Upload
xuan-f-zha
View
212
Download
0
Embed Size (px)
Citation preview
Knowledge-intensive collaborative design modeling and support
Part I: Review, distributed models and framework
Xuan F. Zha a,*, H. Du b
aNanyang Technological University and Institute of Manufacturing Technology, SingaporebNanyang Technological University, Singapore
Received 17 December 2002; accepted 27 April 2005
Abstract
Contemporary design problems embody significant levels of complexity, which make it unlikely that a single designer can work alone on a
complex design problem. Knowledge-intensive collaboration has emerged as a promising discipline for dealing with the modeling and decision-
making processes in distributed design systems. In this paper, we first review previous work and current research status of knowledge-intensive
collaborative design. Then, we develop knowledge-intensive distributed design models and a framework for collaborative design modeling and
decision support. Based on the component-based modeling/design approach, a unified design-with-modules scheme is proposed to model the
distributed network-centric design processes, and a web-based knowledge-intensive distributed module modeling and evaluation framework (KS-
DMME) is developed. The KS-DMME framework intends to link distributed, heterogeneous knowledge-based design models and tools and assist
designers in evaluating design alternatives, visualizing trade-offs, finding optimal solutions, and making decisions on the web. It also enables
designers to build integrated design models using both the local and distributed resources and cooperate by exchanging services. Built on it, the
client (browser)/knowledge server architecture allows product design models to be published and connected over the WWW to form an integrated
intelligent models/modules network. Finally, as an illustration, a simple design problem model for modular micro-robotic systems design is built
upon the KS-DMME framework.
Published by Elsevier B.V.
Keywords: World-wide web; Network-centric design; Design-with-modules; Knowledge servers; Distributed module modeling and evaluation (DMME);
Integrated design and analysis; Collaborative design; Design support system
www.elsevier.com/locate/compind
Computers in Industry 57 (2006) 39–55
1. Introduction
The need for an infrastructure based upon distributed
computing, information and communication technology (i.e., a
cyber-infrastructure) becomes increasing paramount as we
progress closer to a knowledge economy. A recent report of US
National Science Foundation (NSF) indicates that such a cyber-
infrastructure (CI) will play a pivotal role in supporting and
shaping future predicative product realization systems and
processes.
Design process is a knowledge-intensive and collaborative
task. The knowledge-intensive support in a CI becomes critical
DOI of original article: 10.1016/j.compind.2005.04.006.
* Corresponding author. Present address: National Institute of Standards and
Technology, Gaithersburg, MD 20899-8263, USA.
E-mail address: [email protected] (X.F. Zha).
0166-3615/$ – see front matter. Published by Elsevier B.V.
doi:10.1016/j.compind.2005.04.007
in the design process and has been recognized as a key solution
toward future competitive advantages in product development.
Integrated design requires skills of many designers/users and
experts; each participant creates models and tools to provide
information or simulation services to other participants given
appropriate input information. It is the goal that the network of
participants exchanging services forms a concurrent model for
integrated design. Contemporary design problems embody
significant levels of complexity, which make it unlikely that a
single designer can work alone on a design problem. The
continuing growth of knowledge and supporting information
and the ever-increasing complexity of design problems have led
to the increasing specialization. Current computer tools for
product design are generally stand-alone applications. How-
ever, design activities may involve many participants from
different disciplines and require a team of designers and
engineers with different aspects of knowledge and experience
towork together. Thus, there is a need to support and coordinate
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5540
highly distributed and decentralized modeling activities in
CAD systems. Wide-area networks and the internet-based
WWW allow developers to provide remote web-based design
servers, and CAD systems running on these servers can support
a large-scale group of users who communicate over the
network. As such, user interfaces based on the web protocols
provide access to the remote web-based design servers, and
users do not need special hardware or software to consult these
services with appropriate web browsers. The advantages of
using WWWas a cyber-infrastructure to build a design system
are three-folds. First, web browsers can provide a nice human-
interface for a design system, because they can display various
media including hypertext, moving images, sounds, and three-
dimensional (3D) graphics. They can also handle interactive
operations of the media, such as manipulation with a mouse of a
3D object described in VRML. Second, hypertext transfer
protocol (HTTP) can be a standard communication protocol for
design systems, since terminals connected to the Internet can be
accessed from any web site via the protocol. Third, it becomes
possible to use various hardware and software resources
distributed over the Internet together to accomplish a single
mission. Therefore, it is natural to consider developing design
support systems on the web.
This research aims to develop a methodology and a web-
based knowledge-intensive framework for collaborative design
modeling and support. The presentation of the proposed
methodology and framework is divided into two parts:
distributed modeling techniques and framework (Part I) and
system implementation and application (Part II). Part I is the
subject matter of this paper and Part II is covered in the
following paper of this issue. This paper focuses on the review
of previous work and current research status and the
development of distributed models and a web-based knowl-
edge-intensive framework for collaborative design modeling
and decision support. It intends to build up an information
cyber-infrastructure to facilitate the rapid construction of
integrated intelligent design models. A unified design-with-
module scheme is proposed for representing and modeling the
distributed network-centric design process.
The organization of this paper is as follows: Section 2
reviews network-based design tools and the distributed module
modeling and evaluation for the design process. Sections 3 and
4 give a brief of the proposed approach and discuss DMME for
the design process. Section 5 proposes an integrated knowledge
representation scheme. Section 6 provides a description of the
KS-DMME framework. The knowledge server architecture for
supporting different types of collaborative design activities in a
distributed design environment is described. As an illustration,
a simple design problem model is built upon the KS-DMME
framework. Section 7 summarizes the paper.
2. Literature review
There are many research efforts working on enabling
technologies or infrastructure to assist designers in the
computer-aided network-centric design environment [76–
79,81,83–87,90]. Some of them intend to help designers
collaborate or coordinate by sharing product information and
manufacturing services through formal or informal interactions
[82,88,4]. Others propose frameworks that manage conflicts
between design constraints and assist designers in making
decisions [80,89,51]. In this section, we will present a research
review in more detail on the work related to knowledge-
intensive collaborative design.
2.1. Design knowledge modeling and design support
There is a large body of knowledge that designers call upon
and use during the design process to match the ever-increasing
complexity of design problems. Generally, design knowledge
can be classified into two categories [1]: product knowledge
and design process knowledge. Product knowledge has been
fairly studied and a number of modeling techniques have been
developed. Most of them are tailored to specific products or
specific aspects of the design activities. For example, geometric
modeling is used mainly for supporting detailed design, while
knowledge modeling is working for supporting conceptual
designs. Based on these techniques, a design repository project
at NIST attempts to model three fundamental facets of an
artifact representation [2,3]: the physical layout of the artifact
(form), an indication of the overall effect that the artifact
creates (function), and a causal account of the operation of the
artifact (behavior). The recent NIST research effort towards the
development of the basic foundations of the next generation of
CAD systems suggested a core representation for design
information called the NIST core product model (CPM) [4] and
a set of derived models defined as extensions of the CPM (e.g.
[5,88]). The NIST core product model has been developed to
unify and integrate product or assembly information. The CPM
provides a base-level product model that is: not tied to any
vendor software; open; non-proprietary; expandable; indepen-
dent of any one product development process; capable of
capturing the engineering context that is most commonly
shared in product development activities. The core model
focuses on artifact representation including function, form,
behavior, material, physical and functional decompositions,
and relationships among these concepts. The entity-relationship
data model influences the model heavily; accordingly, it
consists of two sets of classes, called object and relationship,
equivalent to the UML class and association class, respectively.
Design process knowledge can be described in two levels:
design activities and design rationale. The importance of
representation for design rationale has been recognized but it is
a more complex issue that extends beyond artifact function. The
design structure matrix (DSM) has been used for modeling
design process (activities) and some related research efforts
have been conducted. For example, a web-based prototype
system for modeling the product development process using a
multi-tiered DSM is developed at MIT. However, few research
endeavors have been found on design rationale [6,7].
