Response to
UML for Systems Engineering Request for Information (SE DSIG RFI 1)
OMG Document # ad/2002-01-17
http://syseng.omg.org/
Part 1: Overview of the Constrained Object (COB)
Engineering Knowledge Representation
May 31, 2002
Russell S. Peak
Georgia Institute of Technology
Manufacturing Research Center
813 Ferst Drive MARC 452
Atlanta, GA 30332-0560 USA
russell.peak@marc.gatech.edu
http://www.marc.gatech.edu/
1 Introduction
This document responds to the subject RFI primarily by overviewing a potentially
relevant engineering knowledge representation known as constrained objects (COBs).
Part 1 is being submitted now as an input for the June 2002 OMG Technical Committee
meeting in Orlando. Part 2 entitled “Potential Applications of Constrained Object
(COBs) to the UML for Systems Engineering Effort” will address the specific questions
in the RFI. It is anticipated that Part 2 will be submitted as an input for the
September 2002 meeting. In brief, COB concepts like constraint schematics may help
form the basis for UML extensions that will be useful for systems engineering.
2 Constrained Object Overview
2.1 Description
Constrained objects (COBs) have been developed over the past 10 years as a means for
integrating design models with diverse analysis1 models (see the Bibliography for
pointers to the main references). Design and analysis information is typically represented
by a collection of interrelated models of varying discipline and fidelity. Thus a means for
capturing diverse multi-fidelity models and their fine-grained relations was needed. It
was also desirable for this method to be independent of the specific CAD/E tools used to
create, manage, and compute these models.
1
In this context “analysis” and “simulation” denote the engineering modeling of physical behavior such as stress,
temperature, and voltage.
1
The COB representation is based on object and constraint graph concepts to gain their
modularity and multi-directional capabilities. Object techniques provide a semantically
rich way to organize and reuse the complex relations and properties that naturally
underlie engineering models. Representing relations as constraints makes COBs flexible
because constraints can generally accept any combination of I/O information flows. This
multi-directionality enables design sizing and design verification using the same COBbased analysis model. Engineers perform such activities through out the design process,
with the former being characteristic of early design stages and vice versa.
The COB representation includes several modeling languages. It has lexical forms that
are computer interpretable as well as graphical forms that aid human comprehension. For
example, the graphical constraint schematic notation, which has strong electrical
schematic analogies, emphasizes object structure and relations among object attributes.
The references given below include tutorial examples of the main COB concepts. To
help validate the COB representation, other work describes electronic packaging and
aerospace test cases implemented in a toolkit called XaiTools™ (an example embodiment
of COB concepts). In all, the test cases utilize some 260 different types of COBs with
some 370 relations, including automated solving using commercial math and finite
element analysis tools. Today some of these COBs are in production usage to help
automate chip package thermal resistance analysis.
Results demonstrate that the COB representation enhances physical behavior modeling
and knowledge capture for a wide variety of design models, analysis models, and
engineering computing environments.
2.2 Summary of Features
The constrained object (COB) knowledge representation has these overall
characteristics:
 Declarative knowledge representation (non-causal)
 Combination of objects and constraint graph techniques
 COBs  (STEP EXPRESS2 subset) + (constraint graph concepts and views).
Test cases and production usage show that COBs provide these advantages compared to
traditional analysis representations:
 Greater solution control
 Richer semantics (e.g., equations wrapped in engineering context)
 Unified views of diverse capabilities (independent of CAD/E methods and tools)
 Capture of reusable knowledge
 Enhanced development of complex analysis models
2
STEP EXPRESS [ISO 10303-11] is an object-flavored information modeling standard geared towards the life cycle
design and engineering aspects of a product. For further information, see http://www.nist.gov/sc4/ . EXPRESS-G,
the graphical form of EXPRESS, is roughly equivalent to the UML class diagram (minus the methods).
2
Technical capabilities and applications are summarized here:
 Capabilities & features:
– Various forms: computable lexical forms, graphical forms, etc.
» Enables both computer automation and human comprehension
– Sub/supertypes, basic aggregates, multi-fidelity objects
– Multi-directionality (I/O changes)
– Reuses external programs as white box relations
– Advanced associativity added to COTS frameworks & wrappers

