A Survey of Engineering Framework Technology with Applications to Computer Aided Engineering

Russell S. Peak, Visiting Researcher {1}
Mechanical Engineering Research Laboratory
Hitachi Ltd., Tsuchiura, Japan

6th Int'l. Conference, Flexible Automation & Intelligent Manufacturing (FAIM 96), May 13-15, 1996, Atlanta, pp 47-56.

Abstract

An HTML version of this survey is available online that links to associated documents on the World Wide Web. Thus, as a secondary effect, this paper demonstrates how the Web can be used as a research tool.

Keywords

Contents

1 Introduction

2 Representative Research & Development Efforts

2.1 Electronic Commerce Resource Centers (ECRCs)

2.2 CAD Framework Initiative (CFI)

2.3 STEP

2.4 Enterprise Integration Technologies (EIT)

2.5 ARPA Knowledge Sharing Effort (KSE), et al.

2.6 Concurrent Engineering Research Center (CERC)

2.7 CALS Technology Center

2.8 Object Management Group (OMG)

2.9 National Industrial Information Infrastructure Protocols (NIIIP)

3 Recommendations & Potential Benefits

4 Conclusion

Acknowledgments

References

Appendix A - A STEP Sampler

Endnotes

1 Introduction {2}

This paper overviews a trend that is likely to broadly impact computer aided engineering (CAE) in the near future. Rather than surveying traditional computational techniques, it focuses on engineering framework technology as a means to incorporate such techniques more effectively into the product development process.

Consider the following scenarios:

1. You have developed a specialized CAE system, targeted at both internal and external customers, based on a finite element analysis (FEA) system from a particular vendor. Some of these customers already have large investments in general purpose FEA software from other vendors. Some customers tell you, "I want to use your system, but I don't want to buy and learn yet another FEA program. Can I use Vendor A's FEA program with your system instead?" With framework technology, the answer would more likely be "yes".
2. You are developing a new CAE package that needs extensive geometric modeling capabilities. Several geometric modelers already are available that will meet your needs. But how can you connect your new CAE program with these tools? Given framework technology, a straightforward solution would be more probable.
3. A certain analysis problem needs to be solved using a sequence of CAE tools. However, data must be exchanged between different versions of these tools which reside on a variety of computers. Because you will perform this type of analysis many times, you are wondering to what degree you can automate such solution processes (i.e., the sequencing of tools and transfer of data). If a framework is used, a flexible, highly automated process may be possible.

Much of the CAE effort to date has focused on the numerical computational aspects of engineering problems and the creation of individual tools. However, as the above examples indicate, applications that require the information-intensive cooperation of these tools are becoming increasingly apparent [Genesereth and Ketchpel, 1994]. The term engineering frameworks (FWs) is used here as the technology that enables such cooperation. The CAD Framework Initiative (CFI) defines a framework as:

"... an infrastructure of software and interfaces that enable the integration and interoperability of [engineering] systems" - CFI FAQ

Recent concepts such as concurrent engineering, collaborative engineering, virtual enterprises and agile manufacturing are the primary drivers behind much of the research and development in engineering frameworks. In the extreme, the goal of such trends is to enable the efficient development of products by geographically distributed, multi-company teams using diverse computer tools. However, the techniques used to solve this more general problem can potentially benefit even small groups within the same company.

From another view, engineering frameworks are part of the answer to the question "How do we connect people, their models, and their tools?". Olsen [1994] poses this question and overviews issues and current approaches. His proposed answer, summarized in the following figure, could be viewed as a kind of agent-based engineering framework [Huyn et al., 1993].

Figure 1. Collaboration as agent interaction [Olsen, 1994]

Given Figure 1, engineering FW technology can roughly be divided into the following areas:

• Tool Communication - how to provide inputs to a tool and request outputs. Topics include interoperability, tool encapsulation, wrappers, and application programming interfaces (APIs).
• Information Representation - how to represent models in a computer-processable manner that can be shared among tools. Related areas are product modeling and knowledge representations like objects and constraints.
• Control - how to manage higher-level issues such as sequencing of multiple tools, coordination of multiple users, and consistency of shared information. Techniques include agents and knowledge-based management systems.

2 Representative Research & Development Efforts

This section highlights representative efforts of varying sizes as starting points to explore this field. These organizations typically publish overviews, technical documents, and public domain software via Web servers to help one study and utilize their techniques. Some of the below descriptions are directly based on associated Web pages.