In terms of representation scenarios, design knowledge can
also be categorized into off-line and on-line knowledge [91].
The former refers to existing knowledge representation,
including design knowledge in handbook and design ‘‘know-
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 41
how’’, etc.; the latter refers to the new design knowledge
created in the course of design activities by designers
themselves. For the off-line knowledge, there are two
representation approaches. One is to highly abstract and
categorize existing knowledge including experiences into a
series of design principles, rationales and constraints. TRIZ is a
good instance of this approach. The other is to represent a
collection of design knowledge into a certain case for
description. Case-based design is an example of this approach
[8]. The current research focus is on the computerization of the
design knowledge representation. For instance, researchers at
the Engineering Design Centre at Lancaster University
established a unique knowledge representation methodology
and knowledge base vocabulary based on the theory of
domains, design principles and computer modeling. They have
developed a radical software tool for engineering knowledge
management. The tool provides an engineering system designer
with the capability to search a knowledge base of past solutions,
and other known technologies to explored viable alternatives
for product design. The on-line knowledge representation is to
capture the dynamic design knowledge in a certain format for
design re-use and archive. A few research efforts have been
found in this area. Blessing [9] proposes the process-based
support system (PROSUS) based on a model of the design
process rather than the product. It uses design matrix to
represent the design process as a structured set of issues and
activities. Together with the common product data model
(CPDM), PROSUS supports the capture of all outputs of the
design activity. The results show that it seems to be a promising
approach. Another focal research area is using ontologies for
product representation (e.g. [10]).
Research suggests, therefore, that there is a need to provide
computer support that will supply clear and complete design
knowledge and also facilitate designer intervention and
customization during the decision-making activities in the
design process [11]. Rodgers et al. [12] describes a design
support system WebCADET using distributed Web-based AI
tools. The system can provide support for designers when
searching for design knowledge. WebCADET uses the ‘‘AI as
text’’ approach, where KBSs can be seen as a medium to
facilitate the communication of design knowledge between
designers.
2.2. Collaborative design frameworks
Computer-supported cooperative design (CSCD) is to use
computer technology to assist people working in the
engineering design field. King [13] considers that the
fundamental issue of CSCD focuses on the computerization
to establish a concept-sharing and seamless coordination
among engineering design participants for concept formation.
While many individuals and organizations may provide
services so that an integrated product model can be constructed,
it is not likely that each participant will disclose the full details
or structure of their proprietary models and data. Providing a
means for encapsulating expert knowledge or know-how is
essential. An object-oriented approach provides a framework
for such knowledge encapsulation. Furthermore, an object-
oriented architecture is also highly suited to a distributed
computing environment [14]. Distributed object technology,
such as Common Request Broker Architecture (CORBA) [15]
and DCOM/ActiveX [16] can be used to address the issue of
distributed computing environment. A computer platform and
language-independent interface definition allows software
applications to communicate with each other provided a
neutral interface has been agreed upon. For example, the
WWW has gained its popularity and momentum through a
platform-independent protocol (i.e., HTTP) and a language-
independent scheme (i.e, HTML) for presenting information.
The software component technology is adopted to build a
collaborative CAD system. Wang and co-workers [17–19]
proposed an approach to collaborative design based on the
software component mechanism. A component is a reusable
application whose data and methods are exposed, and can be
accessed and operated by other applications. The component
agent method is used as the basic system module. However,
their work is only focused on the data, not knowledge.
Distributed design systems might have two distinct forms:
distributed designers with access to centralized resources
(Fig. 1a) or distributed designers with distributed resources
(e.g., engineering models, databases, software applications,
etc.) (Fig. 1b and c). ‘‘Decentralized’’ means that coordinations
between design participants and models are not centrally
modeled or controlled (analogous to the WWW). This is
important because centrally controlling the interactions of all
distributed resources may restrict system growth and flexibility.
If there was a centralized control over theWWW for linking the
hyper documents, it could not evolve so rapidly. Pahng et al.
[20–22] developed a web-based framework for collaborative
design modeling and decision support, based on the distributed
object modeling and evaluation (DOME). The DOME frame-
work asserts that multi-disciplinary problems are decomposed
into modular sub-problems. Modularity divides the overall
complexity and distributes knowledge and responsibility
amongst designers. It also facilitates the reuse of modeling
elements. Thus, DOME allows designers to define mathema-
tical models or modules and integrate or interconnect them to
form large system models. In DOME, a multiple attribute
decision method is used to capture preferences and evaluate
design alternatives from different viewpoints. The resultant
service-exchange network forms an integrated concurrent
system model if module services are connected, as shown in
Fig. 1b. There are other programs (e.g. [23,24], architectures or
frameworks) for network-centric collaborative design. These
include the centralized multi-user system architecture (Fig. 1a),
e.g., the blackboard-based DICE [25,84], DIS [26], data and
model exchange system (Fig. 1c), e.g., SHARE [14], EDN [27],
NIIIP [28] and multi-agent-based distributed system architec-
tures (e.g. [29]). The comparisons between the DOME
architecture and these architectures are described as follows
[21]:
(1) F
or the DOME architecture, as shown in Fig. 1b, the distinctcharacteristic is that when module services are connected,
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5542
Fig. 1. Interactions between modules exchange services.
the resultant service exchange network forms an integrated
concurrent system model.
(2) F
or the centralized multi-user system, multiple users haveaccess to a centralized main system, which stores and
manages information such as product design models, design
information and design history. Although it powerful, a
central system is less suited for loose and flexible
collaborations as it is not an open environment and does
not allow for true knowledge encapsulation. However, such
architecture could be supported within a module of a larger
DOME network.
(3) T
he data and model exchange system architecture tends toprovide an ‘‘over-the-wall’’ sequential interaction between
designers and models. When a designer receives a model or
data from another designer, he/she works on the design and
sends the result of design modification to others. Therefore,
this architecture is not intended to provide concurrent
system modeling functionality.
(4) M
ulti-agent-based architectures are more appropriatefor loosely coupled environments where mutual interac-
tions between objects are not well defined. In the DOME
architecture, the interactions between sub-problems
are explicitly defined through design negotiation so
that a communicating object paradigm is appropriate.
However, within the DOME, agents are useful when
designers are not certain about what modules can provide
the service they require. Agents could locate appropriate
modules.
2.3. Collaborative design information and support systems
There are many information and distributed computing
systems that have been built to support the collaborative design
modeling and decision-making process. According to their
different purposes and/or focuses, these systems can be
generally grouped into the following categories [30]:
(1) C
ollaborative product data/information management sys-tems for engineers to timely obtain the necessary product
data and knowledge [31–37].
(2) N
etwork-based collaborative design systems, which can befurther divided into web-independent [21,22,38–44], and
web-dependent [35,36,45–49] systems.
(3) P
rocess-centered collaborative design and workflow man-agement systems [50–54].
(4) C
onflict detection, management and resolution systems forcollaborative design [53,55–57].
(5) F
lexibility and security focused collaborative design system[58].
(6) I
nteroperability approaches in heterogeneous collaborativedesign systems [38,59,60].
Since most of them are developed for the needs of
collaborative design, current systems can assist designers in
one way or another in collaboration. Below, we provide more
details for some typical or notable systems and comparisons
between them. Of those collaborative design information and
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 43
support systems, notable ones include SHARE [14], NEXT-
LINK [61], DIS [26], DICE [25,84], EDN [26], MADEFast
[23], NIIIP [28], RaDEO [24], etc. Other typical proof-of-
concept systems include NetFeature [62], CooperativeAR-
CADE [63], WebSPIFF [64], CSCW-FeatureM [41], TOBACO
[65], CyberCAD [66], etc. There are also a number of
commercial software packages purporting to perform many of
the services that the system described supplies. Examples
include: Alibre (http://www.alibre.com), Co-Create (http://
www.cocreate.com), NexPrise (http://www.netprise.com),
CollabCAD (http://www.collabcad.com), etc. Modern product
data management (PDM) systems such as PTC Windchill and
UGS Teamcenter (iMAN) also provide some of the function-
ality described for collaborative design and development.
