Evaluation of Computer Aided Software as a Space Analysis ...

Evaluation of Computer Aided Software as a Space Analysis Tool for Outfit Unit Design and Planning

Prepared by:

Patrick D. Cahill, Principal lnvestigator Senior Enpeering Research Associate

University of Michigan transportation Research Institute (UMTRI) Marine Systems Division

for

Designers and Planners

in Support of

Generic Build Strategy mdtenn Sealift Project

July 28, 1995

Evaluation of Computer Aided Software as a Space Analysis Tool for Outfit Unit Design and Planning

Prepared by:

Patrick D. Cahill, Principal Investigator Senior Engineering Research Associate

University of Michigan transportation Research Institute (UMTFU) Marine Systems Division

tor

Designers and Planners

in Support of

Generic Build Strategy Mdterm Sealift Project

July 28, 1995

Table of Contents

........................................................... EXECUTIVE SUMMARY ....................................................... 1

.................................................................. CURRENT RESEARCH AND DEVELOPMENT FOCUS 2

........................................................................................... CURRENTLY AVAILABLE SOFTWARE 4

.................................................................................. CONCLUSIONS AND RECOMMENDATIONS 4

Executive Summary During July, 1995. the Uni~enity of Gchgan Transportation Research Institute. Manne Systems

Division. conducted an extensive literature search to i d e n e and evaluate software currently

commercially available that would support space analysis for outfit h i t design and planning. The long

term objective is to ident* and evaluate specific software packages or combinations of packages that can

be employed in the development of build strategies and the early stage design of sfupboard spaces,

particularly machnery spaces.

The entire University of Michigan library system was searched for pertinent literature about

shipbuilding and other industries where spacial planning is a critical design element. Specific resources

utilized included the NSRP documentation center, the UMTRI library, the Kresge Business Library, the

Dow Engineering Libran and the School of Art and Architecture l i b q .

The overall conclusion of the research effort is that there is not a currently available commercial

sohvare package that will automate the spacial arrangements problem. However, there is current

research that has resulted in working prototypes. and there are available commercial packages that. in

combination with each other, will offer a riable solution.

The Archtecture and Civil Engineering community has made the most progress in the development

of a specific tool that appears to address the issues of concern. ABACUS (Architectural Case Based

Design System) has been he loped into a working prototype at the department of Architecture.

Eidenossische Techrusche Hochschule. Zurich, Switzerland It is a Case Based Design system

implemented in LISP and AutoLISP and using Autocad as a graphic engine.

Other tools of interest include Case Based Design and Case Based Reasoning soha re protoopes,

Artificial Intelligence and Knowledge Based systems without graphical engines, and parametric design

packages that maintain topographical relationships.

Objective

Tlus report represents the findtng of Tasks 1.1 and 1.2 of a three part task statement. The smc

task statements are:

Task 1.1 Literature search to obtain information on specific software currently commercially available. Task 1.2 Comparative assessment of identified software with regard to state-of-the-art capabilities described in CADICAE literature. Task 2 Demonstrations of selected software at vendor sites to assess capabilities to perform space arrangements. analysis and optimization for ERAM and to interface to ERAM product modeling efforts. Task 3 1 . Pro~ide a report detailing the findtngs and pro\iding recommendations. Task 3.2 Provide administrative support for report preparation and task management.

I t quickly became apparent during the literature search that there are no commercially available

packages that meet the requirements of the ERAM and GBS portions of the Midterm Sealift Program. As

a result. the focus of the search was shifted to iden@ research efforts that may result in the appropriate

s o h a r e packages. as well as sofhvare that may be used as a component of an Integrated. Automated.

Intelligent Arrangements (IA): program

Current Research and Development Focus Current sofhvare research and development is focused in three areas that, if combined and integrated

have great potential to yield the desired ( 1 ~ ) ~ system. These three areas are product modeling knowledge

based systems and parametric design. The only area in which shipbuilding specific research is occurring

and ongoing is in product modeling. Knowledge based systems development is prominent in the civil

engineering and architectural design fields, and parametric design efforts are heavily focused on

supporting the mechanical design and manufacturing field An additional field in whch these three areas

are being combined is I electronics, specifically in the design and manufacture of Ver?; Large Scale

Integrated (VLSI) circuits. The literature pertaining to VLSI design is extensive, but may be technically

beyond the scope of what can be readily implemented in the shiphlding industry.

Product Modeling

Much of the current research in shipbudding that may eventually result in a space planning Tstem is

in the area of product modeling. Research has been funded to implement a universal STEP translator

based on the work by NIDDESC. A number of international organizations, both in Europe and in Japan,

are focusing research efforts on the data structure and format of product model database that is

independent of design sohare . Strong efforts are undenvay in both Japan and Europe to integrate

product and process models. \.---.

A sigxuficant part of the focus of the product modeling efforts has been a consistent leaning towards

the object oriented programming paradigm. As long as product and process model attributes can be

defined as objects uithin a software system they are more likely to be universally available for

representation and implementation in both graphcs engines and knowledge bases.

Knowledge Based Systems

Research uithin the civil engneering and architecture industry on knowledge based systems holds the

most promise for near term implementation of an (IA)' design system. Knowledge based systems cover a

number of different paradigms. and can be considered a subset of Artificial Intelligence or Expert

Systems. They are rule based. ofien implemented in a tabular or spreadsheet format, and have sigruficant

amounts of soha re code, generally in either C or LISP, that make up the intelligence engine.

Knowledge based systems research has resulted in a large number of prototypes, particularly in civil

engneering and archtectural design. that use existing designs as the kno~iledge base and integrate them

\\ith rules and adcbtional w r input to generate new designs. Some of the systems are purely specification

based some have graphics engines attached to the A1 backplane.

Case Based Design, or CBD. is one of the most sigruficant knowledge based paradgms. Case Based

Design systems have the follouing properties:

- A CBD system does not require a complete domain model but can produce complete and comples designs based on a relatively small knowledge base. - Design starts from complete cases, implicitly achieting tradeoffs between several functions. Tlus avoids the problem of multi-criteria optimization. - Applying the design history of existing cases can make design problem solving more efficient. - Using cases as the source of kno\vledge allows learning by storing new cases.

The specific properties of case based systems make them ven appealing for use in the shipbuilding

industq. Each subsequent design becomes pan of the knowledge base, and learning obtained in

construction and operation of a specific design can be applied to enhance the kno~vledge base. It is likely

that the soha re is object oriented, or can readily be adapted to object orientation which will complement

the product and process model outputs.

Parametric Modeling Systems

Parametric modeling systems represent the leading edge of design sofhvare. Research and

development is continuing to enable better, smarter and faster parametric design, and a number of

outstandng packages are a l r w available. Parametric modeling allows a designer to quickly define part

geometry, and changes in a specific portion of the geometry will result in a "ripple" effect in whch all

related parts will change in accordance with the design spedcations. Leading edge parametric design

maintains topographical relationships in assemblies, and changes in the parameters of part of an assembly

result in the necessary changes withn the rest of the assembly. Much of the code developed to suppon

parametric design follows the object oriented paradgm, whch supports the theory that the integration of

parametric design. knowledge bases and product models may be well within the reach of near term

technology.

Currently Available Software '

Two of the three focus areas have s o h a r e alreae available, although some is in the prOtOhpe stage

and the commercial availability is unknown. Product and process modeling efforts will move fomard

regardless of the implementation of an (IA)' package. The product and process model s i l l represent data

rather than a specific software package. Several of the abstracts included in Appends B address the

product and process modeling efforts.

There are a number of Knowledge Based systems currently available as protot~pes; however, the

commercial availability and the levels of development are unknown. Case based design systems include

ARCHIE. CADSYN, CADET. ABACUS, CASE, GOAL and SEED. A brief descripuon of these systems

is prolided in Appendis B. An additional abstract on Allegro Common LISP is included in Appendix C

as a potential programming tool to support knowledge based system dmelopment.

Parametric Design systems of note are currently available from Intergraph; including a hvo

dimensional spacial planning system called project layout. Autodesk has released a new version of

designer which provides parametric modeling that maintains topographical relationships of assemblies.