2.1 Electronic Commerce Resource Centers (ECRCs)

The U. S. Congress recently initiated the not-for-profit ECRC program under the direction of ARPA (Advanced Research Projects Agency). The program is a nationwide network of 11 regional ECRCs coordinated by one national office. The main purpose of the ECRCs is to help companies implement electronic commerce and related technologies (including engineering frameworks, design-analysis integration, and STEP). A national Technology Hub reviews and develops needed technologies, and makes recommendations to the regional ECRCs.

The existence of this initiative illustrates the important role that information technology is believed to have in keeping companies competitive. It also emphasizes how information technologies like engineering frameworks are needed to help CAE fit better into a company's overall business purpose.

Examples - ECRC News [ECRC, 1995] reports a wide variety of industrial examples including manufacturing (Harley-Davidson) conversion of existing paper drawings and documents (a railroad equipment manufacturer, and the U. S. Air Force), and better CAE models (Trintex, a rubber tire manufacturer).

2.2 CAD Framework Initiative (CFI)

CFI is developing public specifications which CAD/CAE vendors can use to create interoperable, engineering framework-based tools:

• Design Representation Programming Interface - basically product models for electrical design.
• Computing Environment Services
  • Scheme-Based Extension Language
  • Base Networking Services Guidelines
  • Base System Services Guidelines
  • Base User Interface Guidelines
  • Error Handling Programming Interface
• Message Dictionary Specification
• Tool Encapsulation Specification - defines a standard way to wrap tools (including existing ones) so they can be used in a framework.
• Tool Manager
• Task & Session Model

Combined, the above is similar to the Open Architecture in SDRC I-DEAS and the Falcon Framework by Mentor Graphics; however, CFI standards are public and designed for wide-spread industrial use rather than being vendor-specific, quasi-proprietary interfaces. One aim is that vendors like SDRC and Mentor will adapt their tools to support CFI standards, thus enabling them to interoperate in a much broader framework.

Examples - The Web server at CFI has information about industrial examples to date, including an online interoperability demonstration.

Software - At present CFI sells a CFI DR Toolkit (supporting the Design Representation specification for electrical design). Toolkits for other specifications are planned.

2.3 STEP

STEP (Standard for the Exchange of Product Model Data) primarily plays a role in Information Representation - particularly for product models, which can form the basis for CAE models. STEP Part 104 addresses finite element (FE) meshes, and other STEP efforts exist related to engineering analysis.

Examples, Software, etc. - Appendix A overviews STEP and provides examples of industrial usage, available software, and references.

2.4 Enterprise Integration Technologies (EIT)

EIT is an R&D and consulting company specializing in information technology for electronic commerce, collaborative engineering, and agile manufacturing. Their past and present projects are often agent-based and include PACT [Cutkosky, et al., 1993], the ARPA Knowledge Sharing Effort ( KSE ) and ARPA Manufacturing Automation and Design Engineering ( MADE ). They collaborate with companies like General Motors, Lockheed, Martin Marietta, and Texas Instruments, as well as with universities and industry consortium like NIIIP.

Their server has publicly available information such as project descriptions, demonstrations, technical information, and software. They are the developers of SHTTP (secure hypertext transfer protocol), a more secure version of HTTP that can be used for Tool Communication over the Internet when proprietary issues are involved.

2.5 ARPA Knowledge Sharing Effort (KSE), et al.

Related projects are SHARE (part of MADE) and SHADE involving EIT, Lockheed, Xerox, and others. SHADE reports the development of ontologies (information models) of engineering entities ranging from mathematics to physical laws to elevator components. They emphasize collaborative CAE; however, thus far they apparently have not dealt with problems that require discretized representations like finite element models (FEMs).

Their servers offer public domain tools and many online papers, including Olsen's Concur, his overview of agents, and their engineering applications [1994]. Genesereth and Ketchpel [1994] provide an overview of the agent-based framework theme that is common to much of KSE work.

Examples - PACT [Cutkosky, et al., 1993] demonstrates the multidisciplinary design of a robotic manipulator, including dynamics, controls, and electronics. Khedro, et al. [1993] show how agent-based frameworks can be applied to facility engineering (integrated structural design, architectural design, and project planning for buildings). The Concur pages highlight examples that use agents for Mathematica, MathCAD, and I-DEAS.

Software - Tools for creating and manipulating KIF, ACL, and KQML are available, as well as a series of ontologies (written in KIF). SHADE offers the agents mentioned above.