Table 1 gives an overview of some typical collaborative design
systems.
2.4. Summary of review and observations
From the above review, the requirements for network-based
collaborative design tools are highlighted as follows:
(1) A
Tabl
Ove
Syst
DIC
Alib
Co-C
(O
NetP
Win
Team
Coll
Cyb
NetF
Web
Coo
A
CSC
TOB
DOM
common framework enabling designers to capture, store
and retrieve knowledge efficiently and effectively through-
out the design process.
(2) C
omputer support that supplies clear and complete designknowledge, and also facilitates designer intervention and
customization during the decision-making activities in the
design process.
(3) M
ediation of the information flow between participants andsupport for a heterogeneous computing environment.
e 1
rview of some typical collaborative design systems
em Architecture Platform Communicatio
E Centralized Motif Compatible
Unix Workstation
Specialized M
Collaboration
re Centralized PC Windows TCP/IP, Direct
Cable Modem,
reate
neSpace.Net)
Centralized PC Windows Microsoft.NET
rise Centralized PC Windows Specialized M
Collaboration
dchill (PTC) Centralized PC Windows NT,
Unix Workstation
Java RMI, XM
center (UGS) Centralized PC Windows NT,
Unix Workstation
Microsoft.NET
SOAP, XML,
JSP Communi
aCAD Centralized PC Windows, Intel Linux Java RMI, Jav
erCAD Centralized Platform independent Point-Point Jav
eature Centralized PC Windows CORBA Comm
SPIFF Centralized PC Windows Java 3D Socke
perative
RCADE
Decentralized SGI, HP, SUN Workstation Specialized W
W-FeatureM Decentralized Motif Compatible
Unix Workstation
CORBA Comm
ACO Decentralized Solaris IRIX, Windows NT CORBA Comm
E Decentralized Platform independent CORBA Comm
(4) A
n
essag
(OO
X D
TI
essag
Hub
L
, J2E
catio
a3D
a R
uni
t
orkst
uni
uni
uni
n open environment so that models from new participants
can be added, allowing the design model and tools to evolve
with the design problem.
(5) A
means or mechanism for encapsulating expert knowledgeor know-how using an object-oriented approach, comparing
design solution alternatives and supporting the decision-
making process.
(6) E
mbodiment of a decentralized architecture that thecoordination between design participants and models is
not centrally modeled or controlled.
Further, the current status of research in knowledge-
intensive collaborative design modeling and support can be
summarized as follows. Although there have been efforts that
address design knowledge modeling and support, they all have
deficiencies in their representational capability. They only
focus on product knowledge modeling but none of them address
design process knowledge modeling as well as the relationship
between the product knowledge and the design process
knowledge. None yet provide explicit representation of the
design rationale as part of a comprehensive product knowledge
representation. It is also desirable to integrate a comprehensive
product model with a more formal representation of behavior.
Thus, they could not be directly used for design support. These
issues reinforce the need for consensus among product
development software developers and users regarding the
content and form for more comprehensive representation of
design knowledge.
The overview of the current collaborative design frame-
works and systems shows that they can assist designers in one
way or another in collaboration and their functionality can
Collaboration
granularity
Network
burden
Modeling technique
ing &
DMS)
Fine Heavy 2D/3D parametric
surface/solid modeling
SL,
or faster
Coarse Light 2D/3D, parametric
solid modeling
Fine Light 2D/3D Solid Modeling
ing/ Coarse Light Solid modeling
Coarse Light Feature-based modeling, etc.
E UDDI,
n Service
Coarse Light Parametric,
Feature-based modeling
Coarse Heavy Surface modeling,
solid modeling
MI Fine Heavy Solid modeling
cation Service Coarse Heavy Feature-based modeling
Fine Heavy Feature-based modeling
ation Network Coarse Light Solid modeling
cation Service Fine Light Feature-based modeling
cation Service Fine Light Solid/Feature modeling
cation Service Coarse Light Module modeling
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5544
eliminate a large amount of work during the design process
such as data/information sharing and exchange. However, they
do not provide a structured and formalized framework for
modeling the characteristics of multidisciplinary and multi-
objective design problems. They suffer one or more of the
following drawbacks: a heavy dependence on the experience
and knowledge of designers/users; no built-in ways for
selection or decision-making knowledge representation; no
efficient mechanisms to share, exchange, and reuse utilize the
given knowledge and guide the designer/user; difficult to
modify and not easy to extend/update/maintain knowledge
base, and no efficient mechanism to provide advisory services
and explain the results and what-ifs. Thus, they are unable to
effectively and efficiently support and coordinate highly
distributed and decentralized collaborative design and model-
ing activities. As a result, an intelligent collaborative design
tool that can provide efficient coordination and intelligent
decision-making mechanisms for designers is still needed. To
overcome the above drawbacks suffered in current systems and
improve current methods for collaborative design, this research
proposes a knowledge-intensive design modeling and support
scheme and develops a web-based advisory system to help
designers/users collaborate in making rapid and more
intelligent decisions during the design process.
3. Knowledge-intensive collaborative design: an
overview of the proposed approach
As reviewed above, distributed design systems have two
distinct forms [20–22]: distributed designers with access to
centralized resources, or distributed designers with distributed
resources (e.g., engineering models, databases, software
applications, etc.). This work focuses on the latter architecture.
Thus, the coordination between design participants and models
is not centrally modeled or controlled (analogous to the
WWW). The motivation and vision presented in this research
share some similar themes with Pahng et al. [20–22] but
emphasize knowledge-level or knowledge-intensive design
modeling, decision-making support, and search/optimization
and navigation. Even more, the proposed approach decomposes
design tasks into several ‘‘design components’’ or ‘‘modules’’
that can be developed separately by collaborative product
developers. We investigate the application to modular product
development problems, particularly the design of electro-
mechanical artifacts. The proposed distributed module-based
approach follows the component technology and distributed
object technology, which is called distributed module modeling
and evaluation (DMME) framework. In the component
technology, a component is a reusable application whose data
and methods are exposed and can be accessed and operated by
other applications. Thus, the DMME framework asserts that
multidisciplinary problems are decomposed into modular sub-
problems. Modularity divides overall complexity and dis-
tributes knowledge and responsibility amongst designers,
which facilitates the reuse of modeling elements. It also
facilitates the reuse of modeling elements. Therefore,
distributed modules allow designers to define mathematical
models or modules and integrate or interconnect them to form
large system models. In DMME, a multiple attribute decision
method is used to capture preferences and evaluate design
alternatives from different viewpoints. The resultant service-
exchange network forms an integrated concurrent systemmodel.
More details about the proposed approach are discussed below.
4. Distributed module modeling and evaluation for
design process
In this section, based on the design with modules scheme, a
distributed module modeling and evaluation framework is
developed for modeling the network-centric design process.
4.1. Design with modules scheme
Over half of industrial products have sub-systems or
components or modules as a part of their basic designs. The
modular design concept has been widely used in product design
for flexibility, rapid responsiveness, ease of maintenance and
rapid deployment. As a matter of fact, during the design
process, information processing is inherently model-based
because design object is structural in type. Therefore, an object
orientation scheme is employed so that both calculating and
reasoning work in design can be carried out. The integrated
design object model is, in fact, an attempt to set up a
knowledge-intensive framework in such a way that it becomes
possible to process various types of knowledge in a top-down
design process [29,67–69]. Object-oriented programming
technique allows designers to look at a design problem as a
collection of objects/modules or sub-problems linked together
by rules. Thus, it provides designers with an expressive power
to represent complex problems or information in an effective
manner. If designers can break a design problem into the form
of well defined operable chunks with their own self-containing
information, which are interrelated through a series of rules and
constraints, then the problem is a well object-oriented
programming application and conveniently to be solved (Pahng
et al. [20–22]).