An abstract on PASS by Data One is included in Appends C as one of the few software tools that stood

out during a roiew of over 1000 CADICAM sofbvare abstracts.

Conclusions and Recommendations

The overall conclusion of the project primary investigator is that there are no soft\tare packages

currently commercially available that will meet the needs of the ERAM and GBS projects in terms of

providing an (IA)' tool, with the possible escepuon of ABACUS. There appears to be sigxuficant promise

in the concept of integrating a knowledge based system with a parametric design package. Even if the

integration is not seamless, i.e. the design engineer runs the knowledge based system to achieve a set of

parameters that are then used as input to the parametric design software, the end result may be a more

intelligent and semi-automated design process. The core of the system is the knowledge base, whch will

allow a steady development of design starting points, and learning form construction and operational

experiences.

Appendix A: Bibliography of U of M Library Research Results

GBS Arrangenients Study Literature Search Bibliography: Universily of Michigan Library Systeni

Page I

Year

1995

1995

1994

1994

1994

1994

1994

I994

1994 1994 1994

1993

I993

I993

Publisher

Arlificial lnlelligence for Engineering Dcsign, analysis and manufacluring (AIEDAM) Vol9 Nbr 3 pp161-182 London ; New York : Chapman & Hall New York : American Society of Mcchanical Engineers

Amsterdam ; New York : Elsevier Boston : Kluwer Academic Publishers Dordrecht ; Boston : Kluwcr Acadeniic ~ublistlers Englewood ClilTs. N.J. : Prentice Hall New York. N.Y. : An~erican Society of Mechanical Engineers

New York : ASME Press New York : ASME Press London ; New York : Springer- Verlag New York. N.Y. : American Society of Mechanical Engineers

Rivcr Edge, N.J. : World Scientific Berlin : New York : Springer- Verlag

Title

A franiework for case-based reasoning in engineering design

Advanced CADICAM systems : state-of-the-art and future trendsin fcature technology Advances in design automation, 1994 : presented at the I994ASME Design Technical Conferences, 20th Design AutomationConfierence, Minneapolis, Minnesota, September 1 1-14, 1994 Advances in feature based manufacturing

Analog device-level layout automation

Arlificiai intelligence in design '94

Artificial intelligence in optimal design and manufacturing Concurrent product design

Intelligent systems in design and nlanufacluring Intelligent systems in design and manufacturing Practical knowledge-based systems in conceptual &sign Advances in design automation. 1993 : presented at the 1993ASME Design Technical Conferences, 19th Dcsign AutomalionConference. Albuquerque, New Mexico, September 19-22, 1993 Algorithmic aspccls of VLSl layout

CAD geometry data exchange using STEP : realisation ofinlcrface processors

Author

H. Shiva Kuniar and C.S. Krishnamoorthy

the International Federation for Information Processing (IFIP). Design Automation Conference (ASME) (20th : I994 :Minneapolis. Minnesota)

edited by Jami J.Shah. Martti Mantyla, Dana S. Nau

Zuomin Dong, editor

I994 lnternationalMechanical Eng~neering Congress and Exposition, Chicago.lllinois, November 6- 1 1. I 9 94

Miles, John

Design Autonlalion Conference (ASME) (19th : 1993 :Albuquerque. N.M.)

Locati on

Eng

ENG

ENG

ENG

ENG

ENG

ENG

ENG

ENG ENG ENG

ENG

ENG

ENG

Call#

TS155.6.A35

TA174.D16511

TSI83.3.A381

TK7874.AS-W I

TA174.A7932 1

T S 15-5.6.A77 1

TA174.C6611

TAI 74. I4688 1 TA 174.14688 1 TA345.MS4 1

TA174.D465 1 1

TK7874.AJI91

TA174.C2541

Knowledge aided design Opt imization-based computer-aided model ling and design :proceedings of the first working conference of the new IFIPTC 7.6 Working Group,The Hague, Thc Netherlands. 199 I Title:6lh lnternational Conference on CADICAM. Robotics. andFaclories of the Future 199 I : held at South BankPolytechnic. 103 Borough Road, London

GBS Arrangcmcnts Study Literature Scarch Bibliography: University of Michigan Library Svstcnl

International Confcrcncc on CADICAM, Robotics. and Factoriesof the Future (6th : 199 1 :

Verlag

Year

1993

1993

I993

1993

1993

1993

1993. 1993

1993

I992

1992 1992

London ; South Bank Prcss. I

Publisher

Ncw York : M. Dckkcr

Ncw York : Wilcy

Englcwood CliKs. N.J. : Prcntice Hall Englewood CliKs, N.J. : Prcnlice Hall Amsterdam ; New York : Elscvicr Boston : Kluwer Acadeniic Publishcrs New York : McGraw-Hill AmcricanSociety of Mechanical Engincers.

London ; New York : McGraw- Hill

Boston : Academic Press New York : W i l y

Title

CADICAMICAE systems : justification. in~plcmentation.productivity measurement Computer aided logical design with emphasis on VLSl 4th ed Computer-aided design and manufacturing

Coniputcr-aided design and manufacturing

Concurrcnt engineering : methodology and applications Design automation for timing-driven layout synthesis Design to reduce technical risk Intelligent concurrent design : fundamcnlals, mcthodology,modeling, and practice

Optiniizing engineering designs

Algorithmic and knowledge based CAD for VLSl Published Artificial intelligence in engineering design Intelligent design and manulacluring

SE I O M U.K.. 19th-22nd August, 199 I I South Bank Polytechnic. London) I I :Product modeling for computer-aidcd design and I lFlP TC5lWG5.2 Working I Amsterdam ; Ncw York : North- 1 199 1

Author

Coticchia. Mark E.Titlc

Hill, Fredrick J

:Anlirouche. Farid ML

Amirouche, Farid M. L

Sapatnckar, Sachin S.,

American Society of Mechanical Engineers. Winter Mecting (1993: New Orleans, La.) Krottmaier, J.,

Institution olElectrical Engineers

on ENG

n~anuracturing :&ctcd and expandcd p k r s from the lFlP TC51WG5.2 WorkingConfcrence on

ENG

ENG

Conrercnce on Gconlclric Modcling forProduct Enginecring ( 1990 :

ENG

Holland

ENG

ENG

ENG ENG

ENG

ENG

ENG ENG ENG ENG

ENG

ENG

GBS Arrangements Study Literature Search Bibliography: University of Michigan Library Systcnl

Page 3

Year

I991

1991

1991

1991

199 1

1991 199 1

1991 199 I

1990 I990

I990

1990

Publisher

Berlin ; New York : Springer- Vcrlag New York. N.Y. : American Society of Mechanical Enginecrs

Berlin ; New York : Springer- Verlag Berlin ; New York : Springer- Verlag

Amsterdam ; New York : North- Holland

New York : McGraw-Hill Berlin ; New York : Springer- Verlag London : Chapman and Hall Englwood Clifis, NJ : Prentice Hall Greenwich. Conn. : JAl Press Winter Annual Meeting of the AmcricanSociety of Mechanical Engincers. Dallas. Te.uas, Novcmbcr 25-30 Boston : Kluwer Acadenlic Publishers Amstcrdanl ; New York : North-

Title

Geometric Modcling for Product Engincering,Rensselaerville, U.S.A., 17-2 1 June 19'90 Advanced modelling for CADICAM systems

Advances in design automation, 199 I : presented at the 199 1 ASME Design Automation Conferences- 17th Design AutomationConference, Miami, FL , September 22-25, 199 1 An artificial intelligence approach to integrated circuilnwrplanning CADICAM. robotics, and factories of the future '90 : Sthlnternational Conference on CADICAM, Robotics, and Factoriesof the Future (CARS and FOFe90) proceedings Computer applications in production and engineering :proceedings of the Fourth International IFIP TC5 Conferenceon Computer Applications in Production and Engineering--Integration Aspects, CAPE '9 1, Bordeaux, France, 10- l2September 199 1 Computer-integrated design and manufacturing Engineering optimi7ation in design process :

proceedings oflhe international conference Mechatronics : electronics in products and processes Objcct3riented databases with applications to CASE. networks.and VLSI CAD Advances in computer-integrated manufacturing Advances in integrated product design and manufacturing

Autonlatic programming applied to VLSl CAD software : a caxstudy l ntelligent CAD; I I : proceedings of the IFlP TC

Author

Rcnsselaerville, N.Y .)

