relevant attributes from the given PlanPro file.
Every component in PlanPro has an individual identifier, which is stored as one property within
the PlanPro property set.</p>
    </sec>
    <sec id="sec-6">
      <title>2.2 Implementation strategies</title>
      <p>
        Several initiatives already tried to convert existing object-oriented data structures from or into
the IFC schema. Object-oriented schemas can be seen modeled as graph representations.
Therefore, the main task is finding correlations between both graphs. For identifying links
between IFC and CityGML entities
        <xref ref-type="bibr" rid="ref3 ref4">Beetz &amp; Borrmann (2018)</xref>
        made mainly use of spatial
coincidences, whereas
        <xref ref-type="bibr" rid="ref18">Stouffs, Tauscher, and Biljecki (2018</xref>
        ) applied the triple graph grammar
concept for these formats. The proposed mapping from IMs onto the IFC schema is rather
similar to the latter approach. Both schemas have an object-oriented structure and vary in the
purposes they were intended for. Thus, there is no 1:1 correlation between both graphs, but
there are some classes, which can be correlated. Figure 4 at the end of this document shows
possible links between both data schemas to map point-based products from PlanPro into IFC.
      </p>
      <p>PlanPro
Component Library *.rfa
Type
Number
Name
Sub-Components
*.stl
PreProcessing and
Data Preparation
*.SQLite</p>
      <p>BIM Authoring Tool
(Autodesk Revit)</p>
      <p>XBIM 5 + Modules
IfcAlignmentWriter (*.dll)
IfcAlignmentWriter (*.dll)
PlanProAlignmentExtractor (*.dll)
PlanProElementExtractor (*.dll)</p>
      <p>IFC 2x3
IFC 4
coordinate-based placement
IFC 2x3
IFC 4
IFC 4x1
Coordinate-based Placement or Linear
Placement</p>
      <p>Existing BIM authoring tools offer much flexibility when creating new element groups and
their respective geometry. Therefore, it is one option to use an existing application and extend
it by a plugin that provides the required information in the modeling environment. For this case
study, Autodesk Revit as a well-established BIM authoring application was selected. It has a
powerful “family” editor to create new elements and can be extended using its Advanced
Programming Interface (API).</p>
      <p>A significant disadvantage of this approach is the overall dedication of BIM modeling tools for
buildings and the limited support for linear infrastructure facilities. While applications like
Revit organize their elements in grids and levels, civil infrastructure assets are mainly
referenced to an alignment curve. Thus, Revit itself does not support the export of IfcAlignment
Entities, but only positioning expressions in cartesian coordinate systems.</p>
      <p>In the second approach, the framework XBim was selected to implement a stand-alone
converter, which makes use of the IFC entities suitable to describe alignment curves.
Additionally, the way of storing the placement of an individual component is addressed in
detail. The next sub-sections provide an insight into the implementation process and limitations
of the proposed concepts.</p>
    </sec>
    <sec id="sec-7">
      <title>2.3 Product Catalogue</title>
      <p>In the proposed concept, all available IMs are transformed into a 3D BIM representation to
support coordination and informed decision making. Neither the components in RailML nor in
PlanPro posses geometric descriptions in the form of Boundary Representations (BRep) or
Constructive Solid Geometry (CSG). For some components, however, PlanPro refers to
arrangement drawings. These drawings are independent of the manufacturer, and thus can be
used for vendor-neutral tendering. Because of their generic character, a component library
collects the geometric shape and parameterize these representations.</p>
      <p>Every component has a certain drawing number and can have several types. All types have a
slightly similar shape. Thus, it is possible to parameterize their geometric representation.
Additionally, the arrangement drawings offer some additional semantic parameters like
placement rules (e.g., the required distances to overhead electrification poles), which are
covered by the library as well.</p>
      <p>The generated family library establishes the base for both approaches. Due to restrictions on
publishing and sharing these arrangement drawings, the component library is not big enough to
cover all possible signal types. Hence, a cuboid is used as a geometric representation for a
component if there is no reference to an arrangement drawing, or the arrangement drawing was
not accessible.</p>
    </sec>
    <sec id="sec-8">
      <title>2.4 Using PlanPro Data in a BIM Authoring Tool</title>
      <p>Since Revit is not capable of any typical alignment format yet, a preprocessing tool was
developed to prepare the required data of a PlanPro model for the use inside a Revit Project.