The central design process inherent in design-with-modules
scheme could be represented as the architecture as shown in
Fig. 2 with five main types of objects involved: namely, design
models (S), design objects (O), design algorithms (A), functions
(requirements and constraints) (FRC), and the evaluation
schema (E) [70]. Object operators can express the relationship
between these objects: inheritance, import and message
passing. The architecture in Fig. 2 shows how the particular
instance of a design model, Sk1, is obtained from the design
algorithm, evaluation schema, requirements, constraints and
the design model object. For the pure formulation design or
creative design, a new design model object, S, is defined that
describes the form of the model. A specific instance, Sk1, of thisdesign model can then be created. For pure parametric design
then the design model object S has already been defined and the
design process therefore only involves the determination of a
specific instance, Sk1, of the design model. Note that additional
objects can be defined within the overall architecture.
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 45
Fig. 3. Module definition and embedded model.
Fig. 2. Overall architecture of design with modules (objects).
4.2. Module-based design and modularization process
A design model is created using object attribute variables
and relations between them. Modules are variable containers.
Variables can be grouped together into modules according to
logical, functional or physical component-based decomposi-
tions [71]. This grouping process is normally a bottom-up
decomposition as elementary entities are aggregated to build
more complex entities. A designer may also prefer to use a top-
down approach by defining high-level modules first and then
detailing the low-level variables and relations. Using both
bottom-up and top-down techniques, integrated design models
are built by interconnecting modules corresponding to different
sub-problems. Modules interact by service calls. A module
might request information from other modules to perform
internal computations and then provide the results as services to
other modules. A product module can be seen as a collection of
components that cover one or more sub-functions. It can be
defined by a combination of bill-of-materials (BOM) and
technical drawings. Thus, physically, a modular product system
is a collection of interchangeable modules (e.g., link and joint
units in modular robots) that can be assembled into many
different types and configurations of products. Unlike a
conventional industrial product that is designed for general
tasks or purposes, a modular product system has the advantage
of providing feasible product configurations for specific
customer or task requirements. As a result, the modular design
of products provides the ability to achieve product variety
through the combination and standardization of modules.
On the other hand, a partitioning or module is a building
block capable of performing calculations, reasoning and
exchanging information through service calls invoked by its
user. Defined as shown in Fig. 3, a module represents
knowledge related to different aspects of the design in the
form of variables and relations. The variables contained in the
module are represented as interconnected circles. The directed
arcs imply dependency. The outputs and inputs to the module
constitute the interface of the module. Modules can be
interconnected through interfaces. The calculations internal
to a module constitute its embedded model. Customer-created
computer programs and third-party applications, such as
domain specific analysis tools or CAD systems, can also be
embedded into a module. Modules interact with each other by
exchanging information and services, reacting to each other’s
changes for an integrated systemmodel. Modules can distribute
over the network and collectively form a distributed model for a
collaborative, multidisciplinary and concurrent design evalua-
tion. In this work, modules whose dependencies are to be
analyzed are tasks and parameters contained in design tasks,
and dependencies are constraints among parameters. Details
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5546
Fig. 4. Motor modules.
about modules can be found in [21]. For illustration, an
example module of motor is shown in Fig. 4. The module motor
can be described with IDL (interface description language).
Decomposing a design problem into modules and defining
how modules are related to one another creates the model of the
design problem. This is actually a partitioning or modulariza-
tion process, as shown in Fig. 5. The modularization process is
fulfilled through the following steps [72]:
(1) R
equirement analysis and modeling is carried out both fromtask or customer and designer viewpoints using design
Fig. 5. Modular design and mod
functions deployment technique and Hatley/Pirbhai tech-
nique [73,74]. A function–function interaction matrix is
generated.
(2) T
he combination of heuristic and quantitative clusteringalgorithms is used to modularize the product architecture,
and a modularity matrix is constructed.
(3) A
ll modules in the product are identified through themodularity matrix, and the types of all these modules can be
further identified according to the module classifications
(e.g., robot modules: link modules, actuator joint modules,
passive joint modules, etc.).
ularization process.
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 47
(4) F
unctional modules are mapped to structural modules usingthe function–structure interaction matrix. Module attribute
parameters or features can represent its structure.
(5) H
ierarchical building blocks (modules) are used torepresent the product architecture from both the functional
and the structural perspectives.
(6) O
ptimization algorithms (e.g., genetic algorithm, simulatedannealing algorithm) are used to optimize product
architecture to achieve one main objective. Other design
objectives are transformed into constraints for modules and
their attributes as well as their assemblies or configurations.
In addition, the cost or profit models can also be built as
system constraints.
Fig. 6. Distributed modules.
4.3. Distributed design modeling and evaluation
In a distributed design environment, each group of designers
can define their own modules, loading them into their local
work area and eventually connecting them to the other parts of
the design problem through appropriate networked interfaces
[20].
Fig. 6a shows a DMME model involving two designers and
three modules. Designer 1 defines modules A and B while
Designer 2 defines module C. Two domains communicate
through an Internet connection. Once the whole problem is
loaded and interconnected, each group of designers typically
has write access to local parts of the model, i.e., they can exert
decisions within their local range of influence, and read access
to relevant aspects of remote parts of the model. This allows
them to see the remote effects of their local decisions. Fig. 6b
Fig. 7. Module network modelin
shows that both Designers 1 and 2 might see the complete
problem, but with different access privileges. Designer 1 can
see modules A and B as local and module C as remote.
Conversely, module C is local to Designer 2. The remote part
seen by Designer 2 could showmodules A and B or just a single
distributed object (AB) if Designer 1 restricts their visibility/
access.
The modeling and implementation layers in DMME
[21] are illustrated in Fig. 7. At the modeling layer the
g and implementation [20].
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5548
designer defines the problem in terms of modules and
interactions among modules (Fig. 7a). In this example, an
implementation is provided only for the local modules A and
B, while module C is remote. The designer provides location
constraints for compatible modules that may be used as C.
The unseen implementation layer (Fig. 7b) is created to
provide the functionalities described in the modeling layer.
This layer locates remote modules. The remote modules
must be distributed objects capable of communicating
via a standard communication protocol. A distributed
interface is wrapped around the group of standard modules
(A and B in Fig. 7) to allow the local and distributed
modules to communicate with each other. The external
interface of this distributed module now offers service
calls to and from the remote module. Thus, the distributed
module is considered as a separate application that is capable
of providing services upon request in a design problem
model.
The interactions between distributed modules can be
achieved by publishing and subscribing services. The term
publish refers to making the services of one’s local model
visible together designers. The term subscribe refers to
making use of published services. Such design problem
models are mixed variable, where independent parameters
within modules are set and catalog selections might be used to
substitute entire modules. Design solutions can be assessed
and compared with each other using the decision-making
module embedded in the DMME framework.
Fig. 8. Product, design pr
5. Design knowledge modeling and integrated
knowledge representation scheme
In this section, a design knowledge modeling approach is
proposed. As stated before, both on-line and off-line design
knowledge representations are dealt with in this research, but
more attention is paid to the design process knowledge, as
shown in Fig. 8. Since the design knowledge is very extensive,
the focus is only on product and some selected activities in the
design process. A systematic methodology and the relevant
technologies are developed for knowledge modeling, i.e.,
capturing, representing, organizing and managing knowledge
in the design process.
Design knowledge is very domain specific, which comes
from a variety of sources, such as corporate libraries, textbooks/
handbooks, scientific literature, on-line resource, and in
particular, the experience of individual designers. It is stored
in different forms including documents, drawings, electronic
means and memories of human beings. Therefore, design
knowledge must be classified into different categories and
represented in appropriate ways accordingly. Through an
analysis of the design process, design knowledge is abstracted
and classified into four categories, off-line and on-line, product
knowledge and process knowledge. Each category of knowl-
edge is represented in different ways from multiple views.
Product knowledge includes all information needed for
designing product throughout the whole design process such
as product specifications, concepts, structure and geometry. In
ocess and knowledge.
