Beruflich Dokumente
Kultur Dokumente
Abstract: Integrated project systems hold the promise for improving the quality while reducing the time and cost of architecture/
engineering/construction 共AEC兲 projects. A fundamental requirement of such systems is to support the modeling and management of the
design and construction information and to allow the exchange of such information among different project disciplines in an effective and
efficient manner. This paper presents a methodology to implement integrated project systems through the use of a model-based approach
that involves developing integrated “smart AEC objects.” Smart AEC objects are an evolutionary step that builds upon past research and
experience in AEC product modeling, geometric modeling, intelligent CAD systems, and knowledge-based design methods. Smart objects
are 3D parametric entities that combine the capability to represent various aspects of project information required to support multidisci-
plinary views of the objects, and the capability to encapsulate “intelligence” by representing behavioral aspects, design constraints, and
life-cycle data management features into the objects. An example implementation of smart objects to support integrated design of
falsework systems is presented. The paper also discusses the requirements for extending existing standard data models, specifically the
Industry Foundation Classes 共IFC兲, to support the modeling of smart AEC objects.
DOI: 10.1061/共ASCE兲0887-3801共2005兲19:2共172兲
CE Database subject headings: Construction management; Methodology; Integrated systems; Information management; Falsework.
from the object geometric representation. objects-related knowledge. Computable knowledge can be repre-
Second, a smart object can manage its evolutionary state sented declaratively in the form of production rules 共e.g., configu-
throughout the project life cycle. Moving from conceptual to pre- ration constraints or design rules兲 or procedurally in the form of
liminary to final design stages, smart objects can keep track of computational methods 共e.g., for calculating structural responses
their evolution history. Two possible evolution changes can be or materials quantity兲. Noncomputable knowledge, such as un-
tracked: 共1兲 changes in object definition 共or schema兲; and 共2兲 structured documents, can also be linked to specific object param-
changes in object configuration 共i.e., parameter values兲. Changes eters. Making object-related knowledge accessible through the
in objects’ schemas are tracked by maintaining a “version identi- object model could be used to automate the computation of some
fier” for each schema and defining methods to map between dif- objects parameters and to provide an efficient method to index
ferent schemas. Changes in object configuration are tracked by and access project documents. This will also allow the use of the
maintaining a list of the object “parameter set” along with a object model to capture and represent the design knowledge and
change version identifier and change information. Each object has expertise in the form of object-centered design knowledge bases
a current “parameter set,” from which previous sets 共or object 共Yoshioka et al. 1998a,b兲. As the design process becomes more
states兲 can be navigated. complex and increasingly knowledge-intensive, the need to sup-
Third, smart objects can be arbitrarily complex by aggregating port the capturing, representation, and use of various forms of
or referencing other objects and defining the rules that describe design knowledge becomes even more critical.
the interrelationships and interaction between these objects. The
advantage of defining objects as compositional assemblies of
more primitive objects is two-fold: 共1兲 higher level objects are Implementation of an Integrated Falsework Design
more intuitive and easier to work with from a design viewpoint; System Using Smart Objects
and 共2兲 configuration rules and design constraints can be enforced
to control complex objects configuration at higher level objects A prototype software environment was implemented using the
and, therefore, ensure the consistency and validity of the design at smart objects concept to support the integrated design, layout,
all design stages. structural analysis, cost estimating, and erection planning of false-
Fourth, unlike traditional CAD objects, smart objects’ param- work systems used for constructing cast-in-place concrete box
eters not only describe the geometry but also describe other non- girder bridges. Falsework objects can be thought of as specialized
geometric information such as material, specification, cost, con- assemblies of the IFC columns and beams elements that are de-
struction methods, etc. Smart objects integrate project information signed and erected to satisfy structural, schedule, and site con-
by representing and considering the interaction of different straints, among others. The software was implemented using the
project disciplines at the object level; that is, the object model ObjectARX C⫹⫹ class library that extends the AutoCAD envi-
will glue together different project views pertaining to the object ronment 共AutoDesk 1999兲. The system was developed in collabo-
and enable interoperability of function-specific software tools ad- ration with Hua Construction, Inc., a Taiwan-based falsework
dressing these views. Using the same object model, different subcontractor.
