Towards an Open Building Information Model Server
Report on the progress of an open IFC framework
Jakob Beetz, Ruben de Laat1, Léon van Berlo1 and Pim van den Helm1
Eindhoven University of Technology
2
TNO Built Environment and Geosciences
Den Dolech2 5800 MB Eindhoven
The Netherlands
[email protected], [email protected], [email protected], [email protected]
Key words: Industry Foundation Classes, Model Server, Building Information Modelling
Abstract: In this paper we report on the progress on an Open Source model server for
collaboration between several stakeholders based on models of the Industry
Foundation Classes. After a general introduction on the topic of model servers,
a review on past and present research and development efforts in the field, we
illustrate the general architecture of the framework and report on existing and
planned features.
1. INTRODUCTION
Building Information modelling (BIM) – although it has been on the
research agenda for decades now – has received increasing public interest by
the building and construction industry in recent years. From the early years
of information that was modelled following the approaches defined in the
STandard for the Exchange of Product data (STEP) onwards, the need for
common models shared by stakeholders in an engineering project has been
identified as one of the crucial concepts for a successful increase in
efficiency of collaboration in the building and construction industry (Hannus
et al 2003, Froese 2003). The ISO-certified Industry Foundation Classes
(IFC) which have been modelled in close relation of the STEP initiative,
have seen a number of research projects and approaches to design a
1
2 DDSS 2010
dedicated server for AEC/FM models. However, to date the use of IFC-
based model servers is still not a widely adopted business practice and only a
few experimental real-world light-house projects have made use of model-
based, heterogeneous collaboration in the design, construction and
maintenance of buildings (Plume and Mitchell, 2007, Lê et al 2006,) .
Available commercial model server applications have been developed
mainly focusing on industries other than building and construction. The
amount of resources that are spent on model maintenance in other model-
based engineering processes is significantly higher than in the AEC/FM
field. Here, generally agreed upon implicit knowledge in the form of local
conventions, best practices and rules of thumb often prevents the careful
capturing of ‘the obvious’ in an explicit form in the model. While the main
form to convey information in the field to date still is the technical drawing
and its interpretation by humans, benefits of automation can only be
achieved with good underlying machine-readable data models.
Apart from political and economical issues that have been illustrated by
(Howard and Björk, 2008, Kiviniemi et al 2008) among others, technical
reasons can be identified for low acceptance of IFC models in the field:
Conceptually, their lack of unique identifiers on the level of objects makes
the management of different versions challenging. Even though higher level
objects descending from IfcRoot are equipped with a globally unique
identifier (GUID), the bulk of information – their geometric and topologic
representations – is anonymous and has to be compared with complex,
content-based operations. Even though research projects have aimed at
work-arounds such as heuristic graph-topology comparisons (Weise 2006),
implementations are still not available for end-users.
By providing the Open Source (GPL v3) platform bimserver.org platform
we are aiming at a number of goals:
a) To provide a customizable environment that enables researchers and
developers to re-use, modify and adapt implementations of low-level
tasks such as EXPRESS schema and instance parsing, persistency
management and visualization.
b) Even though the platform is still in a development stage, we would
like to encourage SMEs to make acquaintance and test business
processes based on model server concepts by providing a platform
that is easy to install and use and that can be run on standard PC
hardware.
c) To provide room for experimentation with new business models such
as BIM management consultancy and project hosting.
Towards an Open Building Information Model Server 3
2. EXISTING WORK
In the past, a number of individual and collaborative efforts have been
initiated to provide common platforms and frameworks for the management
of shared IFC models. Several projects have investigated different
approaches for persistency-related research questions for general STEP-
based and IFC models, and a number of prototypical implementations have
been reported upon, none of which seems to have been taken up by a broader
community. The family of BLIS, IMSvr and SABLE (Adachi 2002,
Kiviniemi et al 2005) initiatives which were among the most referred ones
unfortunately have been discontinued. However, experiences, and concepts
that have been made and designed during these projects have been valuable
milestones for further research and more recent developments such as the
Information Delivery Manual (IDM) and the Model View Definition (MVD)
initiatives. The IMSvr project followed the approach of a relational data base
developed from the IFC EXPRESS schema and adding a XML SOAP API
calls on top that exposed server functionality to the web. In the SABLE
project, a domain specific API was added that allowed the retrieval of sub
models categorized by disciplines. In addition to these ‘hard-wired’ query
calls, Adachi developed a query language that allows the retrieval of
complex model subsets, which has been reported in (Adachi 2003), which
we consider an interesting candidate to be implemented in the platform
presented here.