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 49
view of this, a complete product information model is used,
which consists of customer requirements, design specifications,
modules (function—behavior, structure and geometry), module
interfaces, etc., as follows:
Product definition
requirements
specifications
artifacts
features
functions-behaviors-forms
performance objectives and constraints
relationships
design rationale
Product variety
assembly structure
module details
family parameters
product variants
In what follows, we concentrate on an integrated knowledge
representation scheme related to the design process. It deals
mainly with declarative representation, production rules and
object-oriented concepts. Procedural representation using
conventional languages such as C was not emphasized.
Integrating knowledge in its multiple forms, multiple levels
and multiple functions can fulfill design processes and
activities, especially for a more complex type. The integration
is very challenging, as the overall effect may be greater than the
sum of its parts. The integrated knowledge can solve problems,
which cannot be attained by the individual knowledge alone.
Based on a combination of elements of semantic relationships
with the object-oriented data model, a multi-level hierarchical
representation schema (meta-level, physical level, geometric
level) is adopted to represent the modular design process
knowledge in different design stages at different levels. To
effectively manage and utilize design knowledge, a generalized
design knowledge matrix is proposed for organizing design
knowledge. All tasks in the design process are listed in column
while all information and design knowledge is categorized in
rows. The contents of design knowledge for each task are
recorded in the corresponding cells of the design knowledge
matrix with appropriate representations.
More specifically, the object-oriented knowledge represen-
tation is based on a mixed representation method and object-
oriented programming (OOP) techniques, and allows designers
to look at the design problem as a collection of objects or sub-
problems linked together by rules. Thus, it provides the
designers with an expressive power to represent complex
problems or information in an effective manner. The basic
structure of this representation is described as a module. The
class of an object and its instances are described by the module
structure. An object-oriented module is composed of four types
of slots, which are the attribute slot, relation slot, method slot
and rule slot as follows:
(1) T
he attribute slots are used for describing the staticattributes (variables) of design object.
(2) T
he relation slot is used for describing the static relationsamong objects. With the help of the relation slot and
according to the relation of classification, the design
object can be described as a hierarchical structure. Its
class and sub-classes can share the knowledge in super
class. The messages that control the design process can be
sent among all instances of objects. In addition, if needed,
other kind of relation slots can be defined, such as the
resolution, position and assembly, etc. These slots create
the foundation for describing a graph in design. The
hierarchical structure of object-oriented knowledge
representation is formed.
(3) T
he method slot is used for storing the methods of design,sending messages and performing procedural control and
numerical calculation.
(4) T
he rule slot is used for storing sets of production rules. Theproduction rules can be classified according to the
constraints/differences among objects being treated
and stored respectively in rule slots in the form of slot
value.
Thus, the integrated knowledge representation scheme
realizes the advantages of both object-oriented representation
and rule-based representation.
6. Web-based knowledge-intensive collaborative design
framework
In this section, a web knowledge server-based distributed
module modeling and evaluation (KS-DMME) is proposed for
distributed design modeling.
6.1. Knowledge-based systems as knowledge servers
The widespread use of the Internet and WWW provides an
opportunity for making expert systems widely available. By
implementing knowledge-based systems (KBS) as knowledge
servers that perform their tasks remotely, developers can
publish expertise on the Web. Technologies and infrastruc-
tures that make this approach feasible are emerging.
Simultaneously, the interest in artificial intelligence support
for network navigation services is growing. Wide-area
networks and the internet-based WWW allow developers to
provide intelligent knowledge servers [75]. Knowledge-based
expert systems running on servers can support a large-scale
group of users who communicate with the system over the
network. In this approach, user interfaces based on web
protocols provide access to the knowledge servers, and
users do not need special hardware or software to consult
these services with appropriate web browsers. To make
knowledge servers available, developers must distribute the
software front ends that allow users to communicate with the
servers. The remaining parts are actually concerning how
KBS technology can be used to assist designer in navigating
design knowledge present on the WWW, drawing on an
alternative view of KBS, and making decisions in design
process.
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5550
6.2. KS-DMME framework architecture
As discussed above, KS-DMME adopts the design with
modules, modules network and knowledge server paradigms,
which are techniques to utilize the connectivity of knowledge-
based systems provided by the internet to increase the size of
the user base whilst minimizing distribution and maintenance
overheads. Therefore, modules under the KS-DMME frame-
work are connected together so that they can exchange services
to form large integrated knowledge models. The module
structure of KS-DMME leads itself to be a client (browser)/
knowledge server oriented architecture using the distributed
object technology. Fig. 9 shows the main system components of
the proposed client (browser) knowledge server architecture.
Each of these components interacts with one another using a
communication protocol, CORBA, so that it is not required to
maintain the elements on a single machine. As a gateway for
providing services, the interface of a system component
invokes the necessary actions to provide requested services. To
request a service, a system component must have an interface
pointer to the desired interface [21,22]. The resultant service-
exchange network forms an integrated concurrent system
model when module services are connected. The distinct
characteristics of the KS-DMME architecture are described as
follows:
(1) K
S-DMME is well suited for loose and flexible collabora-tions as it is an open environment and allows for true
knowledge encapsulation. The centralized multi-user
system architecture could be supported within a module
Fig. 9. (a) Client-knowledge server architecture and
so that multiple users have the ability to access to the
centralized module.
(2) K
S-DMME is intended to provide concurrent systemmodeling functionality, i.e., concurrent interactions
between designers and models. When a designer receives
a model or data from another designer, he/she works on the
design and sends the result of design modification to others.
(3) I
n KS-DMME, the interactions between sub-problems areexplicitly defined through design negotiation so that a
communicating object paradigm is appropriate. Within the
KS-DMME, agents are useful when designers are not
certain about what modules can provide the service they
require. Agents could locate appropriate modules [30].
6.3. Module interactions for exchanging services
The KS-DMME architecture allows designers or experts to
publish and subscribe design modeling and decision support
services on the. These services operate when information is
received from other clients or knowledge servers. When
module services are connected, the resultant service exchange
network forms a concurrent integrated system model. Any
service requests in the module network can invoke a chain of
service requests if needed to provide correct information. When
a design alternative is evaluated, the local module asks for the
services of subscribed modules. If the subscribed modules
themselves need services from other modules in order to
provide the request services, they again request those services
from their own network to remote models. Thus, the service
requests are propagated through the connected modules.
(b) main components for KS-DMME.
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 51
Fig. 10. Service exchanges between distributed modules.
However, the complete system may not be visible to any given
modules. Since modules can only interact through services, it is
possible for a module or local model to encapsulate its internal
modules and hide intellectual property (IP) if desired. Before a
designer publishes, they can assign access privileges for their
services. Three levels of modules access have been identified:
owner, builder and user. The owner is the original creator of the
module and has access to all the services defined in the model
and control over their publication. The builder can see the
internal details of a model the owner chooses to make public
and can add new modules. However, they cannot destroy
Fig. 11. Simple distributed design model with two modules and a remo
modules created by the owner or other builders. The users can
only subscribe the published services.
The KS-DMME framework provides the methods and
interfaces needed for the interaction with other modules in a
networked environment. These interactions are graphically
depicted in Fig. 10. When Designer B makes a change, the
service corresponding to the request from Designer A will
reflect this change. The enumerated request shows the sequence
for obtaining the service needed by Designer A. The light gray
module seen by Designer A is the remote module published by
Designer B. For more details, please refer to [20–22].
te module: (a) modules A and B and (b and c) remote module AB.
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5552
Table 2
The description of the distributed model
Local modules Remote module
module A ( module Remote_AB (
Variable A1( ) URL: 159.69.1.19//IP address
Variable A2( ) Receive a2
Constraint a1 RuleSet AB (Rule AB1:
Constraint b2 IF (. . .)
EmbeddedModel calculateA2
(A2 = f1(a1,b2))
THEN (. . .)
RuleSet A (Rule A1: )
IF (a1=) )
THEN (. . .) module C (
) Variable C1( )
) Variable C2( )
module B ( Constraint c1
Variable B1( ) Constraint a2
Variable B2( ) EmbeddedModel calculateC2
(C2 = f3(a2,c1))
Constraint b1 RuleSet C (Rule C1:
EmbeddedModel
calculateB2 (B2 = f2(b1))
IF (. . .)