H. Grabowski, R.Andcrl, M.J. Pratt (Eds.). Design Automation Conference (ASME) (1 7th : 199 1 : Miami, FL

Jabri, M. A. (Manvan A.)

International Conference on CADICAM, Robotics, and Factoriesof the Future (5th : 1990 : Norfolk, Va.) International IFlP TC5 Conference on Computer Applications inProduction and Engineering: Integration Aspects (4th : 199 1 : Bordeaux, France) :Ekdworth, David D Karlsruhe Research Center

SclliN, Dorothy E

1FIP TC 51WG 5.2 Workshop on

Locati on

ENG

ENG

ENG

ENG

ENG

ENG ENG

ENG ENG

ENG ENG

ENG

ENG

Call#

TS155.6.A373 1

TA171.AID45199 I

TK7874.J28 1

TS155.6.158171

TS155.6.15861

TS 155.6.B-439 1 TA658.8.E54 1

TS 155.6.M42 QA76.9.D3026 1

TS155.6.A385 TS155.6.A386

QA76.6.S468 1

TA174.E871

GBS Arrangenients Study Literature Search Bibliography: University of Michigan Library System

Page 4

- Title

51WG 5.2Second Workshop on Intelligent CAD, Cambridge. UnitedKingdom. 19-22 Seplember. 1988 Introduction to analog VLSl design automation

Knowledge-bscd design systems

Advances in dcsign automation, 1989 : presented at the I989ASME Design Automation Conferences-- 15th Design AutomationConference. Montreal, Quebcc. Canada. September 17-2 1, 1989 CAD syste~iis in mechanical and production engineering CAD systems using A1 techniques : proceedings of the IFlP TCIO/WG 10.2 Working Conference on CAD Systems Using AITechniques, Tokyo, Japan, 6-7 June 1989 :Factory 2000 : integrating information and material flow Automated modeling for design

Conlpuler-aided systems engineering (CASE)

Design automation : automated fullcustom VLSI layout usingthe ULYSSES design environment Expert systems for engineering design Physical design automation of VLSI systems

An expcn systems approach to computer-aided dcsign ofmultivariable systems Computer-aidcd engineering for manufacture Engineering productivity through CADICAM Intelligenl and integrated manufacturing analysis andsynthesis

Author

Intelligent CAD (2nd : 1990 :Cambridge. England)

Design Automation Conference (ASME) (15th : 1989 : Montreal,Quebec

: Ingham, Peter

IFlP TC IOIWG 10.2 Working Conference on CAD Systems Using AlTechniques

International Conference on Factory 2000

Eisner, Howard

Bushnell. Michael L. (Michael Lce).

Pang. G. K. H. (Grantham K. H.)

Milncr, D. A. (Douglas A,) Chorafas. Dimitris N the Winter Annual Meeting of theAmerican Society of Mechanical

Publisher

Holland Elscvicr ScienccPub. Co

Boston : Kluwer Academic Publ ishers Reading, Mass. : Addison- Wesley New York. N.Y. : American Society of Mechanical Engineers

Oxford : Heinemann Ncwnes

Amsterdam ; New York : North- Holland

London : Institution of Electronic and Radio Engineers New York, N.Y. : American Society of Mechanical Engineers Englewood Cliffs, NJ : Prentice- Hall Boston : Academic Press

Boston : Academic Press Mcnlo Park. Calif. : BcnjaminlCumniings Pub. Co Berlin ; New York : Springer- Verlag New York : McGraw-Hill London ; Boston : Butterworths AmericanSociety of Mechanical Engineers.

Year

I 9 90

1990

1989

1989

1989

1988

1988

1988

1988

1988 1988

1987

1987 1987 1987

Locati on

ENG

ENG

ENG

ENG

ENG

ENG

ENG

ENG

ENG

ENG ENG

ENG

ENG ENG ENG

Call#

TK7874.15955 1

TA174.KS91

TA171.AlD45198 9

TS155.6.158

TA174.11331

TS155.6.1573 1 1

TA174.A92 11

TA168.E371

TAI 74.B871

TA174.E96l TK7874.P471

TJ2 13.P265 1

TS 155.6.MSS 1 TS155.6.CJ95 1 TS 155.6.156 1

Appendix B: Selected Periodical Abstracts and References

Case Based Design and Creatiiiv Gerhard Schnutt Department of Architecture. Eidenossische Technische Hochxhule, Zurich

Case Based Design, CBD. feature presening but not necessarily geometry presening. Combined with Case Based Reasoning CBR relates the present situation to the closest experience in memon and uses that esperience to solve a problem. Setferal case based design systems in use:

1994 Automation based creative design : research and peqxctwes

ARCHIE (Goel et al.. 199 1) CADSYN (Maher and Zhang, 1991) CADET (Sycara et al.. 1991)

Amsterdam ; New York : Elmier Science B.V..

Case based design systems have the following properties: - A CBD system does not require a complete domain model but can produce complete and complex designs based on a relatively small knowledge base - Design starts from complete cases. implicitly achieving tradeoffs between several functions. Ttus a~oids the problem of multicriteria optimization. - Appl!ing the design history of existing cases can make design problem solving more efficient. - Using cases as the source of knowledge allows learning by storing new cases.

ACABAS - Architectural Case Based Design System - protoQped in 1990. implemented in LISP and AutoLISP and uses Autocad as the graphc engine. Results are tested and kisualized on "Stalker" Iisualization software on Silicon Graphcs machines.

Funded t?. the Swiss Nationalfond under the program NFP 23: Artificial Intelligence and Robotics.

Refs :

Faltings. Boi, Gerhard Schnutt. Kefeng Hua and Shen-Guan Shih. 1991. "Case Based Representation of Architectural Design Kno~vledge" in Proceedings of DARPA Workshop on Case based reasoning. Washngton D.C. pp 307-3 16

Goel. A.K. and J.L. Kolodner. 1991. "Towards a case Based tool for aiding Conceptual Design Problem Solving" in Proceedings of DARPA Workshop on Case based reasoning. Washington D.C. pp 109-120

Advances in design automation. 199'1 : Design Automation Conference New York N.Y. : 199 1 presented at the 199 1 ASME Design (ASME) (1 7th : 199 1 : Mami, American Society Automation Conferences--17th Design FL of Mechanical AutomationConference. Miami. FL , Engineers September 22-25. 1991

Motion Planning in ~ l a n i CAD Systems Koichi Kondo and Koichi Ohtomi Mechanical Engineering laboratory Research and ke lomen t Center Toshiba Corporation Kawasaki. Kanagana. japan

"This paper discusses a general and efficient method for solving the motion planning problem defined as checlung the existence of a collision free path among knoun statioxxq obstacles. and also presents it's application to a plant CAD system. The basic approach taken in this method is to resuict the free space referred to in path searching and to avoid executing unnecessary collision detection's. The configuration space is equally quantized into cell s by placing a rectangular grid and two new search strategies whch enumerate restricted cells are introduced for realizing th s method. One is a local strategy which enumerates free space cells only along the boundary of the free space in the configuration space. Another is a global strategy which finds the outer boundary of the free space. This method has been actually implemented and has been applied to an esample in a nuclear power plant."

Rough esample of a plant layout using solid primitives and a method of breaking the CAD model into a series of small cells. each of which either contains a solid member or does not. Separation between elements is calculated and routing for installation is determined based on mo~ing a solid geometry pan through the arrangement.

References:

Kameyama. K., Kondo. K, and Ohtomi. K., 1990. An Intelligent and Interactive CAD Layout CAD system for Industrial Plant. . R o c d n g s ASME DTM'90 DE-Vol27. PP 33-38.

Lozano-Perez. T., 1983. "Spacial Planning: A Configuration Space Approach", IEEE Trans. Comput Vol (2-32. pp. 108-120.

, ~ n : J79 ),

.- and roach.

>-fer- me- raft

rland.

281. for

(Ed. ). -1lter

.XC-

terns.

