Building Information Modelling (BIM) represents a global industry effort to harmonize the way the professions, designs, constructs and builds the environment. The BIM models, which are currently created are however not sustainable and will be unreadable or outdated in a very short time. The EU research project DURAARK presents solutions on how to maintain BIM models and the huge effort put into them by making them usable over long periods and in future projects.
BIM in the building lifecycle
BIM enables the description and communication of our build environment in a comprehensive way, that reflects different hierarchies, scales and resolutions, as well as relations within the elements contained in the model, as those in other domains. Originating in the 1970s the concept of BIM is now more than four decades old. It has taken the industry a very long time to adapt to the idea of an information driven approach and it is still far from being widely implemented. However in countries that have embraced BIM approaches early a proliferation of BIM from the early focus on the construction sector to all other areas can be observed. In Denmark a 2007 governmental demand for BIM resulted for instance in an ongoing transformation of the building industry, where BIM approaches are commonly used across all disciplines involved in mid- to large sized projects. The demand for this originates from the side of the building owners though, who want to reduce risks and costs by easing communication and problem detection during the design phase. A further and increasingly important motivation for them is to receive a Information model of the building, which can serve as base for the maintenance and persistent planning of their facilities in Facility Management (FM).
Through this move, the profession is starting to engage with a lifecycles of building data – often drawn in the infamous circular BIM diagrams (Fig. 1). A persistency of a building model is established, where the model of the building is following a buildings lifecycle, from its design, to construction, maintenance and eventual retrofitting and further use or demolition. This persistency creates a digital mirror of the building, which allows for efficient monitoring and planning. The emerging practice with BIM highlights however challenges, which are more substantial, than the problems of interfaces between programs, formats and standards:
- The data for building construction is different to the one for building maintenance. This results in a purge of data in the construction model, when it enters FM. Building owners might believe, that they are maintaining the BIM model of their building, but essential data on the construction process and building elements is lost during the translation of data to FM.
- The second challenge is to keep a building documentation up to date. Buildings are a life and keeping track of changes in a BIM model is an expansive endeavor. Especially in a setting, where the value generated through this endeavor is not visible on short term, as in real estate management, or in highly dynamic environments as in industry.
The idea of a unified “all-knowing” BIM model is solely theoretical and practice shows, that building related data is often not or just partially existing or already outdated. And even in cases, where up to date 3d Data exists, will practiconers require to verify and eventually update models, before they can be reused in coming building processes. Already now these processes are extremely laborious, imprecise and hence often so expensive, that decision makers opt to base their planning on incomplete data or to measure and “redraw” the building –purging valuable information from the existing models.
The need to establish data rich models is however obvious and digital processes in building industry, architecture and urban planning will increasingly create datasets that are dynamic and semantically related to a range of resources on the web. These resources range from building-related vocabularies and taxonomies, legal and environmental policies or statistical and historical data about a building’s context and infrastructure. This complexity and diversity of the building industry results in models that cannot contain all, but only specific information about a building. By what means can information-rich architectural data be preserved for future use and how can relations between incomplete architectural models be determined?
Project DURAARK: The research
A collaboration of Computer Scientist, Architects and knowledge managers tackles this problem in the European research project DURAARK (Durable Architectural Knowledge). Funded through the European Commission’s FP7 Program it investigates means to generate and maintain semantically rich BIM models from incomplete datasets. Using concepts from linked data and computation allows to create meaningful architectural information from an overlay and linkage of existing architectural data, such as 3d data and 3d Point Clouds. Being semantically rich, allows the building data and the elements it contains to be found and identified in bigger datasets, as they can be linked and compared to other data. The aim of the project is to generate the semantics in a highly automated way, which reduces costs and increases the ease of use for non-specialists.
DURAARK takes point of departure on both the level of metadata and existing properties of elements, as on the geometries within the model. The approaches create new metadata, as geometries. Two general directions exist:
Semantic enrichment of building models
Documentation of buildings, such as 2D/3D models or Building Information Models (BIM), are lacking semantic information about the context of a structure or the model, which would facilitate non-ambiguous interpretation. Lacking information may include detailed material properties, vendor information, provenance data or information about legal, infrastructural and environmental context.
The Web of data is increasingly offering datasets and vocabularies, which provide such additional information. DURAARK is developing automated and semi-automated techniques to enrich building models with corresponding data, in particular Linked Data. Given the continuous evolution of Web data and vocabularies, targeted methods for crawling and archiving Web data in a scalable manner are considered an inherent part of the semantic enrichment approaches developed by DURAARK.
Geometric Enrichment of building models
Especially Point Clouds from Laser Scans or other 3d registration technologies are today the mean of choice for the generation of precise and complete documentations of the current state of building structures. They are however data heavy and cannot be integrated into the existing workflows of the building industries with structured BIM models following international standards such as IFC. DURAARK develops hence approaches to almost automatically relate existing PointClouds and BIM models, find differences between them and to transfer especially Point-Cloud data into BIM formats:
- Data registration and synchronisation: Two different representations of the same building model can be synchronized and overlaid using an automated approach.
- Point Cloud to BIM tools: The DURAARK project developed techniques for the fully automatic generation of BIM models from indoor point cloud data (left image). A high-level semantic segmentation of the formerly unstructured point clouds allows to separate them into stories and rooms and to detect features, such as walls, window, doors and interior objects. The output of this tool are not simple geometrical entities, but fully relational BIM models in IFC format. While this approach creates data, which can be directly used in Facility Management and early stage architectural design processes, the methods produces as well further data, as attributed spatial graphs, that integrate into planning processes of asset managers.
- Semantically aware Difference analysis: Simple tasks, as the display of differences between models, whether BIM to Point-Cloud or Point-Cloud to Point-Cloud are today laborious tasks, and their results are often still open to interpretation. Activating the knowledge about objects in building models increases the quality of difference detection, as it allows to exclude furniture and other scan artefacts (purple in the illustration above) from the analysis and to present users with precise areas were models are identically (grey) and where deviations occur (red). Automatically generated statistics about the amount of deviations help the users further to quickly access the quality of models and quickly grasp the spaces, where deviations actually exist.
Development of Tools and Systems
DURAARK developed a first software prototype for the digital long-term preservation of building information models. The prototype assists producers of building data in the preparation of their 3D models for the submission to a digital preservation system. Following the principles of long-term preservation and future reuse, the software and process is built on established standards and sustainable formats. Covering the pre-ingest phase of long-term archiving, the DURAARK prototype contributes to best practices and standard procedures for the long-term preservation and future use and reuse of such models. DURAARK approaches have been evaluated with stakeholders, such as the Danish land surveyor company LE34 or Denmarks fourth biggest real estate owner Copenhagen Properties.