RuleSet B (Rule B1: THEN (. . .)
IF (. . .) )
THEN (. . .) )
)
)
Design: Two-module design ( Design: design with a
remote module (
Module A Module AB
Module B Module C
) )
6.4. Design modules network formulation
As discussed above, the modularization process decomposes
a design problem into modules and defines how modules are
related to one another. The relationships among modules
specify how outputs of a module are connected to the inputs of
other modules. The embedded model of a module produces
outputs using its internal design resources as well as inputs
from other modules. Fig. 11 illustrates a simple distributed
module network model used for a design process. The variables
of the model are governed by a set of equations and/or rules. As
depicted in Fig. 11, the interface connections between variables
in different modules (e.g., modules A and B) can be established
interactively or defined explicitly using a modified model
definition language (MMDL) (Phang et al. [20–22,15]). The
embedded models defined with the variable declaration can also
be created separately and linked to the model definition using
keywords. Modules A and B are local to the problem. Using the
remote module AB, a new design model (ABC) can be created.
As such, the problem model is available for use as a distributed
module with the outward appearance in Fig. 11a. These
distributed modules allow users to utilize variables and their
dependencies such as module A (A1, A2, a1, !b2), module B
(B1, B2, b1, b2 !), module C (C1, C2, c1,!a2) and module AB
(A1, A2, B1, B2, a1, a2 !, b1, b2). Fig. 11b illustrates the model
from the viewpoint of the ABC designer. Module C is local to
the designer. Fig. 11c illustrates the true integrated model
created when the remote module AB and the local module C are
connected. The problem model ABC is thus created, which
Fig. 12. Module network configuration under KS-DMME framework.
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 53
requires additional information such as the distributed module’s
name and IP address. The description of the distributed model
can be illustrated in Table 2. It is shown that the relations
between modules do not need to be changed even if the
embedded model of a remote module (i.e., module AB) is
changed. This flexibility enables a designer to define a model
independently from the actual location (i.e., local or remote) of
embedded models. When the designer utilizes the remote
module AB in conjunction with the local module C, the
resulting integrated model forms a distributed computing
system comprised of two autonomous computing elements.
Fig. 12 illustrates the configuration process for distributed
modules using the system components, including the Internet
and web resources. The embedded model of the module AB in
design problem model ABC contains module connector that
manages the design information exchange with the distributed
design module AB.
7. Summary and conclusions
In this paper, we presented a design-with-modules scheme,
distributed modeling and a web-based knowledge-intensive
framework (KS-DMME) for collaborative design modeling and
support. Because of the heterogeneous structure, product design
and simulation requires different grades of abstraction and needs
the cooperation and collaboration of different disciplines and
resources. The developed framework can provide distributed
designers with an intelligent tool to collaboratively build
integrated system models. The advantage of the demonstrated
module concept consists in the flexibility of the structure and the
reduction of costly software support by integrating design
tools and simulators. Large problems are decomposed into
sub-problems with modules; models or other software applica-
tions are encapsulated in modules. A module can provide
information services through its interface, and the network of
modules exchanging services form a concurrent intelligent
design model. Therefore, the behavior of complex systems and
the interactions of components/modules can be analyzed and
optimized during the design process, resulting in shorter design
cycles.
As the knowledge server-based framework was built to
provide the module network architecture for integrating
intelligent modeling services available on the network, it can
accommodate top-down and bottom-up approaches in the
context of both the traditional sequential design processes and
the concurrent intelligent design. In the module network,
design resources, models, data and activities are not
centralized nor concentrated in one location, but distributed
among many companies, designers, or design participants
working together over the web. The module network
architecture is extended to a network-centric environment,
focusing on the knowledge-intensive and collaborative design
modeling and support. Design modules and modules network
are created by fully implementing the locally defined modules
and subscribing to the services of the remote modules. Design
modules and modules network are created. In the module
network architecture, when modules services are connected,
the resultant service exchange network creates an integrated
concurrent intelligent system model or module network
that invoke a chain of service requests if needed to
provide correct information. The following paper (Part II)
will present in detail the implementation and application of
the web-based knowledge-intensive collaborative design
support system.
Compared to existing or ongoing research efforts, the
framework presented in this paper differs in its focus to create a
concurrent intelligent design modeling scheme that handles the
different variable types and knowledge needed in engineering
design, integrate multiple objective evaluation and optimiza-
tion with design models, and provide an object-oriented design
methodology to facilitate the intelligent integration of design
models and their utilization in an open and distributed
intelligent design environment.
Acknowledgements and disclaimer
The bulk of the work was done while the first author was
with Nanyang Technological University and Singapore
Institute of Manufacturing Technology (SIMTech), Singapore.
The authors would like to express their gratitude to the
anonymous reviewers of this paper for their valuable comments
and suggestions.
References
[1] M. Stokes, Managing Engineering Knowledge: MOKA Methodology for
Knowledge Based Engineering Applications, MOKA Consortium, Lon-
don, 2001.
[2] S. Szykman, R.D. Sriram, W. Regli, The role of knowledge in next-
generation product development systems, ASME Journal of Computing
and Information Science in Engineering 1 (1) (2001) 3–11.
[3] S. Szykman, Architecture and implementation of a design repository
system, in: Proceedings of ASME DETC2002, 2002, Paper No.
DETC2002/CIE-34463.
[4] S.J. Fenves, A core product model for representing design information,
NISTIR 6736, NIST, Gaithersburg, MD, 2001.
[5] X.F. Zha, R.D. Sriram, Feature-based component model for design of
embedded system, in: B. Gopalakrishnan (Ed.), Intelligent Systems in
Design and Manufacturing, Proceedings of SPIE, vol. 5605, SPIE, Bel-
lingham, WA, vol. V, 2004, pp. 226–237.
[6] F. Pena-Mora, R.D. Sriram, R. Logcher, SHARED DRIMS: SHARED
design recommendation and intent management system, in: Enabling
Technologies: Infrastructure for Collaborative Enterprises, IEEE Press,
1993, pp. 213–221.
[7] F. Pena-Mora, R.D. Sriram, R. Logcher, Conflict mitigation system for
collaborative engineering, AI EDAM—Special Issue of Concurrent Engi-
neering 9 (2) (1995) 101–123.
[8] W.H.Wood III, A.M. Agogino, Case based conceptual design information
server for concurrent engineering, Computer-Aided Design 8 (5) (1996)
361–369.
[9] L.T.M. Blessing, A process-based approach to computer supported engi-
neering design, Ph.D. Thesis, University of Twente, 1993.
[10] L. Patil, D. Dutta, R.D. Sriram, Ontology-based exchange of product data
semantics, IEEE Transactions on Automation Science and Engineering 2
(3) (2005) 213–225.
[11] A.M. Madni, The role of human factors in expert systems design and
acceptance, Human Factors 30 (4) (1988) 395–414.
X.F. Zha, H. Du /Computers in Industry 57 (2006) 39–5554
[12] P.A. Rodgers, A.P. Huxor, N.H.M. Caldwell, Design support using dis-
tributed web-based AI tools, Research in Engineering Design 11 (1999)
31–44.
[13] S. King, Co-Design: a Process of Design Participation, Van Nostrand
Reinhold, New York, 1989.
[14] G. Toye, M.R. Cutkosky, J.M. Tenenbaum, J. Glicksman, SHARE: a
methodology and environment for collaborative product development, in:
Proceedings of Second Workshop on Enabling Technologies: Infrastruc-
ture for Collaborative Enterprises, Morgantown, West Virginia, 1993), pp.
33–47.
[15] J. Siegel, CORBA: Fundamentals and Programming, OMG, 1996.
[16] D. Chappel, Understanding ActiveX and OLE, Microsoft Press, Red-
mond, WA, 1996.
[17] F.J. Wang, J.J. Mills, V. Devarajan, A conceptual approach managing
design resource, Computers in Industry 47 (2002) 169–183.