Automation In Construction 1 (1995) 101-110

Software design of maintainable knowledge-based systems for building design .'

Tom Andersen * , Niels Vejrup Carlsen CABDL', Cornpurer Aided Building Design Unil, c / o Department of Bulldit~g Deslgn, Building 118. Tecl~n~cal L;nr~ ersln oJDe~lmori.

DK2800 L> ngb}, Denmark

Abstract

Identifying and establishing a basic structure for knowledge representation is one of the keys to successful design of knowledge-based computer systems. In Building Design and Construction, this initial knowledge structure can be achieved by utilizing a query driven approach to software engineering. As (user) queries reflect the user's demand for in/output, i t is natural to link the overall user dialogue with key elements in the knowledge base. Direct connections bewcen user screens and objects in the knowledge base support prototyping and testing the application during development. However, the price for pursuing this approach in its pure form can be high, as needs for later maintenance and augmentation of the system can be very hard to fulfil. T o overcome these problems, a strict user interface software separation strategy must be introduced at early stages of software design, and implemented as a global control module as independent of the knowledge processing as possible.

Keywords: bowledge-based; Query driven: Software design: User interface; Separation: hlaintainablc systems

1. Introduction

At the Computer Aided Building Design Unit (CABDU), our research efforts are directed towards the provison of an intelligent building design assistant (IBDA).

IBDA must be an integrated environment con- sisting of several computer tools working around a central model of the (building) artifact [7,12].

Discuss~on is open until December 1995 (please submit your discussion paper to the Editor on Architecture and Engineering. Y.E. Kalay).

Corresponding author. I Current address: Siemens AG, Otto-Hahn-Ring 6. 81739

Miinchen. Germany.

09!6.5805/95/S09.50 @ 1995 Elsevier Science B.V. All rights reserved SSD1 0 9 1 6 - 5 8 0 5 ( 9 4 ) 0 0 0 4 2 - 5

This system will employ a variety of information technologies spanning from logistics systems. traditional CAD systems and computational tools to knowledge-based design support systems. Some observations on the overall configuration of such an integrated environment are briefly presented in [3]. Issues like user dialogue and the diversity of design domains are outlined.

In this paper we focus on knowledge-based tools, and our aim here is to explore pertinent software engineering aspects for this category of computer applications.

In the following we will present and discuss the BYGSYS system which at this point is a stand-alone knowledge-based system supporting roof design. BYGSYS can be regarded as a tool

1 10 . T. Andersen, N. b'ejrup Carlsen / Airtomatron in Consrntction 4 f 19951 101 -110

possible links which must be reestablished according to the nature of the augmented function- ality.

This also covers the introduction of cltanged order o fke) , screen. The nature of the domain. the implications and rules etc. which can be derived by acquisition, cannot be neglected i f one wants to pursue a sensible consultation process. In the example presented above, the selection of roof covering (Block2 in control function) do impact level 1 to a certain degree, thus reducing the choices of preferable roof shapes (which is controlled by Blockl). This is simply augmented to the functional core by adding the relevant rules and functions. Test of the system has proven that it performs equally well in both scenarios, and that our distributed separation method formed a sound base for this augmentation of BYGSYS.

5. Conclusions

User (query) driven modelling and design does without doubt imply incremental prototyping. However, in the case of larger systems and/or complex knowledge domains, this is only feasible if care is taken to have optimal separation between the user interface and the functional core.

of agents in an integrated software environment (IBDA) supporting building design.

Acknowledgements

We acknowledge support from the Danish Technical Science Foundation (STVF). grant # 16-5243-1 0s. A special thanks to Professor Chris Hendrickson, Carnegie Mellon University, Pitts- burgh for providing valuable suggestions and comments.

References

[ I ] T. Andrrsen and A. Gaarslev. Building faults-prevent- ing future faults by utilizing past experience, In: V. Ire. land (Ed.). P r c , of ClBW F ' . 7. 1990.

[1] T. Andersen. BYGSYS technical report. Department of Construction Management, June 1992 (in Danrsh).

[3] E. Borchersen. Man-machine interaction and the intelligent design assistant, Proc. 4rh Int. Symp. on Syrrems Research, lnfomat~cs and Cybemerics. Baden-Baden. Germany, 1993.

[4] B.C. Bjsrk. Basic structure of a proposed building product model, CAD I., 21 (2) (1989).

[5] G. Cockton. A New Model for Separable lnteractlve Systems. Proceedings of INT'ERACT'87. Elsevirr Science Publishers. 1987.

[6] A. Dix et al., Human Compurer Inreracrion. Prentice-Hall. An initial prototype can easily be developed with- Englewood Cliffs, N J . (1993).

out separat ion but later revisions will most likely [7] S. Fenves et al.. Concurrent Computer-lntegrared Building Design. Prenticc-Hall. Englewood Cliffs. N.J. (1994).

be prohibitively complex. [8] R.A. Guedj and H.A. Tucker. ,\lerhodology in Computer So, user demands must be taken seriously when Graphics. North-Holland. Amsterdam (1979).

designing knowledge-based decision support systems. We do recommend user (query) driven knowledge acquisition and modelling but combined with an overall strategy for software separation as a minimum of effort.

I t should be pointed out, that our findings and related proposed guidelines are based on research in the spectrum of technical building design. These can be classified as engineering domains. wherein a reductive problem solving ap- roach can be amlied.

-19) R.A. Guedj et al., iClerhodology of Inrrractron, North- Holland. Amsterdam (1980).

[ lo] D. Hix and H.R. Hartson, Der~elopmg User Inrerfoces. Wiley, New York (1993).

[ l l ] Intellicorp, User's Manual, W P A . P C , 1991. [I21 J. Pohl, Computer-based cooperative agents. an emerging

paradigm, Proc. 41h Inf. Symp. on Sysrems Research. Infomarics and Cybcnurics, Baden-Badrn. Gennany. 1993.

1131 1. Sommerville, Software Engineering. 4th ed.. Addison. Wesley. Reading. Mass. (1992).

[I41 D.A. Watennan. A Guide to Expert Systems. Addison- Wesley. 1986. . .

If a domain satisfies these conditions, the query 1151 D.A. Young, The X Window System-Programmmg and

driven approach combined with a firm strategy of Applicarrons wrrh .Yt (OSF/MOTIF ed.). Prentice-Hall,, Engleupood Cliffs, N.J. (1990).

pursuing a high degree of independent user dia- [I61 S. kchariassen. Knowledge based computer systems in logue. is highly recommendable, and can form a engineering design. Master Thesis. CADBU. Tech. Univ. basic software engineering strategy for a number of Denmark. 1993 (in Danish).

ELSEVIER Autdrnatlon in Construction 3 (1994) 229-237

Knowledge-based design research at the key centre of design computing *

M.A. Rosenman *, J.S. Gero, M.L. Maher Key Centre of Des~gn Computing, Deporrment of Architectural and Design Science, Unrtersrr). of Sydney, Sydney, Ausrralio -

Abstract

This report outlines the current focus of the research at the Key Centre of Design Computing (KCDC) in the field of knowledge-based design and its future directions. The research is classified into seven main areas as follows: - Computer-Supported Synchronous Collaborative Design (CSSCD) - Design Creativity - Evolutionary Models of Design - Shape Emergence - Design Intent and Multiple Abstractions - Integration of Specific and General Knowledge - Concept Learning - Machine Learning

While each such area will be discussed separately, there are many interrelations between them. For example, the research on CSSCD includes the research into shape emergence, design intent and multiple abstractions and their integration in the one environment; design creativity includes research into shape emergence and evolutionary models; CSSCD, evolutionary models, design creativity, case-based design and design intent/multiple abstractions all deal with object-oriented or schema-based representations of design objects and classes, using the design prototype schema developed at the then Design Computing Unit (DCU). The work on memory organization for heterogenous design knowledge includes aspects of concept learning and machine learning. Evolutionary models are also seen as an approach to machine learning.

All the above areas of research are concerned with producing computable models of design processes based on integrating artificial intelligence approaches with CAD systems to provide better design support for designers using CAD systems. Other areas of research of the KCDC, not described-in this paper are design theory and practice.