3. FRAMEWORK ARCHITECTURE
The architecture of the bimserver.org framework is based on a number of
flexible model based layers. At the core is an original model schema defined
in ISO 10303 EXPRESS. Although subsequently generated meta models and
code fragments have been manually optimized for the use of current IFC
schema in the 2x3 version, other model schemas, such as earlier or future
versions of the IFC model or other models such as the ISO 12006
“International Framework for Dictionary” (IFD) could in principle be used
with some additional adaption. From this initial model layer, a Meta Object
Facility (MOF) described as an XML Metadata Interchange (XMI) model is
generated. The specific implementation is based on the Eclipse Modelling
Framework (EMF). The current state of implementations is limited to basic
model information and uses straight-forward mappings between EXPRESS
and EMF. Basic Information such as EXPRESS TYPE, SELECT, ENTITY,
and attributes are mapped to their respective XMI/EMF counterparts classes,
interfaces, and attributes. Particular differences in each modelling languages’
4 DDSS 2010
characteristics and features require creative software engineering solutions.
Difficulties and specifics of such a transformation have been reported by
other authors in the past (Arnold and Podehl, 1999, Lubell et al 2004, Krause
and Kaufmann 2007). One such prominent example is the necessity to box
simple TYPEs which allow the extension/aliasing of simple data types
(integer, float, bool etc.) in EXPRESS into fully fledged classes in EMF and
the subsequently generated Java code, since inheritance from primitive data
types is disallowed in this environment. Other necessary work-around
solutions concern inverse relationships which occasionally cannot be
automatically transformed due to inheritance issues. Even though all data
found in part 21 STEP Physical File Format (SPFF) or part 28 ifcXML
instance models can be processed by our framework, at present, the
conversion from STEP to UML/XMI/EMF is far from feature complete.
Advanced modelling features such as derived values WHERE rule
constraints and STEP FUNCTIONs are striped off of the initial schemas, we
are very interested at looking into possible solutions such as the Object
Constraint Language (OCL) to preserve the richness of the IFC model in
other modelling languages. Even if a complete automation of a
transformation between these rule language might not be achieved, manual
conversions seem feasible, since only few occasion are found in the IFC
schemas and their change over the last version revisions has been minimal.
Based on the EMF meta model, a code generator is included in the
platform that creates early-binding Java classes from the model. Although
the number of alternative libraries and implementations in different
languages is currently growing, we would like to encourage fellow
researchers and developers to use parts of the bimserver.org platform as a
convenient development tool for their own purposes on this level.
To achieve the main goal of the bimserver.org project – a free, Open
Source model server for IFC information - a persistency layer is added on
top of the layers described earlier. As a reference implementation, the open
source Berkeley-DB has been chosen as its concrete realization, but other
database implementations – both key-values-based or traditional relational
databases – might be used as a replacement. Earlier attempts to manually
craft object-relational and object-oriented database mappings from the model
or to use automated frameworks such Hibernate have not been successful
due either to structural mapping issues or performance and scalability
limitations. Similar issues have been reported by (Nour 2009).
Concrete instances of IFC objects that are uploaded to the model server
system are stored on a per-object level. A unique object key is generated for
every new instance. This makes it possible to unambiguously identify,
retrieve and manage even those instances that have not been equipped with a
Globally Unique Identifier (GUID) by its source application as required by
Towards an Open Building Information Model Server 5
the IFC schema only for a small fraction of entire model population. This
mechanism is a first step to track changes between different model versions
over its entire lifecycle. However a number of problems have yet to be
solved before fully overcoming some of the inherent structural shortcomings
of STEP-based data models concerning their versioning and the integration
of multiple domain specific models stemming from different applications.
