Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Model Interoperability in Building
Information Modelling
Jim Steel, Robin Drogemuller
Queensland University of Technology/
CRC for Construction Innovation
Alternate First Page Only
Summary
• Problem Domain: Architecture/Engineering/Construction industries
• Building Information Modelling
• Industry Foundation Classes as a Domain-Specific Modelling Language
• Using IFC as an interoperability mechanism
• File/syntax
• Visualisation
• Semantic
• Conclusions
The Domain: Stakeholders
• A lot of different parties can be involved in a building project
• Client/owner
• Financier
• Architect/s
• Engineers (structural, mechanical, electrical, acoustic, thermal perf.)
• Government/s (code checking, building approval)
• Builders & Sub-contractors
• Cost Engineers
• All of these need to interact in some way with the building model/plan
The Domain: Concepts
• A lot of different things to model, including
• Structure
• Architecture
• Heating/Ventilation/Air-conditioning (HVAC)
• Electrical
• Interiors
• Landscaping
• Processes, scheduling, costing, ownership/responsibility
• Most of these have complex inter-dependencies
The Domain: What happens
• Traditionally, information exchanged as paper
• Building plans (drawings), cost plan, bill of quantities, building
specification
• Increasing use of 3D tools in the design process
• Often still rendered as 2D for exchange with other parties involved in
the process
• 2D is well-understood, esp. for legal liability implications
The Domain: Building Information Modelling
• Building Information Modelling
• Exchange of rich models
• Full 3D, but also with metadata about materials, processes, purposes,
etc
• Allows for more detailed examination and processing of models by
people and increased use of automation techniques for model analysis
and preparation
• “Tipping point”
• In 2008, 45% of architects, engineers & contractors use BIM on a
significant proportion of their projects
Building Information Modelling
• Noun
– An information model representing a buildingproject across all disciplines and all stages ofthe project lifecycle (initiation, design, bid,construct, maintain, refurbish/recycle)
• Verb
– Activity of sharing the complex informationacross the “owners” of all systems involved inproject procurement
• Better term
– Virtual Design & Construction (VDC)
Why BIM?
• Build virtually before building actually
– Explore more alternatives at each stage
– Directly link analysis tools to BIM
(interoperability)
• Better intergration
– Reduce information entry
– Reduce errors
Why is BIM a problem?
• One Island East
WATER
PlumbingPlumbing
PlumbingFIRE
SPRINKLERS
ELECTRICAL
HVAC
STRUCTURE
COLUMNS
CORE
FLOORS
CLADDING
Complexity?
x 71 floors
Industry Foundation Classes
• Industry standard for BIM exchange
• Since 1996, now version 2x3 (2x4 in draft)
• Defined as a STEP/Express schema
• Mapping to file format (Part 21)
• Mappings to programmatic APIs
• Mapping to XML (ifcXML)
• Very similar architecture to OMG-style MDA
• Wide uptake
• Does vary from between disciplines
The IFC Metamodel (schema)
• BIG!
• 653 entity types, 327 data types, 317 property sets
• BROAD!
• Modelling constructs for:
• Basic building elements, e.g. walls, slabs, beams, columns, doors
• 3-dimensional geometries
• Electrical/Ventilation elements, e.g. wires, ducts, fans, vents
• Facilities management (tasks, etc)
• Identity (people, companies, etc)
• much, much more…
The IFC Metamodel (schema)
The IFC Metamodel (schema)
• Alternative modellings
vs
• Extension mechanism
• IfcProxyObject
• For things that cannot be modelled using another IFC type
IFC Models: example
• Mechanical services (and some
structure as a guide) for a 19-
storey office tower
• Plant rooms, verticals, 2 example
floors (not a complete model)
• 360MB file
• 7.3 million objects
Interoperability: Drivers
Why is interoperability an important issue for the domain?