Key wordr: Knowledge-based design; CAD; Collaborative design; Design creativity; Evolutionary design; Shape emergence; Design intent; Case-based design; Concept learning; Multiple abstractions

i ' L . Correswndinn author.

1. Computer-supported sjmchronous collaborative design

1.1. Aims

$ ~iscus i ion is open until December 1994 (please submit your discussion paper to the Editors on Architecture and Enginecr- The aim of this research [ I , 21 is to develop a ing, G. Smeltzer and H. Wagter). computer-based design environment in which de-

0926-5505/94/S07.00 O 1994 Elsevier Sc~ence B.V. All rights reserved SSDI 0 9 2 6 - 5 8 0 5 ( 9 4 ) 0 0 0 2 5 - 1

136 .\1..4. Rosenman er 01. auto matron in Construction 3 (19941 229-237

more concerned with the issue of knowledge representation, organization and access.

Further to the work related to the project on case-based design. on-going research is concerned with the organization of case memory with flexible indexing of cases for case retrieval and browsing using a multi-media approach [22, 231.

6.2. Concept leanling - machine learning

6.2.1. Aims The aim of this research is to examine the

application of machine learning techniques in the generation of design concepts from design examples. More specifically, the research is aimed at learning empirical knowledge to assist preliminary design ['A, 251.

6.2.2. Description Existing machine learning techniques are in-

sufficient in a number of ways. For example, much of the work on conceptual clustering builds upon Fisher's [26] COBWEB which provides a methodology for partitioning data spaces. How- ever. Alem and Maher [27] show that this is insufficient in itself as i t does not accommodate varied attributes across training examples nor learning associations among attributes within a cluster. In addition, associations between numerical and nominal data are not generated.

The work by Maher and Li (251 considers a target representation of design knowledge, called a design concept which is based on a group of design attributes and associations among these attributes and develops an approach to learning such concepts. The model for automatically learning design concepts from project data is an aggregation of machine learning and numerical analysis techniques and has two major components: concept aggregation and concept characterisa- tion. Concept aggregation involves decisions about which entities or features are to be selected and grouped into a concept. In concept characterisa- tion, the design concepts formed by the concept aggregation process are further defined by the empirical associations among the design attributes, in the form of empirical formulae and design patterns.

6.2.3. Summary The proposed model has been tested using 72

bridge designs collected from the Roads and Traffic Authority (RTA) in Sydney. Four concept spaces were generated that, although unnamed by the system, could be recognized as represent- ing four bridge design types: simply supported girder bridges; plank bridges; continuous bridges; and cable-stayed bridges. The empirical networks produced by the learning system represent a gen- eralization of the examples from previous design

' .- projects. This generalised knowledge can be used I ;

to provide assistance in the preliminary design of new design problems by associating sets of attributes with concepts and providing initial values'

I

for the attributes.

7. Summary

This report has outlined some of the research areas in knowledge-based design in which the Key Centre of Design Computing is involved. All these areas have in common the aim of contribut- ing to more useful CAD systems as design aids. All the areas have an effect on creative design and computer-supported collaborative design and their understanding and development are necessary for the understanding and development of creative design and computer-supported collaborative design.

Future research in knowledge-based design in the Key Centre of Design Computing has been outlined in each of the areas above, but in general, will concentrate strongly on creative design methods, emergence and computer-supported collaborative design. The latter area brings into focus the area of multi-media communication and research will also be directed in this area.

References

[ I ] M.L. Maher, J.S. Gero and M . Saad. Synchronous support and emergence in collaborar~ve CAAD, in: U. Flrm- ming and S. van Wyk (Eds.) . CA-ID Fur1rrrs'93. North. Holland. Amsterdam ( 1993) 455-470.

[2] M. Saad and M.L. Maher, A compul~t ional model for synchronous collaborative design. Ii'orktng hbtes AG11.9.1

,\1.,4. Rosenniun et a/. / Autoniarron m Co~~~rrrcrron 3 1199.11 229-237 -- 7 ; -

,<pt ~ m e d

It-

ed ges:

\ - .ks :n-

sign used

of at-

~ l u e s

lhch 1 the 1 411

11-

dids. *' 0"

Ijorkshop on ..(I m Collaborarr~~e Dcsrgn. July 1993. Wash- in: K.S. hlathur. h1.P. Betts and K . & Tham (Ed$.). Ington. D.C. (1993). .\ianagemrnt of It~formatron Technolog? for Consrrucrron .

[!I J.S. Gero. Des~gn prototypes: a knowledge representa- World Scientific. Singapore (1992) 239-251. [Ion schema for des~gn. .4rtrf Inrell. .\lag. I1 (4) (1990) 1161 M. Balachandran. R. Villamayor and X1.L. hlahrr. Using 26-36.

(41 J.S. Gero and M.A. Rosenman. A conceptual framework lor knouledge-based research at Sydney University's De- sign Computing Unlt. Artif. Inrell. Engrg. 5 (2) (1990) 65-77.

151 J.S. Gero, Creat~vity. emergence and evolution in design. In: J.S. Gero and F. Sudweeks (Eds.). Prepr~nrs of the 2nd In!. Conf, on Compurarlonal hlodels of Creatrre Deslgn. Department of Architectural and Des~gn Science, Sydney ( 1992) 1-26.

[6] I . Hybs and J.S. Gero, An evolutionary process model of design. Design Stud. 13 (3) (1992) 273-290.

[7] J.S. Gero. S.J. Louis and S. Kundu. Evolutionary learning of novel grammars for design improvement, Working Paper. Key Centre of Design Computing. University of S>dney. Sydney (1993).

[8] D.E. Goldberg. Generic Algorrrhmr m Search, Opr~miza- ~ I O I I , and .\lackme Learnmg. Addison-Wesley, Reading.

pas1 design cases to support structural s!stem design tor buildings. Progress Report for Acrr Wargon Ch~pman Associates. Deslgn Computing U n ~ t . Department of Ar- ch~techral and Design Sc~ence. Unlvers~t! of Sydner. Sbdney (1992).

1171 D.M. Zhang and M.L. Maher. Using case-based reabon- Ing for the synthesis of structural systems. In: hironkdgr Based System rn Cilrl Engmeenng. IABSE. Zur~ch (1993) 113-152.

[I81 M.A. Rosenman. Issues in the standard~zat~on of the communication of design intent. Unpublished ?iotes. Dc- sign Computing Unit. Department of Architectural and Design Sc~ence. University of Sydney. Sydney (1992).

(191 M.A. Rosenman and J.S. Gero. The what. the hou and the why in design. Iibrkrng ,Voles rtL41.93 I.lorishop on Reasonmg about Fio~ctron. Washington, D.C. (1993) 131- 136.

[20] W. Wang and J.S. Gero, A memor, organlzatlon for hlass. ( 1989). heterogeneous design knowledge. in: hirowledge-Based

[9] R.F. Woodbury, Design genes, Prepnnrs .Clodelling Cre- Sysremr In Clrrl Engmeenng, IABSE. Zur~ch (1993) 99- arlrrn and Aironledge-Based Creatrt,e Desrgn, Design 107. Computing Unit. Department of Architectural and De- 1211 M.L. Maher and D.M. Zhang. CADSYN: Using case and sign Science, University of Sydney. Sydney (1989) 133- 154. Also in: J.S. Gero and M.L. Maher (Eds.), Modeling Creari~,rry and Knowledge-Based Creatr1.e Desrgn. Lawrence Erlbaum. Hillsdale. N.J. (1989) 21 1-23?.

[ lo ] J.S. Gero. Shape emergence and symbolic reasoning us- Ing max~mal lines, Unpublished Notes, Design Comput- Ing h i t . Department of Architectural and Design Sci- ence. Universiv of Sydney. Sydney (1992).

[ I l l J.S. Gero and M. Yan. D~scovering emergent shapes using a data-driken symbolic model, in: U. Flemrn~ng and S. van Wyk (Eds.). C A D Flrrures'93. North-Holland. Amsterdam (1993) 3-17.

[I21 A.T. Purcell and J.S. Gero. The effects of examples on the results of a design activity, in: J.S. Gero (Ed.). Artifi- cral Inrellrgencc m Desrgn '91, Butteworth-Heincmann. Oxford (1991) 52-54?.