We are investigating approaches suggested by (Weise 2006) and (Arthaud
and Lombardo, 2006) and an approach based on a spatial heuristic that is in
the early stages of planning. A sub-project mechanism implemented in the
database core allows the creation of model branches and design variants.
Upon this database layer a number of operations – referred to a as
‘actions’ – are implemented which are exposed in the form of SOAP calls.
These operations include general database administration routines such as
user management and access right management. Query operations include
the possibility to request all instances of a specific (EXPRESS entity) type,
objects descending from the IfcRoot entity identified by a Globally Unique
Identifier (GUID) and more sophisticated queries in the form of Java code
snippets that can be uploaded and compiled at server runtime. Configurable
and flexible filter mechanisms that allow the explicit inclusion or exclusion
of object types or specific instances will be extended in future versions
aiming at MVD functionality.
4. FUNCTIONALITY OF THE CLIENT REFRENCE
IMPLEMENTATION
Based upon the aforementioned SOAP API which is published by a
WSDL interface, clients interacting with the server can be written by third
party developers. A reference implementation has been created in the form
of a web-browser based client which is included in the binary distribution of
the server platform. With this client, basic interactions such as the creation
and management of users and their access rights, projects and subprojects
can be executed by non-technical end-users. Import and export filters that
have been added to the system include ifcXML, KMZ, Collada and
CityGML. Basic navigation and inspection of the model content is provided
by a structural browser that presents a revision sorted by the corresponding
IFC Entity types. A graphical visualization module that is based on Google’s
O3D format allows the interactive navigation of complete model revisions or
the result of queries model queries issued through the client. While at present
all geometrical operations are facilitated by a multi-platform version of
Bonsma’s IFCEngine.dll (TNO IFC engine, 2002) that operates on the server
side, we are looking for a tighter integration of alternative geometry engines
6 DDSS 2010
in future. Also based upon this external engine, clash detection functionality
has been added that allows the investigation of potential design and
engineering issues during a construction project.
5. PRELIMINARY USER FEEDBACK
The bimserver.org platform has received quite some attention both by an
international community of developers through mailing lists and the end-
users such as architects, engineers and consultants. On several occasions
works-shops and presentation have been organized, targeting mainly the
Dutch landscape of software developers and design and construction
practitioners. Based on feedback provided by the community which is very
positive in general future developments, refinements and extensions are
being planned. Among the important concerns of end-users is the usability of
the client interface. Most frequently requested features are the improvements
of model merging capabilities, finer control and tracking mechanisms of
different users working on the model and advanced model consistency
checkers.
Features that are appreciated the most include the ability create dynamic
links to Google Earth that allows clients to monitor a project’s progress at
any given moment. A compression feature that reduces bandwidth is
similarly popular as is the automated update of project changes through RSS
feeds.
Several developers across different domains and countries are currently
making experiences with tailoring the platform to their own needs, and their
feedback and contribution are very valuable to the advancement of the
project.
6. SUMMARY, DISSCUSSION AND FUTURE WORK
In this paper we have reported on the ongoing research and development
of an Open Source IFC Model server based on a scalable, extendable and
multi-platform architecture. We illustrated our approach using a modular
architecture with a key-value based database and described functionality of
the server that is exposed through web service interfaces. With outlined
functionality that has been implemented in a reference client and reported on
feedback we received both from developers as well as end users.
Using an extensive suite of test models that in some cases exceed one
million entity instances, we experienced the platforms capability of handling
real-world models stemming from a spectrum of different applications with
Towards an Open Building Information Model Server 7
IFC support. Future investigations will include benchmarks and comparisons
to existing applications. The bimserver.org platform is not intended to be a
full alternative to the few existing commercial solutions, but we hope that it
may serve as an experimentation environment both for end-users as well as
the research and development community and especially the work of the
buildingSMART organization.