Although Revit is not intended to be a modeler for railway equipment design, the family engine
and the possibility to work with large property sets make it adequately generic and suitable for
the proposed converter.</p>
      <p>
        All coordinates of the alignment segments in a PlanPro file are described in a world coordinate
system (DB Ref as a specialized system of the well-known Gauss-Krüger-System). Since Revit
cannot handle large coordinate values, a preprocessing tool translates the given world
coordinates into a local project coordinate system1
        <xref ref-type="bibr" rid="ref13 ref19 ref3 ref4">(Markič, Donaubauer, &amp; Borrmann, 2018)</xref>
        .
In a PlanPro instance model, the placement of components is given by the track it belongs to,
the station kilometer and the lateral offset to the rail axis. Revit itself cannot process this kind
of placement description; thus, the preprocessing tool calculates the (projected) cartesian
coordinate. Thus, the BIM authoring tool itself is mainly used to place the right family inside
into the modeling environment and sets the required semantic attributes. The preprocessing
approach also offers the flexibility to use the prepared data in other tools.
      </p>
      <p>
        After finishing the modeling process, the IFC export interface of Revit was used to create the
desired IFC model. Although the export interface offers the possibility to adjust some
preferences
        <xref ref-type="bibr" rid="ref19 ref3 ref4">(Trzeciak &amp; Borrmann, 2018)</xref>
        , it cannot create IFC models using IfcAlignment and
IfcLinearPlacement entities.
      </p>
      <p>The resulting IFC model in the versions 2x3 and 4 can be displayed in all common IFC viewer
applications.
1 It must be noted that this local coordinate system is still a projected one with the respective distortions that must be taken into account.</p>
    </sec>
    <sec id="sec-9">
      <title>2.5 Using XBim Framework for a Refined IFC Model</title>
      <p>
        As discussed above, the aim of mapping an existing IM into an IFC model can be solved with
the first approach using existing modeling software. The resulting model contains signals with
an individual geometric representation depending on the given arrangement drawings and
semantic attributes. The alignment is modeled with a volumetric boundary representation as
well. However, the plugin-approach leads to the loss of a direct relationship between the
alignment curve and the implicit placement indication (station and axis reference) of all
elements since cartesian coordinates now describe the position of any element.
The second approach remedies the disadvantage of using Cartesian Coordinates for every
individual element placement. Instead, each element’s position is described relative to an
alignment curve in the resulting IFC model. Once the alignment curve in the  -plane is
modeled, it is possible to locate an element by a new coordinate system ( ,  ) in which 
describes the stationing axis and  the height at a specific station
        <xref ref-type="bibr" rid="ref14">(Markic et al., 2018)</xref>
        . With
additional offset parameters in the longitudinal, lateral and vertical direction the exact
component placement is defined.
      </p>
      <p>To realize the second approach, we implement a direct mapping to IFC without using an
existing BIM authoring tool in the toolchain anymore. Instead, the XBim Toolkit is employed
to create the desired IFC model using IfcAlignment and the linear positioning concept along the
alignment curve according to ISO 19148. XBim is an open-source framework for importing,
editing and exporting IFC models and is licensed under the CDDL Open Source License
(“xBIM Toolkit,” 2019).</p>
      <p>The required methods are split up in several modules to make the code as reusable as possible.
This concept enables the developers to extend the implemented prototype for further
requirements or other applications, which need similar methods. The modules
IfcAlignmentWriter and IfcElementWriter are entirely decoupled from the libraries extracting
data from a given PlanPro model. The modules PlanProAlignmentExtractor and
PlanProElementExtractor extract the necessary data and insert suitable IFC entities into a given
IFC model.
As shown in Figure 1, the created component library is used in both presented approaches.