[I31 J.S. Gero and M. Yan. Shape emergence by symbolic reasoning, Working Paper, Design Computing Unit, De- partment of Architectural and Design Science, University of Sydney. Sydney (1992).

[I?] M.A. Rosenman and M.L. Maher. Formal representa- tlons for communicating design intent, Technical Repon. Key G n t r c of Design Quality, University of Sydney, S!dney (1992).

[I51 X1.A. Rosenrnan. J.S. Gero and Y-S. Hwang, Multiple concepts of a design object based on mult~ple functions.

decomposition knowledge for design synthesis. in: J.S. Gero (Ed.), Arrificial lnrelligence rn Design '91. Butter- worth-Heinemann. Oxford (1991) 137-150.

[22] M. Balachandran and M.L. Maher. Towards an organlza- tion of case memory with flexible indexing of cases. Unpublished Notes. Design Comput~ng Un~t . Depart. ment of Architectural and Design Sc~ence. L'n~\ers~t) of Sydney. Sydney ( 1993).

[23] M.L. Maher and B.M. Balachandran. A mult~mcdia approach to case-based structural design. Compur. Cii.11 Engrg.. submitted.

1241 M.L. Maher and H. Li. Learning emp~rical knouledge to assist preliminar). design, in K.S. hfathur. h1.P. Bctts and K.W. Tham (Eds.). ~lianagcmtnt of In(ormarron Technol- o m for Conrrccrion, World Sc~entific, Singapore (1993) 301-317.

[25] M.L. Maher and H. Li. Adapting conceptual clustering -for preliminary structural design, Amer. Soc. Cil Eng., to appear.

[26] D. Fisher, Knowledge acquisition via incremental conceptual clustering, Machine Learning ? ( 1987) 139- 172.

[27] L. Alem and M.L. Maher, Using conceptual clustering to learn about function. structure and behaviour in design, in: D. Herln-&me, R. Dieng. J.P. Regourd and J.P. Angoujard (Eds.). Ejronledge blodrNing and Erpmrse Transfer. IOS. Amsterdam (1991) 163-177.

for .. 93

and ' d o d s .

ELSEVIER Automatron in Construct~on 3 (1994) 203-217

Central or distributed organization of design systems?

Ivan K. Petrovic The IMS Inrrirure, Belgrade, Yugosla~ia

Abstract

The paper describes two design systems prototypes developed at the IMS Institute: GIMS-EXPERT, a centrally controlled system made in 1987, and GIMS-DDS, a distributed design system still under development. Both systems produce design sketches of the family houses to be built in "GIMS", a catalogue-based, prefabricated building system. Presented and discussed are the two systems gestations, basic concepts, and examples of output. The more rigid EXPERT produces the expected designs and is suited for the well-defincd requirements and contexts. while the more flexible DDS produces satisficing answers in ill-defined contexts, and can even produce "unexpected" designs. I t can also simulate the performance of GIMS-EXPERT.

Key words: CAAD; Design expert systems; Distributed design systems

1. Introduction why, while a "Design methods problem" relates

The paper describes two design systems prototypes developed at the IMS Institute: GIMS-EX- PERT, a centrally controlled system, and GIMS- DDS, a distributed design system. Both systems produce design sketches of the family houses to be built in "GIMS", a catalogue-based, prefabricated building system [I]. The application domain suggests that all possible design solutions that can be produced (or "selected") from the system cat- alogued elements are determined in advance. If this is the case, a "Design object problem" relates to the question of which design to select and

to the question of which design methods and tools to apply for the selection of the required designs, and eventual generation of the alternative designs from the existing elements.

Two design systems mentioned reflect two types of organization:

(i) a closed, "deterministic machine" (GIMS- EXPERT) and

(ii) an open, distributed, "almost self-organizing system" (GIMS-DDS). How do these systems treat the aforementioned design problems? Which, of the two, should be selected for further development into the complete "GIMS Design ~ystern"? We shall firstly

1 * Discussion is open until December 1994 (please submit your describe both systems' traits and after a limited

. * discussion paper to the Editon on Architecture and Engineer- attempt answers !hese ques- lng, G. Smeltzer and H. Wagter). tions.

0926-5805/94/507.00 0 1994 Elsevier Science B.V. All rights reserved SSDI 0 9 2 6 - 5 8 0 5 ( 9 4 ) 0 0 0 2 3 - G

the situation changes. Then, this tool becomes a fast and challenging interface, a full body contact between man and machine, where each partner can fool the other. The most important thing here is that every idea is represented as a 3-d object within seconds. As i t happens, i t is just possible that the PDP-AAM may be an under- cover "agent-anarchist" [23]; when it enters the design process, anything can happen (even.creative leaps?).

4.4. What kind of sysrem is the GIMS-DDS?

The development of DDS tools started as an attempt to re-make the GIMS-EXPERT to be faster and more efficient. As the various tools appeared, doubts were risen as to what control, if any, to introduce. The pragmatic and empirical approach, "experiment first and propose and check a theory later" took its toll in sometimes unnecessary work (e.g, in making several very similar expert system shells), but the produced "kit of tools" proved to be an interesting simulat-

ments on this manuscript more than once, and produced the DDS illustrations. The author thanks Ulrich Flemming for his inquiry on the reasons why we put an unstable member ltke PDP-AAM into a respectable company of the tools based on logic; ~t really opened up a serles of new questions still unanswered but challenging. Thanks are also due to Jens Pohl for showing an example of what a real distributed design

- system ought to look like. The last but not least

I . mention goes to Gianfranco Carrara who thought . - - - -- - I

up an interesting conference theme and hosted ; 5~ i*, the presentation of this work, and the co-chair- man, Yehuda Kalay, for his patience in having to , cope with our ever-changing text. The project has been supported by the research grants from the Scientific Council of the Republic of Serbia and the IMS Institute.

References

[ I ] 1. Petrovic and 2. Lazovic, GIMS system, LWS Imr. B~rll . ,

ing (and stimulating) device. GIMS-EXPERT was Belgrade 9 (2-3) (1982) 2-50.

reincarnated, but we are still charting possible [2] 1. Petrovic, M. Novkovic, M. Cubric. I. Svetel and 2. Minjevic, GIMS-EXPERT: consultant for des~gn of fam-

future paths of the DDS development. Obviously, ily houses in GIMS building system. froc. I l l Int. Conf: the adapting organization of the DDS enables the on Research and Der.elopmenr of Building Technolog~es formation of "static and dynamic frameworks" [ 5 ] 1,tIS '87. Belgrade (1987) 227-245. I

but would it enable the creation of something I31 C. Alexander, Nores on the S.vnrltesic of Form, Harvard I I

that would perform like a "self-organizing University Press. Cambridge, Mass. (1964). [4] C. Alexander, S. Hirshen. S. Ishikawa. C. Coffin and S.

system"? Why would anyone need such a system? Angel, Houxs generated by patterns. Center for Enviro- Further development of DDS plus all sorts of mental Structure. Berkeley, Calif. (1969). new PDP-agents such as "analoaists". "mutants" 151 K.E. Boulding, General system theory-the skeleton of . .

and othersW[24], point out to new directions of science, an& Sci. 2 (3) (1956) 197-208.

unpredictable methodological turns in CAAD (61 C. Hewitt, Open information system semantics for dls- tributed artificial intelligence, An$ Intell. 47 (1991) 79-

theory and practice. 106.

[7] L. Gasser and R.W. Hill, Engineering coord~nated problem solvers, AM. R ~ L . . Compul. Sci. 4 (1990) 203-253.

Acknowledgments [8] L. Gasser, Social conceptions of knowledge and action: DAI foundations and open systems semantics, Artif. In- tell. 47 (1991) 107-138.

This project have been impossib1e with- [9] J. Pohl, L. Myen, A. Chapman, J . Snyder, H. Chauvet. J . out the talent and enthusiasm of many full- and Cotton. C. Johnson and D. Johnson, ICADS work~ne part-time members of the IMS Design Research model version ? and future directions. Technical Report. workshop learn in the period 1987-91, especially CADRU-05-91, CAD Research Center. Design Institute.