[18] M. Rosenman, F.J. Wang, CADOM: a component agent-based design-
oriented model for collaborative design, Research in Engineering Design
11 (1999) 193–205.
[19] M. Rosenman, F.J. Wang, A component agent-based open CAD system for
collaborative design, Automation in Construction 10 (2001) 383–397.
[20] F. Pahng, N. Senin, D.R. Wallace, Modeling an evaluation of product
design problems in a distributed design environment, in: Proceedings of
ASME DETC, Sacramento, CA, 1997 (CD ROM).
[21] F. Pahng, N. Senin, D.R.Wallace, Distribution modeling and evaluation of
product design problems, Computer-Aided Design 30 (6) (1998) 411–423.
[22] F. Pahng, S.H. Bae, D.R. Wallace, Web-based collaborative design
modeling and decision support, in: Proceedings of DETC’98, Atlanta,
Georgia, USA, 1998.
[23] MADEFast, http://madefast.stanford.edu/, 1999.
[24] RaDEO, http://elib.cme.nist.gov/radeo/, 1998.
[25] R.D. Sriram, R. Logcher, The MITDICE project, IEEE Computer (1993)
64–65.
[26] P.I. Bliznakov, J.J. Shah, D.K. Jeon, S.D. Urban, Design information
system infrastructure to collaborative design in a large organization, in:
Proceedings of ASME DETC’95, vol. 1, Boston, MA, 1995), pp. 1–8.
[27] J.W. Lewis, K.J. Singh, Electronic design notebooks (EDN): technical
issues, in: Proceedings of Concurrent Engineering: a Global Perspective,
McLean, VA, 1995), pp. 431–436.
[28] NIIIP, http://www.niiip.org/, 1999.
[29] X.F. Zha, Knowledge intensive methodology for intelligent design and
planning of assemblies, Ph.D. Thesis, Nanyang Technological University,
Republic of Singapore, 1999.
[30] Y.L. Li, X.Y. Shao, P.G. Li, Q. Liu, Design and implementation of a
process-oriented intelligent collaborative product design system, Com-
puter in Industry 53 (2004) 205–229.
[31] Y.M. Chen, M.W. Liang, Design and implementation of a collaborative
engineering information system for allied concurrent engineering,
International Journal of Computer Integrated Manufacturing 13 (1)
(2000) 11–30.
[32] M. Hardwick, D.L. Spooner, T. Rando, K.C. Morris, Sharing manufactur-
ing information in virtual enterprise, Communications of the ACM 39 (2)
(1996) 46–54.
[33] Y.P. Zhang, C. Zhang, H.P. Wang, An Internet based STEP data exchange
framework for virtual enterprises, Computers in Industry 41 (1) (2000)
51–63.
[34] M. Rezayat, The enterprise-web portal for life-cycle support, Computer-
Aided Design 32 (2) (2000) 85–96.
[35] Y. Kim, S.-H. Kang, S.-H. Lee, S.B. Yoo, A distributed, open intelligent
product data management system, International Journal of Computer
Integrated manufacturing 14 (2) (2001) 224–235.
[36] Y. Kim, Y. Choi, S.B. Yoo, Brokering and 3D collaborative viewing of
mechanical part models on the web, International Journal of Computer
Integrated Manufacturing 14 (1) (2001) 28–40.
[37] P.-Y. Chao, Y.-C. Wang, A data exchange framework for networked CAD/
CAM, Computers in Industry 44 (2) (2001) 131–140.
[38] S. Abrahamson, D. Wallace, N. Senin, P. Sfereo, Integrated design
in a service marketplace, Computer-Aided Design 32 (2) (2000)
97–107.
[39] Y.-C. Kao, G.C.I. Lin, Development of a collaborative CAD/CAM system,
Robotics and Computer-Integrated Manufacturing 14 (1) (1998) 55–68.
[40] N. Shyamsundar, R. Gadh, Internet-based collaborative product design
with assembly features and virtual design spaces, Computer-Aided Design
33 (9) (2001) 637–651.
[41] A. Stork, U.V. Lukas, R. Schults, Enhancing a commercial 3D CAD
system by CSCW functionality for enabling cooperative modelling via
WAN, in: Proceedings of 1998 ASME Design Engineering Technical
Conference, Atlanta, GA, 13–16 September, 1998 (CD).
[42] G.W. Tan, C.C. Hayes, M. Shaw, An intelligent-agent framework for
concurrent product design and planning, IEEE Transactions on Engineer-
ing Management 43 (3) (1996) 297–306.
[43] J. Sun, Y.F. Zhang, A.Y.C. Nee, A distributed multi-agent environment for
product design and manufacturing planning, International Journal of
Production Research 39 (4) (2001) 625–645.
[44] R.I. Whitfield, A.H.B. Duffy, G. Coates, W. Hills, Distributed design
coordination, Research in Engineering Design 13 (2002) 243–252.
[45] F.-C.F. Wang, P.K. Wright, Web-based CAD tools for a networked
manufacturing service, in: Proceedings of 1998 ASME Design Engineer-
ing Technical Conferences, Atlanta, GA, 13–16 September, 1998 (CD).
[46] C.S. Smith, P.K. Wright, CyberCut: a World Wide Web based design-to-
fabrication tool, Journal of Manufacturing Systems 15 (6) (1996) 432–
442.
[47] C.-Y. Kim, S.-H. Kang, N. Kim, P. O’Grady, Internet-based concurrent
engineering: an interactive 3D system with markup, in: Proceedings of
ASME 18th Computers in Engineering Conference, Atlanta, GA, 13–16
September, 1998 (CD).
[48] K. Cheng, P.Y. Pan, D.K. Harrison, Web-based design and manufacturing
support systems: implementation perspectives, International Journal of
Computer Integrated Manufacturing 14 (1) (2001) 14–27.
[49] U. Roy, S.S. Kodkani, Collaborative product conceptualization tool using
web technology, Computers in Industry 41 (2) (2000) 195–209.
[50] G.Q. Huang, K.L. Mak, Web-integrated manufacturing: recent develop-
ments and emerging issues, International Journal of Computer Integrated
Manufacturing 14 (1) (2001) 3–13.
[51] S.C.Y. Lu, J. Cai, W. Burkett, F. Udwadia, A methodology for collabora-
tive design process and conflict analysis, Annals of CIRP 49 (1) (2000)
69–73.
[52] G.Q. Huang, J. Huang, K.L. Mak, Agent-based workflow management in
collaborative product development on the Internet, Computer-Aided
Design 32 (29) (2000) 133–144.
[53] Y. Jin, L. Zhao, A. Raghunath, ActiveProcess: a process-driven and agent-
based approach to supporting collaborative engineering, in: Proceedings
1999 ASME Design Engineering Technical Conferences, Las Vegas, NV,
12–16 September, 1999 (CD).
[54] H. Kim, J.Y. Lee, S.-B. Han, Process centric distributed collaborative
design based on the web, in: Proceedings of ASME 19th Computers in
Engineering Conference, Las Vegas, NV, 12–15 September, 1999 (CD).
[55] S.T.C. Wong, Coping with conflict in cooperative knowledge based
systems, IEEE Transactions on Systems, Man and Cybernetics—Part
A. Systems and Humans 27 (1) (1997) 57–72.
[56] M. Klein, Supporting conflict resolution in cooperative design systems,
IEEE Transactions on Systems, Man and Cybernetics 21 (6) (1991) 1379–
1390.
[57] M.P. Case, S.C.-Y. Lu, Discourse model for collaborative design, Com-
puter-Aided Design 28 (5) (1996) 333–345.
[58] L.M. Camarinha-Matos, H. Afsarmanesh, A.L. Osorio, Flexibility and
safety in a web-based infrastructure for virtual enterprises, International
Journal of Computer Integrated Manufacturing 14 (1) (2001) 66–82.
[59] G. Zhao, J. Deng,W. Shen, CLOVER: an agent-based approach to systems
interoperability in cooperative design systems, Computers in Industry 45
(3) (2001) 261–276.