Since Revit families are stored in the proprietary *.rfa format, the contained geometric and
semantic data are not accessible from applications outside the Autodesk product family. Thus,
an export of the geometry into the open geometry format “STereoLithography” (STL) was
required. STL stores a given body as a triangulated boundary representation and is, therefore,
an explicit geometry representation. It can easily be converted into an IFC geometry
representation like IfcTriangulatedFaceSet.</p>
      <p>
        Model View Definitions (MVDs) are used to define a subset of the complete IFC schema and
add additional requirements such as geometric representation method or necessary attributes
        <xref ref-type="bibr" rid="ref19 ref3 ref4">(Beetz, Borrmann, &amp; Weise, 2018)</xref>
        . The chosen IFC entities used in the standalone approach
are based on the Reference View of IFC 4. Unfortunately, there is yet no official MVD for
IfcAlignment and IfcLinearPlacement entities. However, the generated IFC models can be
opened and visualized by several applications. We tested the alignment data with the
commercial software ProVI
        <xref ref-type="bibr" rid="ref16">(Obermeyer, 2018)</xref>
        and the FZK-Viewer
        <xref ref-type="bibr" rid="ref11">(Karlsruher Institut für
Technologie, 2019)</xref>
        . The latter one is already capable of processing IfcLinearPlacement
entities.
      </p>
    </sec>
    <sec id="sec-10">
      <title>3. Discussion</title>
      <p>The paper has presented two different approaches for integration domain-specific information
models into a common model representing the overall railway project by means of the vendor
neutral data format IFC. Both approaches presented the feasibility to convert domain-specific
data models into an IFC model. By using the IfcAlignment classes as well as the
IfcLinearPlacement concept, it is possible to set up an IFC model based in semantically
powerful IMs, which uses alignment expressions for the overall project orientation. The concept
of property sets enriches the data schema and provide the flexibility to adapt the model to
specific needs. The presented approach of mapping semantically powerful but geometrically
sparse IMs into IFC can be applied to many existing data concepts which are used for
domainspecific exchange scenarios. By converting them into IFC models, it is now possible to
coordinate elements and their related attributes across different domains. However, the
translation and analysis of PlanPro have unveiled significant limitations: The geometric
granularity of arrangement drawings can only provide an estimation of the component’s shape
to be built in the real world. Thus, the modeled geometry can differ from real product shape in
the built environment, and this must be considered when performing simulations based on the
shown IFC models. In order to use the created models on site, refinements of the used geometric
representations are inevitable. In conclusion, the proposed mapping process is possible and
enriches the concept of seamless collaboration and communication along different domains in
a project.</p>
      <p>IMs often provide fictive components, which do not have any geometric representation but are
of high importance for the design process. Such concepts cannot be directly covered by the IFC
schema yet but are depictable in an IFC model. The use of placeholder geometries like cuboids
or other simple geometries is one possible solution for such issues. The obligation that every
class derived from IfcProduct must have a ProductRepresentation is restrictive in that point but
solvable.</p>
      <p>User-defined property sets enrich an IFC model due to project or domain-specific requirements.
However, the possible data types for property values are limited to simple numeric or textual
types and Booleans. An IM can have much more specialized subclasses, which structure
relevant attributes and relationships better than a simple IfcPropertySet can do. Thus, use cases
are conceivable that could be better served with the original IM. The focus of our research was
the preparation of data for project-wide communication rather than supporting advanced and
specialized use cases. Another option inside the IFC schema is the use of element assemblies,
which should be addressed in future research.</p>
      <p>Due to the yet missing standardization of railway components in IFC, it is complex to define
filters or to process the models in subsequent simulations. It is possible to filter specific
components by their names, but the defined filters are dependent on modifiable attributes. A
better approach is the detection of elements by their element type. However, the necessary
changes in the implementation described in Section 2 will not have a massive impact: most of
the IFC classes representing railway equipment components will probably derive from
IfcProduct.</p>
      <p>In summary, the proposed conversion from IMs into the IFC format can be used to utilize
domain-specific information in federated models and combine those in a coordination model.