Maria Cubric, Dejan Simovic and Alexander College of Architecture and Environmental Design. Cal Poly. San Luis Obispo, Calif. (1991).

.Miric. The Project Director is Ivan Petrovic. Par- [ lo ] J , Pohl. L, Myen, I, Cotton, A, Chapman, J , Snyder, H, ticular acknowledgments are due to Ig0r Svetel Chauvet, K. Pohl and J . La Porta. A com~utrr-based who has developed all DSS toois, made corn- design environment: implemented and planned exten-

the : 1 e s 1 Ig -

~ , i n g . gn

. 1st .

. ght i

~qted t

r 1 ; t 0

: has !

n e

i

i

11.. i

4d z,, farri-

? i

na rd i i

5, * s l ro- s

i

Of j 1 79-

I S - 1 b-

2 .

tlon: ' $1.

c,, J . -k~ng

1 I * t

; 1. i

slons of the ICADS work~ng model. Technical Report. CXDRU-06-92. CAD Research Center, Des~gn and Con- structlon Institute. College of Architecture and Environ- mental Design. Cal Poly. San Luis Obispo. Calif. (1992).

[ I I ] G. Polya. How ro Solr,e 11. Pr~nceton Univers~ty Press. Pr~nceton. N.J. (1945).

[I:] G.H, von Wnght. The logic of action - a sketch. in: N . Resher (,Ed). The Lngrc of Dvcrsron and Acrron. Univer- srt: of Pittsburgh Press. Pittsburgh (1967) 121-136.

1131 I. Petrovic. Systems and design: decisions between reality and conjecture. Ph.D. Thesis, unpublished, Department of Arch~tecture and Plann~ng, The Queen's University of Belfast (1972).

(141 A. Hickling, Beyond a linear iterative structure, in: B. Evans. J . Powell and R. Talbot (Eds.), Changing Design, Wilry. New York (1982) 275-194.

[ l j ] I . Svetel. Distributed systems for computer-aided architectural design. Proc. Con! on Informat~on Sysrems and Con~plrrer Sofrware rn B~iildrng. Arandjelovac. Yugoslavia ( 1992) 53-64.

[ lh ] I . Svetel. ARCH: an architectural layout synthesis program. /,\IS Bull.. Belgrade 5 ( 1 ) (1990) 17-25

(171 D.E. Rumelhart and J.L. McClelland (Eds.). Parallel Drsrrrhured Processmg. MIT Press. Cambridge, Mass. (1986).

1181 R.R. Coynr. Tools for esplor~ng assoclatr\e reasoning In design, in: hl, hlcCulloch. W.1, hlitchell and P. Purcell (Eds.1. Tltc Elvrrronrc Desrgn Srrrdro. MIT Press, C j m - bridge. Mass. (1990)..

119) 1. Petrovic. Integrative knowledge-based des~gn s\srems a v~ru,, in: T. Terai (Ed.). Comptrrcr lt~rcrgrarcd Co~~srn,c- rrott. CIB Proceedings. Publication 138. Ch~ha Insi~tu[e (11'

Technoloa and Arch~tectural Inst~tute of Japan. Tok!o (19911 129-134.

1201 1. Petrovic and I . Svetel. Arch~tectural computer.cl~ded design systems: an example. CIB Y? Itvrld Cotrqrrss. Montreal (1992).

[? I ] I. Svetel. PDP.rUh.i: an analog~cal arch~tectural mod. eller, Proc. ARECDAO '93, Barcelona 11993) 179- 190.

[?2] A. Miric. I. Svetel and I . Perrovic. Development of expert system prototypes using OYSTER, lhlS Bull.. Belqr~~dt~ 5 ( 1 ) (1990) 26-34.

[23] P. Feyerabend. Agarnsr Jlethod. Verso. London (1975). [24] J.S. Gero and M.L. hlaher. hlutatlon and analog) to

support creativity in computer-a~ded design. In: G.N. Schm~tt (Ed.), C M D Futures '91. Department of Archl- tecture. Swiss Federal Institute of Technolog), Zur~ch (1991) 241-249.

ELSEVIER Automat~on in Construction 3 (1994) 187-202

I Knowledge-based design support '

James H. Rutherford, Thomas W. Maver * ABACUS, Un~i,ers~ry of Stmrhclyde. Glasgow G4 ONG, UK

1. Historical perspective

In 1971 the Architects' Journal featured a paper entitled PACE l: Computer Aided Building Appraisal [I]. The sofnvare known as PACE (Package for Architectural Computer Evaluation), initially developed on the Systemshare time-shar- ing system accessed on-line from a teletype termi- nal over the ordinary voice grade telephone network, has, in concept at least, survived the subsequent 22 years and remains to this day a crucial aid in the teaching of architectural design in the University of Strathclyde's Department of Archi- tecture and Building Science.

The intention of PACE, subsequently known as GOAL (General Outline Appraisal of Lay- outs) was to make explicit to students and practi- tioners the "cause and effect" of decision-making at the early conceptual and formative stages of the design activity, i.e. how her/ his design choices regarded form and fabric impacted on the range of cost and performance consequences which characterise the design (Fig. 1).

The user of GOAL created form by shaping and placing volumes, and chose fabric by selection from a constructional database. By naming

* Corresponding author. Discussion is open until December 1994 (please submit your

discussion paper to the Editors on Architecture and Engineer. ing, G. Smeltzer and H. Wagter).

volumes (e.g. classroom. store, stairway) the user ensured that the software found the relevant functional and environmental requirements of the space (e.g. air change rate, proportional limits, requirement for daylight, contiguity constraints). By choosing a particular construction, the user ensured that the software found the relevant physical properties of the building fabric (e.g. thermal transmissivity, and mass, acoustic attenu- ation transparency, colour).

The topographical and thermo-physical properties of the hypothesised design were available to a range of calculation routines some of which were well established (e.g. energy consumption algorithms) and some of which had to be invented (e.g. algorithms for the computation of planning efficiency).

The detailed output from GOAL, which was organised hierarchically to suit the needs of particular users, was also available as a summary of design and cost/ performance variables, known as the "fingerprint" of the design (Fig. 2).

The fingerprint of all design hypotheses relat- ing to a particular brief could be progressively saved and subsequently analysed. The analysis offered a mie ty of options:

- The cost/performance profile of any indi. vidual design could be displayed as a histogram in relation to a base line (or benchmark). The base against which a design was to be compared could be selected by the user and could be, for exam-

Elsevier Science B.V. SSDI 0926-5805(94)0002?-F

lild~ng ions

tcnus. i s ions

tion. .lore

s plan

lild- .Iputa- values

mtal ... ig in

- pro- raise

urrent :pes of

(Fig. , to-

' the ,,OSSl-

stand- ~gies, .lake

design

m a 1 . a v i n g in the IS to _ the

make :ma- tio on

d they lPSS of

,. the

FIE. 17. Automated elicitatton of a des~gner's belief structures and the derivation of models of the des~gn process.

back-end of the framework would normally be used once a complete description of a building had been resolved. The use of sophisticated design tools of this form would therefore normally be used in a post hoc checking mode. The framework, adding contextually sensitive defaults, makes these applications accessible at an earlier stage of the design process, providing an opportu- nity to identify and resolve potential problems before the production drawing stage, thus saving valuable time and resources.

5.1. F~tntre research

The KNODES environment blackboard, by monitoring and recording transactions between knowledge resources, provides a mechanism for recording and tracing design decisions.

This facility is to be used to provide on-line protocols of design activity. It is believed that where i t is possible to enumerate designs by means of a classification or a diagnostic system (Fig. 171, design fingerprints, measured against known metrics, can be used to elicit a designer's belief structures and therefore provide some insights into the nature of design decision making.

References

[ I ] T.W. Maver. PACE 1: computer-aided build~ng appraisal, Archrtects'l. (July 1971) 207-211.

121 N. Vidovic. Design framework tool kits for concurrent engineering environments. Engineerrng Drsign Research Centre. CMU. Pittsburgh. PA 15213. 1990.

[3] Encyclopaedia Britannica Inc.. .\lrcro .\ledlo. Vol. 3. page 45. 15th editron (1975).