Although at present there are many possible areas for enhancement, we
hope a broader community is developing an interest in the project and
considers the platform for own collaboration projects, developments, or
extensions. In future, we would like to enhance the merging functionality
and further work on queries to allow MVD partial model extraction. An
interface more convenient for non-technical end users and further
enhancements of the revision control on an object level are additional items
on the project’s agenda.
7. REFERENCES
Arthaud, G., and J. Lombardo. 2006. “Automatic Semantic Comparison of STEP Product
Models”. Innovations in Design & Decision Support Systems in Architecture and Urban
Planning, Proceedings of the 2006 conference on Design and Decision support systems.
447–463.
Adachi, Y., 2002. “Overview of IFC model server framework”. Proc. of the 2002 conference
EWork and EBusiness in Architecture, Engineering and Construction, 367.
Adachi, Y., 2003. “Overview of partial model query language”. In Proc. of the 10th Int.
Conf. on Concurrent Engineering. pp 549-555
Arnold, F. and G. Podehl. 1999. “Best of Both Worlds-A mapping from EXPRESS-G to
UML. The Unified Modeling Language.UML” in Bézivin, J. and P.-A. Muller (eds.) The
Unified Modeling Language. «UML»’98: Beyond the Notation’98: Beyond the Notation,
514–514.
Froese, T., 2003. “Future directions for IFC-based interoperability”. ITCon, (8). pp 231-246.
Hannus, M., M. Blasco, M. Bourdeau, M. Böhms, G. Cooper, F. Garas, T. Hassan, et al.
2003. “Construction ICT roadmap”. Public report of ROADCON project IST-2001-37278,
Deliverable WP5/ D, 52, 30.
Helm, P. van den, M. Böhms and L. van Berlo. 2010.”IFC-based clash detection for the open-
source BIMserver”. To appear in the Proceedings of the International Conference on
Computing in Civil and Building Engineering ICCCBE 2010
Howard, R., and B.C. Björk, 2008. “Building information modelling - Experts' views on
standardisation and industry deployment”. Advanced Engineering Informatics 22, no. 2
(April): 271-280.
Kiviniemi, A., M. Fischer and V. Bazjanac. 2005. “Integration of multiple product models:
Ifc model servers as a potential solution”. Proc. of the 22nd CIB-W78 Conference on
Information Technology in Construction.
Lê, M.A.T., F. Mohus, O.K. Kvarsvik, M. Lie. 2006. “The HITOS project-A full scale IFC
test”. eWork and eBusiness in architecture, engineering and construction: proceedings of
8 DDSS 2010
the 6th European Conference on Product and Process Modelling, 13-15 September 2006,
Valencia, Spain. p. 191.
Lubell, J, R.S. Peak, V. Srinivasan and S.C. Waterbury 2004. “STEP, XML, and UML:
Complementary technologies. In Proc. of the ASME 2004 Design Engineering Technical
Conferences and Computers and Information in Engineering Conference.
Krause, F. and U. Kaufmann, U. 2007. „Meta-Modelling for Interoperability in Product
Design”. CIRP Annals - Manufacturing Technology, 56(1), 159-162.
Nour, M. 2009. “Performance of different (BIM/IFC) exchange formats within private
collaborative workspace for collaborative work” ITCon 14 p.736-752
Plume, J. and J. Mitchell. 2007. “Collaborative design using a shared IFC building model--
Learning from experience”. Automation in Construction, 16(1), 28-36.
TNO IFC engine, (2002),”TNO IFC Engine series”, http://www.ifcbrowser.com/. Last
accessed 01.06.2010
Weise M. (2006). “Ein Ansatz zur Abbildung von Änderungen in der model-basierten
Objektplanung. Doctoral Thesis, Institute of construction Informatics, Report 4 (Ed. R. J.
Scherer), Technische Universität Dresden, Germany, ISBN 3-86005-557-7