However, limitations exist with respect to the granularity and expressiveness of the converted
data.</p>
    </sec>
    <sec id="sec-11">
      <title>4. Summary</title>
      <p>In this paper, we showed different approaches to convert domain-specific data formats into a
more generic data model. These concepts offer much flexibility when querying data in a model
and using it for in cross-domain simulations.</p>
    </sec>
    <sec id="sec-12">
      <title>5. Acknowledgments</title>
      <p>This study has been performed in the frame of the RIMcomb project. We gratefully acknowledge
the support of the Bavarian Research Foundation for funding this project.
cad.de/
xBIM Toolkit. (2019). Retrieved February 12, 2019, from http://docs.xbim.net/
Basis_Objekt
Punkt_Objekt</p>
      <p>Punkt_Objekt_TOP_Kante</p>
      <p>Punkt_Objekt_Strecke
ID_TOP_Kante
Abstand
Wirkrichtung
Seitlicher Abstand
ID_Signal_Befestigung
ID_Regelzeichnung
Befestigung_Art
Hoehe_Fundamentoberkante
Obere_Lichtpunkthoehe</p>
      <p>Regelzeichnung
IfcObjectDefinition
IfcObject
IfcElement
Tag
IfcBuildingElement
IfcBuildingElementProxy
PredefinedType</p>
      <p>IfcProductRepresentation</p>
      <p>IfcObjectPlacement
IfcGridPlacement
PlacementLocation
PlacementRefDirection</p>
      <p>IfcLinearPlacement</p>
      <p>IfcLocalPlacement</p>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          <string-name>
            <surname>Amann</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Jubierre</surname>
            ,
            <given-names>J. R.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Flurl</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          (
          <year>2014</year>
          ).
          <article-title>An alignment meta-model for the comparison of alignment product models</article-title>
          .
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          <string-name>
            <surname>Augele</surname>
            ,
            <given-names>V.</given-names>
          </string-name>
          (
          <year>2017</year>
          ).
          <article-title>Comparative Analysis of Building Information Modelling (BIM) and RailTopoModel/railML in View of their Application to Operationally Relevant Railway Infrastructure</article-title>
          . Retrieved from https://www.railml.org/en/public-relations/scientificpapers.html?file=files/download/papers/230117_TUDresdenAugele_TermPaperCompari sonRailTopoModelIFCrailML.pdf
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          <string-name>
            <surname>Beetz</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          (
          <year>2018</year>
          ).
          <article-title>Benefits and limitations of linked data approaches for road modeling and data exchange</article-title>
          . In Workshop of the European Group for Intelligent Computing in Engineering (pp.
          <fpage>245</fpage>
          -
          <lpage>261</lpage>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          <string-name>
            <surname>Beetz</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Weise</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          (
          <year>2018</year>
          ).
          <article-title>Process-Based Definition of Model Content</article-title>
          .
          <source>In Building Information Modeling</source>
          (pp.
          <fpage>127</fpage>
          -
          <lpage>138</lpage>
          ). Springer.
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          <string-name>
            <surname>Bosschaart</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Quaglietta</surname>
            ,
            <given-names>E.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Janssen</surname>
            ,
            <given-names>B.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Goverde</surname>
            ,
            <given-names>R. M. P.</given-names>
          </string-name>
          (
          <year>2015</year>
          ).
          <article-title>Efficient formalization of railway interlocking data in RailML</article-title>
          .
          <source>Information Systems</source>
          ,
          <volume>49</volume>
          ,
          <fpage>126</fpage>
          -
          <lpage>141</lpage>
          . https://doi.org/10.1016/j.is.
          <year>2014</year>
          .
          <volume>11</volume>
          .007
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          <string-name>
            <surname>Bray</surname>
            ,
            <given-names>T.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Paoli</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Sperberg-McQueen</surname>
            ,
            <given-names>C. M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Maler</surname>
            ,
            <given-names>E.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Yergeau</surname>
            ,
            <given-names>F.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Cowan</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          (
          <year>2006</year>
          ).
          <article-title>Extensible Markup Language (XML)</article-title>
          . https://doi.org/DOI:
        </mixed-citation>
      </ref>
      <ref id="ref7">
        <mixed-citation>
          <article-title>BuildingSMART international</article-title>
          . (
          <year>2017</year>
          ).
          <article-title>Ifc 4x1 Final Documentation</article-title>
          .
          <source>Retrieved February 13</source>
          ,
          <year>2019</year>
          , from http://www.buildingsmart-tech.org/ifc/IFC4x1/final/html/
        </mixed-citation>
      </ref>
      <ref id="ref8">
        <mixed-citation>
          <article-title>BuildingSMART international</article-title>
          . (
          <year>2018</year>
          ).