project disciplines can access data relevant to their domains. Also, Falsework systems are complex structures in terms of the
integrating different project views would enable developing meth- number of geometric and functional parameters that a designer
ods to automate mapping between these views 共e.g., automatically must consider in order to develop a structurally safe and economi-
performing quantity takeoff兲. cal system. Falsework systems are typically designed and erected
Fifth, three-dimensional geometric representation of smart ob- by a specialty contractor who needs to share and exchange project
jects can be constructed given the set of object geometric param- information with both design and construction teams. The input to
eters. Addressing configuration design problems requires accurate this process typically consists of the bridge design documents, the
geometric representation of objects and their spatial relationships site topography 共including soil data兲, and the construction sched-
共e.g., for objects layout, interference detection, etc.兲. The use of ule. The falsework subcontractor team will then define the design
3D models would provide a wide range of benefits to various parameters along with the specification, cost estimating, and erec-
project processes. Project team members can more easily review tion planning to meet the bridge design and construction schedule
and evaluate the design details and the construction process from constraints. This process typically requires several iterations and
multiple perspectives to identify potential problems. This could is subject to stringent code requirements to ensure the stability
significantly improve the communication and collaboration and safety of the system.
among project teams. 3D models would also enable effective un- To demonstrate the importance of sharing an integrated data
derstanding of project information through visual interaction with model in a typical falsework project, let us consider the following
and manipulation of design entities. The construction team can scenario. Suppose that design changes have modified the layout
or configuration of some falsework components. The objects rep- tioned at regular or irregular grid points. In laying out a segment,
resenting these components need to rebuild the structural model the user specifies the segment parameters, which include length,
and perform structural analysis on the new model, recalculate the width, start and end offsets, number and spacing of towers in each
quantities to be used for cost estimating, or perform checks to direction, the dimensions of each tower, and the ground elevation
ensure that the new changes are within the code limits. By encap- and top elevation at the start and end of the segment. The user can
sulating the data relevant to each of these domains in the objects, define any number of segments. The user can also modify seg-
it would be possible to automate and support interrelated project ments’ global properties 共e.g., its elevation兲 as well as the loca-
tasks across multiple domains while ensuring the consistency and tion, height, and dimensions of individual towers either directly
integrity of the project data. Fig. 1 demonstrates the role of smart on the 3D models or using the dialogs associated with the in-
objects in integrating falsework design and erection processes. tended operation. The user can also place layers of beams longi-
Falsework smart objects were developed to support efficient tudinally or transversely in a segment. The system has been de-
design as well as the sharing and exchange of project information veloped to primarily support the following use cases: 共1兲 create,
between various project disciplines. The objects are implemented configure, and modify falsework segments; 共2兲 perform structural
as derived C⫹⫹ classes that inherit from the ARX AcDbEntity modeling and analysis; 共3兲 produce layout drawings; 共4兲 produce
class. These objects included: falsework segments, towers, beams, bill of materials and cost estimates; 共5兲 perform planning and vi-
and grids 共Fig. 2兲. The objects also encapsulated rules to ensure sual simulation of the erection schedule in relation to the bridge
the consistency and correctness of the design. For example, beam construction schedule; and 共6兲 edit the object library.
objects explicitly define rules that ensure a beam is supported by A number of different configurations for each object are stored
at least two towers and segment objects define rules that ensure in a library, from which a falsework designer would select the
the total height of the segment’s towers will be less than the components and configurations that are most suitable to the
height of the bottom of the bridge deck or girder. If the user project at hand. Designers then specify values for the predefined
specifies objects parameters that may violate these rules, an ap- objects parameters. Based on these values, the objects determine
propriate warning message is displayed and the user will be re- their 3D geometric configuration and can perform several func-
quested to revise these values before proceeding. As such, the tions, such as generating a bill of materials for cost estimating
design of the falsework system will be kept in a consistent state 共Fig. 4兲 or a finite-element model to check the stability against
throughout the design process. By adding more such rules, the different loading conditions. Designers position the objects in
objects will be smarter and more capable of supporting end users their exact location by referencing points on the bridge structure
to reach an optimum and viable design in the shortest time pos- or on the site. Fig. 5 shows an example of a complete falsework
sible. system. The objects’ behavior was modeled and implemented in
The software defines a falsework system model as a series of the form of methods to control their response to user modifica-
straight and/or curved segments. Users can enter and edit the tions in order to ensure that the objects remain in a consistent and
values of the falsework system parameters through a set of dialog valid state and that the relationships to other objects are main-
boxes that are designed to offer users the flexibility to easily tained. Designers can modify different objects 共e.g., tower
configure an entire falsework system by setting the values of a heights兲, and the impact of any changes is automatically propa-
small set of parameters which can be used to derive values for the gated by the object methods to other dependent objects 共e.g.,
remaining parameters required to fully define the falsework sys- beam layers兲.