2.6 Concurrent Engineering Research Center (CERC)

The Concurrent Engineering Research Center at West Virginia University is part of the DICE initiative begun in 1988. Their primary mission is "the advancement of enabling technologies for the collaborative enterprise." CERC concentrates on the development of generic computer technologies to help manufacturing organizations become collaborative enterprises. Validation of these generic approaches has been carried out using engineering applications (e.g., aircraft engine turbine blade design at General Electric and electronic system design at Westinghouse), and has continued in other arenas as well (e.g., real-time patient treatment for the National Library of Medicine). CERC's Web server contains Concurrent Engineering Research in Review, collections of concurrent engineering abstracts and references, and an archive of their technical reports.

Software - The purpose of CERC's Information Sharing System (ISS) is disseminating heterogeneous information in a heterogeneous computing environment. In the Tool Communication area, this system uses the HTTP protocol to support client-end interoperability and OMG CORBA specifications for server-end interoperability. Web*, a major component of ISS, is a tool for publishing dynamic information.

2.7 CALS Technology Center

Other Examples - Warner Robbins Air Force Base has been a major sponsor for several years with an emphasis on improved geometric data conversion via IGES. Recent additions to the Center include the Atlanta Regional ECRC and a program funded by the FAA to re-engineer aircraft-related safety inspection processes.

The ARPA-sponsored TIGER project is developing techniques to enable electronic commerce and advanced collaborative engineering between DoD prime contractors and their subcontractors. This work (being carried out jointly between Arthur D. Little, Inc., Boeing, Georgia Tech, Holaday Circuits, ITI, and SCRA) emphasizes the use of STEP-related techniques and utilizes a slice of the printed circuit assembly life cycle as a demonstration vehicle (including thermomechanical analysis).

2.8 Object Management Group (OMG)

Thus far, OMG has developed the following guidelines and specifications to achieve the above goal:

• The Object Management Architecture Guide contains the central design guideline (the Reference Model) used by OMG to create a distributed object computing environment (Figure 2).

  

  Figure 2. The Object Management Architecture [OMG]

• The Common Object Request Broker: Architecture and Specification (CORBA) defines the mechanisms by which objects make requests and receive responses. The Object Request Broker is "the cornerstone of OMG's architecture," providing interoperability between applications on different machines in heterogeneous distributed environments.
• CORBAservices include the basic functionality needed to create and maintain objects in a distributed computing environment.
• CORBAfacilities are a collection of higher level services "broadly applicable to many applications or high value capabilities for specific domains or vertical markets."

Examples - Engineering applications of OMG's CORBA include an NIIIP demo noted below and the CERC ISS tool mentioned above.

2.9 National Industrial Information Infrastructure Protocols Consortium (NIIIP)

For example, their Reference Architecture (Figure 3) proposes a combined use of existing protocols from the Internet, OMG, STEP, and CFI efforts. Their document defines the four key technologies in their reference architecture, and discusses existing and needed solutions for each. Roughly, technologies #1 and #2 are related to Tool Communication, while #3 corresponds to Information Representation, and #4 to Control.

Figure 3. NIIIP Reference Architecture [NIIIP, 1995]

The NIIIP document can be used as a mini handbook for collaborative computing and engineering frameworks in general. Besides defining the above architectural vision, it indirectly overviews the state-of-the-art and contains a glossary of terms and acronyms. Overall, it provides one view of how the technological pieces highlighted in the above R&D efforts can be brought together to "connect people, their models, and their tools."

Examples - The NIIIP approach appears promising based on a demonstration at a recent STEP meeting. They demonstrated collaborative Web access to axle information (including geometry and assembly) between CAD tools like AutoCAD and ProEngineer using STEP AP 203, OMG CORBA, HTTP, and HTML protocols.

3 Recommendations & Potential Benefits

Suggested Approach

If a company is interested in learning more about framework technology, one approach is to evaluate the NIIIP Reference Architecture [NIIIP, 1995] and try out its recommendations where appropriate. As NIIIP membership includes a diverse group of significant players in this area (e.g., CFI, IBM, EIT, NIST, and STEP vendors), their approach is likely to be acceptable to a wide industrial audience.

A company could begin by reproducing a similar prototype framework internally and demonstrating it for a CAE application. This exercise would help a company become more familiar with the key technologies and better understand the benefits as well as challenges. Some of the software components are available in the public domain (e.g., Internet protocol daemons and STEP parsers).

Though agent technology has apparently not received as much industrial exposure as the above components, interested companies may do well to explore it also. The software and demonstrations mentioned from KSE, et al. provide one starting point.