          <article-title>Infrastructure: Ongoing Projects</article-title>
          .
          <source>Retrieved July 15</source>
          ,
          <year>2018</year>
          , from https://www.buildingsmart.org/standards/activities/
        </mixed-citation>
      </ref>
      <ref id="ref9">
        <mixed-citation>
          <string-name>
            <surname>Bukhsh</surname>
            ,
            <given-names>Z. A.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Hartmann</surname>
            ,
            <given-names>T.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Stipanovic</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          (
          <year>2016</year>
          ).
          <article-title>From analysis of information needs towards an information model of railway infrastructure</article-title>
          . In 23rd International Workshop of the European Group for Intelligent Computing in Engineering, EG-ICE
          <year>2016</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref10">
        <mixed-citation>
          <string-name>
            <surname>Daum</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          (
          <year>2014</year>
          ).
          <article-title>Processing of topological BIM queries using boundary representation based methods</article-title>
          .
          <source>Advanced Engineering Informatics</source>
          ,
          <volume>28</volume>
          (
          <issue>4</issue>
          ),
          <fpage>272</fpage>
          -
          <lpage>286</lpage>
          . https://doi.org/10.1016/j.aei.
          <year>2014</year>
          .
          <volume>06</volume>
          .001
        </mixed-citation>
      </ref>
      <ref id="ref11">
        <mixed-citation>
          <source>Karlsruher Institut für Technologie</source>
          .
          <article-title>(</article-title>
          <year>2019</year>
          ). KIT - FZKViewer.
          <source>Retrieved February 13</source>
          ,
          <year>2019</year>
          , from https://www.iai.kit.edu/1648.php
        </mixed-citation>
      </ref>
      <ref id="ref12">
        <mixed-citation>
          <string-name>
            <surname>Luteberget</surname>
            ,
            <given-names>B.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Johansen</surname>
            ,
            <given-names>C.</given-names>
          </string-name>
          (
          <year>2018</year>
          ).
          <article-title>Efficient verification of railway infrastructure designs against standard regulations</article-title>
          .
          <source>Formal Methods in System Design</source>
          ,
          <volume>52</volume>
          (
          <issue>1</issue>
          ),
          <fpage>1</fpage>
          -
          <lpage>32</lpage>
          . https://doi.org/10.1007/s10703-017-0281-z
        </mixed-citation>
      </ref>
      <ref id="ref13">
        <mixed-citation>
          <string-name>
            <surname>Markič</surname>
          </string-name>
          , Š.,
          <string-name>
            <surname>Donaubauer</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          (
          <year>2018</year>
          ).
          <article-title>Enabling Geodetic Coordinate Reference Systems in Building Information Modeling for Infrastructure</article-title>
          .
          <source>In Proceeding of the 17th International Conference on Computing in Civil and Building Engineering</source>
          , Tampere, Finland.
        </mixed-citation>
      </ref>
      <ref id="ref14">
        <mixed-citation>
          <string-name>
            <surname>Markic</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Schlenger</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Bratoev</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          (
          <year>2018</year>
          ).
          <article-title>Tangible Alignment Design</article-title>
          . In 30. Forum Bauinformatik, Weimar, Germany.
        </mixed-citation>
      </ref>
      <ref id="ref15">
        <mixed-citation>
          <string-name>
            <surname>Maschek</surname>
            ,
            <given-names>U.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Klaus</surname>
            ,
            <given-names>C.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Gerke</surname>
            ,
            <given-names>C.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Uminski</surname>
            ,
            <given-names>V.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Girke</surname>
          </string-name>
          ,
          <string-name>
            <surname>K.-J.</surname>
          </string-name>
          (
          <year>2012</year>
          ).
          <article-title>PlanPro - Durchgängige elektronische Datenhaltung im ESTW-Planungsprozess</article-title>
          .
          <source>Signal + Draht</source>
          ,
          <volume>104</volume>
          ,
          <fpage>22</fpage>
          -
          <lpage>26</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref16">
        <mixed-citation>
          <string-name>
            <surname>Obermeyer.</surname>
          </string-name>
          (
          <year>2018</year>
          ). ProVI: Home: Trassierungssoftware für Tiefbau, Bahnplanung, Straßenplanung, Kanal, Vermessung.