[60] C.S. Han, J.C. Kunz, K.H. Law, An internet-based distributed service
architecture, in: Proceedings of the 1999 ASME Design Engineering
Technical Conferences, Las Vegas, NV, 12–15 September, 1999 (CD).
[61] C. Petrie, M. Cutkosky, H. Park, Design space navigation as a collabora-
tive aid, in: Proceedings of Third International Conference on Artificial
Intelligence in Design, Lausanne, Switzerland, 1994.
X.F. Zha, H. Du / Computers in Industry 57 (2006) 39–55 55
[62] J.Y. Lee, H. Kim, S.B. Han, Network-centric feature-based modeling, in:
Proceedings of Pacific Graphics’99, Seoul, Korea, 1999), pp. 280–289.
[63] A. Stork, U. Jansnoch, A collaborative engineering environment, in:
Proceedings of the TeamCAD’97 Workshop on Collaborative Design,
Atlanta, 1997), pp. 25–33.
[64] R. Bidarra, E. van den Berg, W.F. Bronsvoort, Collaborative modeling
with features, in: Proceedings of 2001 ASME Design Engineering Tech-
nical Conferences, Pittsburgh, 2001 (Paper No. DETC2001/CIE-21286).
[65] U.V. Lukas, Collaborative geometric modeling using CORBA services, in:
Advance Proceedings of the ECSCW’97 Workshop on Object-Oriented
GroupWare Platform (OOGP’97), Lancaster, UK, 1997), pp. 91–92.
[66] F.E.H. Tay, A. Roy, CyberCAD: a collaborative approach in 3D-CAD
technology in a multimedia-supported environment, Computer in Industry
52 (2003) 127–145.
[67] J.S. Gero, Design prototypes: a knowledge representation schema for
design, AI Magazine 11 (4) (1990) 6–36.
[68] J.S. Gero, Knowledge-Based Design Systems, Addison-Wesley, Reading,
MA, 1990.
[69] S.R. Gorti, A. Gupta, G.J. Kim, R.D. Sriram, A. Wong, An object-oriented
representation for product and design process, Computer-Aided Design 30
(7) (1998) 489–501.
[70] P. O’Grady, W.Y. Liang, An object-oriented approach to design with
modules, Iowa Internet Laboratory Technical Report TR98-04, 1998.
[71] N. Senin, D.R. Wallace, N. Borland, M.J. Jakiela, A framework for mixed
parametric and catalog-based design problem modeling and optimization,
MIT CAD Lab-Technical Report 97.02, 1997a.
[72] X.F. Zha, W.F. Lu, Knowledge intensive support for product family
design, Proceedings of ASME, DETC 02/DAC-34098, 2002a.
[73] S. Sivaloganathan, P.T.J. Andrews, T.M.M. Shahin, Design function
deployment: a tutorial introduction, Journal of Engineering Design 12
(1) (2001) 59–74.
[74] G.J. Rushton, A. Zakarian, Development of Modular Vehicle Systems,
Department of Industrial and Manufacturing Systems Engineering, Uni-
versity of Michigan, Dearborn, 2000.
[75] H. Eriksson, Expert systems as knowledge servers, IEEE Expert 14 (3)
(1996) 14–19.
[76] P.I. Bliznakov, Design information framework to support engineering
design process, Ph.D. Dissertation, Arizona State University, 1996.
[77] C.A. Ellis, S.J. Gibbs, G. Rein, Groupware: some issues and experience,
Communication of ACM 34 (1) (1991) 38–58.
[78] M. Klein, Capturing geometry rationale for collaborative design, in:
Proceedings of the 1997 6th IEEE Workshops on Enabling Technologies:
Infrastructure for Collaborative Enterprises, WET-ICE, Cambridge, MA,
USA, 18–20 June, Journal of Engineering and Applied Science 1997) 24–
28.
[79] G. Konduri, A. Chandrakasan, Framework for collaborative and distrib-
uted web-based design, in: Proceedings of Design Automation Confer-
ence, Proceedings of the 1999 36th Annual Design Automation
Conference (DAC), New Orleans, LA, USA, 21–25 June, 1999), pp.
898–903.
[80] A. Kusiak, J. Wang, Dependency analysis in constraint negotiation, IEEE
transactions on systems, Man and Cybernetics 25 (9) (1995) 1301–1313.
[81] S.E. Lander, Issues in multi-agent design systems, IEEE Expert: Intelli-
gent System and Their Application 12 (2) (1997).
[82] J. Owen, STEP—an Introduction, Winchester, 1993.
[83] U. Roy, B. Bharadwaj, S.S. Kodkani, M. Cargian, Product development in
a collaborative design environment, Concurrent Engineering Research and
Applications 5 (4) (1997) 347–365.
[84] R.D. Sriram, Distributed and Integrated Collaborative Engineering
Design, Sarven Publishers, Glenwood, MD 21738, December, 2002.
[85] A.K. Singh, CONSENS—an IT solution for concurrent engineering, in:
Proceedings of Concurrent Engineering: a Global Perspective, McLean,
VA, 1995), pp. 635–644.
[86] N. Senin, N. Borland, D.R. Wallace, Distributed modeling of product
design problems in a collaborative design environment, in: CIRP Inter-
national Design Seminar Proceedings: Multimedia Technologies for
Collaborative Design and Manufacturing, Los Angeles, CA, 8–10 Octo-
ber, 1997.
[87] M.W. Sobolewski,W. Erkes J., CAMnet: architecture and applications, in:
Proceedings of Concurrent Engineering: a Global Perspective, McLean,
VA, 1995), pp. 627–634.
[88] R. Sudarsan, Y.H. Han, S.C. Feng, U. Roy, F. Wang, R.D. Sriram, K.
Lyons, Object-oriented representation of electro-mechanical assemblies
using UML, NISTIR 7057, NIST, Gaithersburg, MD, 2003.
[89] K.-L. Wang, Y. Jin, Managing dependencies for collaborative design, in:
Proceedings of 2000 ASME Design Engineering Technical Conferences
and Computers and Information in Engineering Conference, Baltimore,
MA, 10–13 September, 2000 (CD).
[90] A.W. Westerberg, R. Coyne, D. Cuningham, A. Dutoit, E. Gardner, S.
Konda, S. Levy, I. Monarch, R. Patrick, Y. Reich, E. Subrahmanian, M.
Terk, M. Thomas, Distributed and collaborative computer-aided environ-
ment in process engineering design, in: Proceedings of ISPE, 1995.
[91] X.F. Zha, M.F. Fu, W.F. Lu, S. Ma, C.F. Zhu, Knowledge modeling in
design process-knowledge modeling in product family design for mass
customization, SIMTech Technical Report, MIT/02/031/PDD, Singapore,
2002.
Xuan F. Zha is currently with the Design and
Process Group, Manufacturing System Integration
Division, National Institute of Standards and Tech-
nology (NIST), USA. He is a guest professor of
Shanghai JiaoTong University (China). He received
his PhD from Nanyang Technological University,
Singapore. He is a senior member of IEEE and
SME. His current research interests are in virtual
and collaborative product design and development,
product lifecycle management, knowledge manage-
ment in design, artificial intelligence and soft computing in design and
manufacturing, supply chain management, micro electro mechanical systems
design and simulation, embedded systems, robotics and so forth. Dr. Zha has
authored or coauthored over 100 papers, book chapters and reports in these
areas.
H. Du is currently at School of Mechanical and
Production Engineering, Nanyang Technological
University, Singapore. He obtained his BEng and
MEng from Nanjing University of Aeronautics and
Astronautics, Nanjing, China, in 1983 and 1986,
respectively. He obtained his PhD from Imperial
College of Science, Technology and Medicine, Lon-
don, UK, in 1991. Since then, he has been working at
School of Mechanical and Production Engineering,
Nanyang Technological University, Singapore. His
research interests include micro-electro-mechanical system (MEMS), topology
optimization design, concurrent design as well as smart materials and struc-
tures.