tem design. Fig. 3 shows a tabbed dialog box that is used to enter Objects also implemented methods to store, access, and man-
and edit the parameters of a typical falsework segment. age the multidisciplinary project information. By encapsulating
Each falsework segment contains an array of towers posi- data relevant to different project disciplines, it would be possible
to automate and support interrelated project tasks across multiple could incorporate object behavior within the model could capture
disciplines while ensuring the consistency of the project data. For this notion of common object behavior and may improve the ef-
example, if the objects’ configuration changes, the objects can ficiency of software development, because the behavior would not
modify the structural model and reperform structural analysis, need to be recreated independently within each application.
recalculate the quantities to be used for cost estimating, or per- While the current IFC model does not address behavioral or
form checks to ensure that the new changes are within the code functional aspects of the objects, traditional knowledge-based
limits. techniques do not address the modeling of the multidisciplinary
Components of the falsework systems are represented, orga- objects’ data in a complete and comprehensive manner. Repre-
nized, and managed as parts of an integrated object model. Users senting a hybrid of the two approaches, smart objects are posi-
could navigate through the object model using the “project ex- tioned to play an integrative role that would build on the strength
plorer” interface 共Fig. 6兲. This interface provides a hierarchical of both approaches to allow software tools to share and exchange
view of various pieces of project information such as the false- semantically rich AEC object models. By supporting the capabili-
work product model, schedule, cost estimate, and documents. ties of smart objects in the IFC model, systems will no longer
Users could associate pieces of information between different merely represent and exchange objects’ static data; more knowl-
views to indicate “relationships” between different project as- edge and semantics about design, including behavior, objects’ re-
pects. For example, users could drag schedule activities and drop lationships, design rules, and constraints will also be represented
them onto falsework components to indicate that a particular and exchanged. Extending the IFC model to support smart objects
component is dependent on one or more activities. As a result, a would enable the addition of more semantics 共i.e., behavior and
4D simulation of the erection process could be automatically gen- knowledge兲 to IFC objects and allow the definition of application-
erated. specific custom complex objects that are composed of more
primitive objects.
The IFC model defines a flexible, yet powerful, mechanism
Extending Industry Foundation Classes Schema that allows extensions to the model through the use of the If-
to Support Modeling of Smart Objects cPropertySet entity. An IFC property set could be used to define a
set of properties 共IfcProperty entities or other nested IfcProperty-
Currently, the IFCs are used to exchange the form and structure of Set entities兲 and can be linked to any number of IFC objects using
a project model between applications. Each application is respon- the IfcRelAssignsProperties entity. Representing and linking ob-
sible for implementing the appropriate object behavior; there is no jects’ behavior could be supported following a similar approach.
notion of exchanging or standardizing this behavior. Yet, users A new IfcBehaviorDef entity could serve as a container for
might reasonably expect these objects’ behavior to be similar be- object-related behavioral if-then rules as well as procedures, both
tween different applications. Extending the IFCs so that they of which are supported by the EXPRESS language. IfcBehavior-
Fig. 3. Sample dialog for entering and editing falsework segment parameters
Def entities can be linked to any number of IFC objects using while a constraint stating that the beams’ elevation equals the
another entity called IfcRelAssignsBehavior. towers’ top elevation could be used to propagate the second
Objects’ behavioral constraints can be modeled more naturally change.
using if-then production rules. In a typical IfcBehaviorDef entity, Rules could also be used to enforce spatial constraints among
rules may reference the attributes of the same object or other different objects 共e.g., adjacency or connectedness constraints兲.