          <source>Retrieved May 1</source>
          ,
          <year>2018</year>
          , from https://www.provi-
        </mixed-citation>
      </ref>
      <ref id="ref17">
        <mixed-citation>
          <string-name>
            <surname>Preidel</surname>
            ,
            <given-names>C.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          (
          <year>2015</year>
          ).
          <article-title>Seamless Integration of Common Data Environment Access into BIM Authoring Applications : the BIM Integration Framework</article-title>
          .
        </mixed-citation>
      </ref>
      <ref id="ref18">
        <mixed-citation>
          <string-name>
            <surname>Stouffs</surname>
            ,
            <given-names>R.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Tauscher</surname>
            ,
            <given-names>H.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Biljecki</surname>
            ,
            <given-names>F.</given-names>
          </string-name>
          (
          <year>2018</year>
          ).
          <article-title>Achieving Complete and Near-Lossless Conversion from IFC to CityGML</article-title>
          . ISPRS
          <source>International Journal of Geo-Information</source>
          ,
          <volume>7</volume>
          (
          <issue>9</issue>
          ), 355. https://doi.org/10.3390/ijgi7090355
        </mixed-citation>
      </ref>
      <ref id="ref19">
        <mixed-citation>
          <string-name>
            <surname>Trzeciak</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          (
          <year>2018</year>
          ).
          <article-title>Design-to-design exchange of bridge models using IFC : A case study with Revit and Allplan</article-title>
          .
        </mixed-citation>
      </ref>
      <ref id="ref20">
        <mixed-citation>
          <string-name>
            <surname>Turck</surname>
          </string-name>
          , F. De,
          <string-name>
            <surname>Ongenae</surname>
            ,
            <given-names>F.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Verstichel</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Dhaene</surname>
            ,
            <given-names>T.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Dhoedt</surname>
            ,
            <given-names>B.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Dings</surname>
            ,
            <given-names>P.</given-names>
          </string-name>
          , …
          <string-name>
            <surname>Loeve</surname>
            ,
            <given-names>L.</given-names>
          </string-name>
          (
          <year>2010</year>
          ).
          <article-title>Efficient data integration in the railway domain through an ontology-based methodology</article-title>
          . Transportation Research Part C: Emerging Technologies,
          <volume>19</volume>
          (
          <issue>4</issue>
          ),
          <fpage>617</fpage>
          -
          <lpage>643</lpage>
          . https://doi.org/10.1016/j.trc.
          <year>2010</year>
          .
          <volume>10</volume>
          .003
        </mixed-citation>
      </ref>
      <ref id="ref21">
        <mixed-citation>
          <string-name>
            <surname>Vilgertshofer</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Jubierre</surname>
            ,
            <given-names>J. R.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Borrmann</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          (
          <year>2016</year>
          ).
          <article-title>IfcTunnel-a proposal for a multiscale extension of the IFC data model for shield tunnels under consideration of downward compatibility aspects</article-title>
          .
          <source>In 11th European Conference on Product and Process Modelling</source>
          ,
          <string-name>
            <surname>ECPPM</surname>
          </string-name>
          <year>2016</year>
          (pp.
          <fpage>175</fpage>
          -
          <lpage>182</lpage>
          ). Retrieved from https://www.scopus.com/inward/record.uri?eid=
          <fpage>2</fpage>
          -
          <lpage>s2</lpage>
          .
          <fpage>0</fpage>
          -
          <lpage>85016608612</lpage>
          &amp;partnerID=
          <volume>40</volume>
          &amp;md5=
          <fpage>b7dbf8d0bb46d0c0b4ddea7388a1c2c1</fpage>
        </mixed-citation>
      </ref>
      <ref id="ref22">
        <mixed-citation>
          <string-name>
            <surname>Wunsch</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Lehnert</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Krimmling</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          , &amp;
          <string-name>
            <surname>Easton</surname>
            ,
            <given-names>J.</given-names>
          </string-name>
          (
          <year>2016</year>
          ). Datenformate,
          <article-title>-modelle und - konzepte für den Eisenbahnbetrieb</article-title>
          . Der Eisenbahningenieur, (November),
          <fpage>8</fpage>
          -
          <lpage>14</lpage>
          .
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>