Analysis module/template applications (XAI/MRA):
–
–
–
–
–
Analysis template languages
Product model idealizations
Explicit associativity relations with design models & other analyses
White box reuse of existing tools (e.g., FEA, in-house codes)
Reusable, adaptable analysis building blocks
– Synthesis (sizing) and verification (analysis)
Target users and associated business benefits include the following:

COB end user : Designer
(uses COB instances & COB-based applications)
– Automation  Time savings & consistency
– More analysis  Improved designs

COB creator : Analyst
(creates templates with COB definition language)
– Modularity & reusability  Faster, consistent modeling
– Semantic richness  Increased understanding
– Knowledge capture  Enhanced corporate memory

COB application developer: Programmer
(uses COB API to create COB-based custom applications)
– Modularity & reusability  Faster, consistent application development
While originally intended for integrating engineering design and analysis models, COBs
may have broader usage wherever relation-intensive models are present.
3 Bibliography
These documents are included with this RFI response zip file and are prefixed with the
indicated numbers.
1. Wilson MW, Peak RS, Fulton RE (June, 2001) Enhancing Engineering Design and
Analysis Interoperability - Part 1: Constrained Objects. First MIT Conference
Computational Fluid and Structural Mechanics (CFSM), Boston.
http://eislab.gatech.edu/pubs/conferences/2001-mit-cfsm-1-wilson-cobs/
Illustrates the primary capabilities of constrained object (COBs) [Wilson, 2000] via basic spring
system examples.
3
2. Peak RS and Wilson MW (June, 2001) Enhancing Engineering Design and Analysis
Interoperability - Part 2: A High Diversity Example. First MIT Conference
Computational Fluid and Structural Mechanics (CFSM), Boston.
http://eislab.gatech.edu/pubs/conferences/2001-mit-cfsm-2-peak-xai-example/
Describes the main concepts of the multi-representation architecture (MRA) via a basic flap link
example. Uses COBs (described in Part 1) as the basis for CAD-CAE interoperability in the MRA.
3. Peak RS (Oct. 9, 2001) Techniques and Tools for Product-Specific Analysis
Templates: Towards Enhanced CAD-CAE Interoperability for Simulation-Based
Design and Related Topics. Invited Seminar. NIST. Gaithersburg, Maryland USA.
http://eislab.gatech.edu/pubs/seminars-etc/2001-10-nist-peak/
Overviews COBs and the MRA analysis template methodology with several tutorial and industrial
examples. This seminar is effectively a subset of the below referenced short course.
4. Peak RS (July 6, 2001) XAI Central: Applications, Techniques, and Tools.
http://eislab.gatech.edu/research/XAI_Central.doc
Contains summaries and an index to the main techniques, software tools, and application domains in
our work on X-analysis integration (XAI). It includes material on COBs and their usage in an analysis
template methodology. Here X represents product life cycle stages like design, manufacture, and
maintenance. The index includes pointers to other overviews and short course slides. An update of this
document is anticipated in the July 2002 timeframe.
The work is continuing in similar sponsored projects and as part of the Engineering
Framework Interest Group (a consortium-based effort): http://eislab.gatech.edu/efwig/
4 Submitter Biosketch
Russell S. Peak is a Senior Researcher in the Manufacturing Research Center at the Georgia Institute of
Technology. He received all his degrees from Georgia Tech in the School of Mechanical Engineering. His
industrial experience includes business telephone design at AT&T Bell Laboratories and analysis
integration investigation as a Visiting Researcher at the Hitachi Mechanical Engineering Research
Laboratory in Japan.
Dr. Peak is the lead developer of constrained objects (COBs), the multi-representation architecture
(MRA) for CAD-CAE interoperability, and context-based analysis models (CBAMs) - a knowledge pattern
that explicitly captures design-analysis associativity using object and constraint graph techniques. His
research specialty is analysis integration for simulation-based design (SBD), with applications including
electronic packaging and structural analysis.
He has authored and co-authored a variety of publications, holds several U. S. patents, and is a member
of ASME, IEEE, and the U. S. Association of Computational Mechanics. He serves on the Technical
Advisory Committee of PDES Inc., an international consortium furthering the development and usage of
engineering information technology standards.
4