• Large number of participants on any given project
• Complex, highly interdependent models
• Tool-independent Model versioning & management facilities
• Many analyses are manual and labour intensive, that could be at least
partially automated
• Constructibility (process simulation)
• Quantity takeoff and cost estimation
• Environmental analysis, including emissions, carbon, air quality
• Thermal/Acoustic/Lighting performance
Interoperability: Drivers
Interoperability: Drivers
• A lot of different things to model, including
• Structure
• Architecture
• Heating/Ventilation/Air-conditioning (HVAC)
• Electrical
• Interiors
• Landscaping
• Processes, scheduling, costing, ownership/responsibility
• Most of these have complex inter-dependencies
IFC Interoperability: File/Syntax
• Interoperability almost universally by Part-21 file interchange
• Occasional problems due to file size
• 7.3 million objects is a large number to parse and store
• 7.3 million objects is a large number to render in 3D
IFC Interoperability: Visualisation
• The main goal of IFC
• Generally good
• Occasional misunderstandings to do with relative locations of copied/reused
objects
• Columns or openings “floating in space”
• Increasingly rare
IFC Interoperability: Alternative visualisations
• Different stakeholders need different visualisations
• An architect wants photo-realism
• A quantity surveyor prefers elements to be visually distinct based on
element type or material in order to classify and quantify them
• A structural engineer sees elements differently again, tied to structural
properties and not visual properties
• IFC supports multiple visualisations
• Managing, grouping and ordering visualisations is not especially well-
supported in and between tools
IFC Interoperability: Semantics
• The most challenging type of interoperability
• Increasingly in demand
• Major stumbling block for uptake of automated analysis tools
• Garbage in, garbage out
• “My crappy model doesn’t produce beautiful analysis results, the
software must be terrible!”
• Some small issues
• Some (significant) tools have a lax approach to preserving globally
unique object identifiers
• Some bigger issues
IFC Interoperability
• Modelling style
• Heterogeneity of attribute values
• If a beam is made of steel, has a compression strength of 450, and a
cross-section of UB470.4, which information goes in the “material”
property, and which goes in the description as text?
• Too many possibilities -> nightmare for anyone writing a tool that
classifies and quantifies the tonnage of steel beams in a design
• Conventions even for naming are different in different jurisdictions
• Solution (technical): ontology-type approaches for standardised modelling
objects
• Solution (non-technical): agreement by stakeholders on standard modelling
styles
IFC Interoperability: Semantics
• Coverage is an issue
• IFC is too big
• Tools cannot visualize everything: rare – geometry is common
• Tools do not (and should not have to) provide palette objects for every
IFC type
• kerb, or a low wall or small slab
• IFC is too small
• There is no construct for modelling a water tank
• Build it out of slabs and curved walls, OR use a proxy object
• Solution: Conventions & guidelines for good modelling practice
IFC Interoperability: Semantics
• Alternative representations
• Beyond just alternative visualisations, some stakeholders think about the
model using very different paradigms
• In road modelling, models often begin as string-based models with
vectors for edges instead of surfaces. Need to infer surfaces later in the
process to change paradigms
• When converting a model for thermal/acoustic analysis, walls, windows,
doors, etc from the original are split or merged to simplify the analysis
process. This mapping needs to be preserved
• Solution: Investigating the use of bidirectional/roundtripped domain-specific
model transformations to maintain correspondences
Conclusions: IFC as a modelling language
• Language is very big, very broad
• Models are very big: scalability challenges
• Idea: interesting case study for both metamodel modularity and model
modularity
• 3D visualisation makes visualisation-level interoperability more challenging
than diagrammatic languages, but it seems to work
• Progression from drawing language to true modelling language, used to
drive analysis and automation.
• Reminds me of another language
• Unlike UML, semantic interop issues are not about semantic interpretation,
but clear expression.
Conclusions: The IFC Process
• Compared to languages I know: UML/MOF
• Standardisation by committee/compromise
• Something for everyone: “unionised” modelling language
• Initial set of implementing tools:
• standardisation before implementation?
• shared user understanding, vs shared machine understanding
• threat of the elephant in the room (Autodesk, Rational)
• Transition to partial tool coverage: “family of languages”
• jury still out!
• Compliance testing issues
Conclusions: Human Processes
• Importance of human processes
• In many cases, technical solutions exist provided that they are
combined with agreement and effort amongst stakeholders
• Agreement on modelling “style” and standardised building blocks
• Compliance testing and interoperability testing
Conclusion
• Questions?
The Domain: Concepts
• A lot of different things to model, including
• Structure
• Architecture
• Heating/Ventilation/Air-conditioning (HVAC)
• Electrical
• Interiors
• Landscaping
• Processes, scheduling, costing, ownership/responsibility
• Most of these have complex inter-dependencies
The Domain: Concepts
• A lot of different things to model, including
• Structure
• Architecture
• Heating/Ventilation/Air-conditioning (HVAC)
• Electrical
• Interiors
• Landscaping
• Processes, scheduling, costing, ownership/responsibility
• Most of these have complex inter-dependencies
The Domain: Concepts
• A lot of different things to model, including
• Structure
• Architecture
• Heating/Ventilation/Air-conditioning (HVAC)
• Electrical
• Interiors
• Landscaping
• Processes, scheduling, costing, ownership/responsibility
• Most of these have complex inter-dependencies
Content Page