Finally, organizations desiring more immediate results may consider existing product data managers (PDMs) available from a variety of vendors (e.g., Computervision, IBM, Intergraph, SDRC, and Sherpa). Such systems typically offer the advantages of robustness and fielded techniques, but lack in the more advanced capabilities identified above.

Possible Applications

• Creation of specialized CAE tools based on general purpose analysis tools.
• Design-analysis integration, including wrapping of existing FEA tools for highly automated use [Peak, et al., 1995] .
• Interaction of global/local or micro/macro models based on a common higher level description (e.g., via CIF, a kind of product model for LSI packages [Shepherd, et al., 1994]).
• Creation of general CAE tools from functional parts (e.g., using existing geometric modeling tools).
• Development of one's own CAE tools such that can be used by other applications (i.e., as tools that fit into a broader framework).

4 Conclusion

Since engineering framework technology is a rapidly growing field, it is easy to get lost in all the information available. It is helpful to keep in mind Olsen's basic question (how to connect people, their models, and their tools), and to consider solutions to the three basic areas identified: Tool Communication, Information Representation, and Control. One must be watchful for the general techniques that exist amidst the diverse applications driving this technology - techniques that can be applied to CAE and benefit a company. The R&D efforts highlighted in this paper can be used as starting points to learn more about such techniques.

Acknowledgments

The views expressed in this paper are those of the author and may not necessarily reflect the views of Hitachi, Ltd. or his current employer.

References

Cutkosky, M. R., Engelmore, R. S., Fikes, R. E., Genesereth, M. R., Mark, W. S., Tenenbaum, J. M., Weber, J. C. (Jan. 1993) PACT: An Experiment in Integrating Concurrent Engineering Systems. Computer, 28-37.

ECRC, Winter-Spring 1995, ECRC News, Electronic Commerce Resource Center, Concurrent Technologies Corp., Johnstown PA, USA.

Genesereth, M. R. and Ketchpel, S. P., July 1994, Software Agents, Comm. ACM, 37, 7, pp. 48-53, 147.

Huyn, P. N., Genesereth, M. R.; Letsinger, R., Jan. 1993, Automated Concurrent Engineering in Designworld, Computer, 74-76.

Khedro, T., Genesereth, M. R., Teicholz, P. M., 1993, Agent-Based Framework for Integrated Facility Engineering, Engineering with Computers, 9, 94-107.

NIIIP, Jan. 1995, NIIIP Reference Architecture: Concepts and Guidelines , NIIIP Consortium , NTR95-01.

Olsen, G. R., Dec. 12 1994, Concur Overview, WWW, http://piano.stanford.edu/concur/tutorial/tutor1.html.

Peak, R. S., Fulton, R. E., Nishigaki, I., Okamoto, N., March 1995, Integrating Engineering Design and Analysis Using a Multi-Representation Approach, in review.

Shephard, M. S.; Sham, T.-L.; Song, L.-Y., Garimella, R.; Tiersten, H. F.; Lwo, B. J.; Le Coz, Y. L.; Iverson, R. B.; Fish, J. (1994) Global/Local Heat Conduction and Thermomechanical Analyses of Multichip Modules. Proc. Intl. Symp. on Highly Advanced Computing - Computational Mechanics for Electronic Devices & Components (ISAC '94), Chiba, Japan, 123-134.

Stephens, E. R. (1993) LEGEND: Laboratory Environment for the Generation Evaluation, and Navigation of Design. Doctoral Thesis, Georgia Institute of Technology, Atlanta.

Appendix A - A STEP Sampler

Sampler Contents {3}

• Suggested Starting Points
• Internet Resources
• References
• Software Providers
• Industrial Usage Examples

Endnotes

{1} Author affiliation as of January, 1996: Russell S. Peak, Research Engineer, Georgia Institute of Technology, CALS Technology Center, peak@cad.gatech.edu.

{2} Because of the survey nature of this paper, it is also available as a World Wide Web document (with a table of contents) at: http://eislab.gatech.edu/pubs/conferences/faim96/

All underlined phrases are hypertext links to other documents containing more information. Thus, the best way to read this paper is from a Web browser. Select a link and the related document will be displayed.

As many of the links are to Web pages on computers maintained by other organizations, the author cannot guarantee that they will always be available or that they will continue to contain the same information described here. Thus, Postscript files of key Web pages (as of the time of this writing) are available for archival purposes.

{3} Because the STEP Sampler is long and contains numerous hyperlinks, the full contents are available in a Web document at: http://eislab.gatech.edu/pubs/web/step-sampler/