related objects. The rules could be used to enforce design con- Changing the location of an object may require that other related
straints to maintain the consistency and correctness of the design. or connected objects be relocated according to certain rules. For
For example, in the falsework model described in this paper, a example, moving the falsework towers would also require moving
constraint was implemented to specify a relationship between a the supported beams in order to maintain the correctness of the
tower type and its minimum dimensions, and the change of a system. Supporting the propagation of this type of change could
tower type would require changing certain dimensions. Also, be supported by adding an entity that represents “anchors” be-
changing the towers’ height would require changing the sup- tween different objects as well as a mechanism or “reactors” to
ported beams’ elevation. Rules could be formulated to enable au- detect that changes have occurred in some objects. Anchors can
tomatic propagation of these changes. For example, a rule stating be thought of as another form of relationships between objects
that “if tower type is X then dimension Y = Z” could be used to that indicates “spatial-dependencies” between objects.
represent the dependency between a tower type and dimensions, The definition of objects’ behavior would typically become
part of the IFC model-development process. The domain and using standard component interfaces 共e.g., a standard COM inter-
modeling experts that initially define the IFC objects would define face兲, or using web services. Applications could then interact with
the constraints and procedures, which would then be available for standardized component libraries or web services to execute these
use by software developers and end users. procedures. For example, a “CalculateFalseworkBillOfMaterials”
There are several possibilities for how these procedures and procedure may be implemented in a standard interface 共e.g.,
constraints in object behavior definitions might be executed by IFalseworkBillOfMaterials兲, or as a web service. Other proce-
applications. For example, each application could implement or dures could be implemented to check some object values or to
interface with a rule-processing engine, possibly as part of a stan- retrieve some data 共e.g., form an online product repository兲.
dard IFC toolkit. Alternatively, procedures could be defined in Another possible extension is the capability to represent arbi-
behavior sets by their interface 共i.e., signature兲 and implemented trarily complex object assemblies where an object can be com-
Fig. 6. Project explorer, product model, and process model of an example falsework system
posed of more primitive objects. For example, a tower object is that defines the object attributes will be added. In addition, the
composed of a set of column and beam primitives. The IFC mod- property set will define an attribute for the version number, an
els a set of predefined complex or container objects 共e.g., build- attribute to reference the previous version property set, if one
ing, building story兲. A new IfcProduct-derived class 共e.g., Ifc- exists, along with attributes that record change information 共e.g.,
ComplexObject兲 could be used to support this extension. This cause for change, time and date, etc.兲. Using this capability, the
class could define a list of references to its component entities. IFC model will evolve with the project while representing a com-
This would enable modeling, managing, and accessing arbitrarily plete recording for design changes throughout the project. Such a
complex and hierarchical objects as one unit. To define the inter- model could be later transferred and used to support the facility
relationships among the set of component objects, an IfcBehav- management activities.
iorDef entity could be used to specify the rules of interaction
between these objects. Similarly, the IfcBehaviorDef object could
then be linked to the IfcComplexObject instance using an
IfcRelAssignsBehavior object.
Objects are initially defined using a simple set of parameters Summary and Conclusion
and evolve into a more comprehensive and detailed representation
as more design iterations are performed. The IFC model repre- This paper presented a model-based approach that employs
sents a static snapshot of objects data with no ability to record or “smart objects” to implement integrated project systems. Smart
track changes that an object has gone through during various objects are an evolutionary step that builds on almost two decades
design stages. The IFC model defines an entity, IfcOwnerHistory, of research and experience. The paper discussed the main charac-
which supports tracking the agent who performed the change. teristics of smart AEC objects and presented the requirements to
However, it does not define any mechanism to track the changes extend the IFC data model to support the modeling of smart ob-
themselves. An important feature of smart objects is their ability jects. Besides serving as data models that integrate multiperspec-
to record different changes that occur to the object state during tive project views and encapsulate behavioral object intelligence,
different design stages. To add the capability to track and manage smart objects also enable the exchange of semantically-rich data
evolving objects’ data, a mechanism for representing and manag- models between different software tools.
ing the evolution and change of the objects’ data sets is required. Smart objects could potentially offer many benefits to the de-
One possible approach would involve the use of property sets to sign process, including 共1兲 modeling behavioral aspects within
track the current version and the history of changes in each ob- individual objects as well as between interdependent objects, and
ject. For each IFC object, a corresponding version property set integrating those aspects with the structural and configuration as-