141 D. Gentner and A.L. Str\ens. ,\lc~ri/ui .\lodeis. Laurence Erlbaum Assocrares. London ( 1983).

[5] P. Chu and J.J . Elam. Induced system restricti\encss: an experimental demonstrauon. IEEE Trans. 5,vstrrns. .\fun. Cykmer.. 20 ( 1 ) (1990) 195-201.

(61 S. Weis. C. F;ulikowski, C. Apte and bl. Uschold. Build. ing expert systems for controlling complex programs. in: . W l 1952 (1982) 322-326.

[7] J. Morton. P.J. Barnard. N. Hammond and J. Long. Interacting with the computer: a framework, in: E.J. Boutmy and A. Denthe (Eds.). Trlrmn~ormarrcs 79 (1979).

[S] M.D. Wilson. Task analysis for knowledge acquisition. Proc. SERC Il'orkshop: Knonbciggl. .-lrqrturrton for Engr- neering Appbcarrons. Abingdon (1987).

[9] P. Hart. D~rections for A1 in the Eighties. SIGART h'cwslrrt., 79 ( 1982) 1 1 - 16.

[ lo ] P.J. Hayes. Naive physics I - onlolop for liqutds. hlemo Centre pour les etudes Semanttques el Cognillves. Geneva. 1979.

( I I ] I. Lansdown. Requirements for knowledge based ssslems in design. in: A. Pipes (Ed.). C,-t4D Atrures. /'roc. of Internar. Con! on C.&4D. September 1985. Butteruorths. London (1985) 120- 127.

[I!] R.S. Englemore and Morgan. A.J. (Eds.). Blackboard S,vsrems, Insight Series in Artificial Intelligence. Addison-H'esle). Reading. Mass. (1988)

1131 hl. Prime. User Interface Management Systems (UIMS) - a current product review. Informatics Division. Rutherford Appleton Laboratory, 1988.

[I41 hi. Mead. Data modelling language "Express". Informat- ~ c s Department. RAL, 1990.

[l j l STEP 1988. General AEC Reference Model (GARM). Standard for the Exchange of Product Data I S 0 TC184/ SC1/NGI (Draft) Document 3.2.2.1. 12th October 1988

[16] J .H. Rutherford, Forms - A multi-faceted user interface, MSc D~ssertation, Department of Architecture, University of Strathclyde, Glasgow. UK 1987.

[I71 J.H. Rutherford, An intelligent design suppon environment. PhD Thesis, Department of Architecture, Univer- sin. of Strathclyde, Glasgow, UK. 1990.

[I81 P. Totterdell and P. Rautenbach, Adaptation as a problem of design, in: D. Browne, P. Totterdell and M. Norman (Eds.), Adaptice User Inrerfacu, Academic Press. Boston (1990) Ch. 3.

[I91 Autodesk. AutoCAD R11 Reference Manual. Autodesk. Guildford Surrey.

[ZO] D.N. Chln. Modelling what the user knows in UC, in: L'ser .2fodels m Dialogue Sysrem, Symbolic Compurar~on, Spnnger. Berlin (1989) 74-107.

[21] W. Kaemmerer and J. Larson. A graph oriented knowledge representation and unification technique for auto- matlcally selecting and invoking software functions, In: R441.86, Philadelphia. Pa. (1986).

(221 Ontos. OKTOS Object Database - Development and User Guide. Ontological Inc. Three Burllng~on Woods. Burlington. MA 01803. 1990.

[23] B.R. Anderson. A.J. Clark. R. Baldwin and X . 0 . Mil- bank, BREDEM - BRE Domestic Energy Model: background, philosophy and description, Building Research Establishment, Department of the Env~ronmenl. 1985.

1241 N. Baker, D. Hoch and K Steemen, The LT method version 1.2 - energy design tool for non domestic build. ings, Commission of the European Communities Direc- torate-General XI1 for Science, Research and Develop- ment and Directorate-General XI11 for Telecommunica- tions. Information Technology and Innovation.

[25] T.A. Markus. T.W. Maver, P. Whyman. J. Morgan. D. Whitton, D. Canter and J. Flemming. Bulldlng Perfor- mance, Halsted Press. New York (1972).

[26] D.R. Reddy, L.D. Earman and R.B. Neely, A model and a system for machine recognition of speech, IEEE Trans. Audio Electro-acoust.. 21 (1973) 229-238.

[27] B.C. Bjork. Basic structure of a proposed building product model. CAD, 21 (2) (1989).

Bal- l..,

3f' 5 ?

5 9 3 t 1 L

A -2 T a 5

i I

Z !

ELSEVIER Automalion in Construction 3 (199.1) 157-175

Knowledge-based computational support for architectural design

Gianfranco Carrara a , Yehuda E. Kalay b-*, Gabriele Novembri ' a Dlparrrmenro di Archlterrura Tecnlca e Tecnica Urbanirrlca. Uni~,ersiti degli Srudi La Sap~en:a, Rome. Irah

Department of Archirecrure, Uni~,ersrr) of Cclliforni~. Berkeley, USA Carresiana Consorrlum, Rome. Italy

Abstract

The process of architectural design aims to define a physical form that will achieve certain functional and behavioral objectives in a particular context. It comprises three distinct, but highly interrelated, operations: (1) definition of the desired objectives; (2) production of alternative design solutions: (3) evaluation of the expected performances of the solutions and their comparison to the predefined objectives. Design can be viewed as a process of search for a solution that satisfies stated needs, while at the same time adapting the needs to the opportunities and limitations inherent in the emerging solution.

Computational techniques were developed to assist each one of the three operations, with varying degrees of success. We propose to integrate all three operations into one whole, by developing a computational model that will facilitate smooth transition from one operation to another. The role of computers in supporting this model will include providing a database of prototypical design objectives and solutions. storing project-specific design goals and solutions, and predicting their expected performances. This paper discusses the rationale and background for developing such a knowledge-based design system, and presents the parameters for implementing it as a computational tool to support architectural design. Examples from a protoppe implementation serve to illustrate the discussion.

Key words: Design process; Knowledge base; Partnership paradigm: Design goals; Prototypes

I.. lntroduction riences and knowledge. The purpose of the process is to define an object (or an environment)

We consider architectural design a goal-di- that achieves some desired behavioral and spatial rected search process which relies on prior expe- characteristics, while conforming to and relying

upon cultural, social, environmental, and other norms. We refer to the object that is being designed as the solution, and to the desired behav-

* Corresponding author. " Discussion is open until December 1994 (please submit your ioral and spatial characteristics it strives to discussion paper to the Editors on .4rchitecture and Engineer- achieve as the goals. We view the Process of ing, G. Smeltzer and H. Wagtcr). design as a dialogue between the goals and the

09?6.5805/94/S07.00 0 1993 Elsevier Science B.V. All rights reserved SSDl 0926-5805(94)00020.N

f :

i C. edge

'legro ; of the

n im- ter-

: was !

? 'cent, sing !

t h e 1 ; i j

, i the

~t the ! e sig- 1

ter- de-

lesign

5 i that f zome

reas B i r' aiall-

!d for i

ready --t of

la in

; that itec- f

. . -*ntiy

ap- 'ully

1 that pom- :ries

G. Carraro et al. /Automutton ul Construcrron 3 (194.1) 157-175

achieved outstanding results doing so without the aid of computers. It is our contention that it is not necessary to fully automate each and every one of the design process activities in order to significantly improve design productivity and quality. Rather, it is more prudent to develop a practical symbiosis between the capabilities of designers and machines.

The implementation of our approach is based on viewing architectural design as a process of search, which aims to reconcile the differences between a set of requirements, given by the client, and the expected behavior of a design solution, proposed by the architect. In the course of nego- tiating the differences between the two'ends, tradeoffs must be made on both sides. The process also contributes to better understanding the problem itself, and informs both the architect and the client of initially hidden opportunities and irreconcilable conflicts.

The prototype we have developed forms a framework for implementing our proposed design partnersirip paradigm. In such partnership, the role of the computer can be shifted dynamically between passive representation/evaluation and active generation/evaluation of design solutions. Such dynamics would allow the designer and the system to respond to changing requirements, un- foreseen problems, and emerging opportunities as they arise during the design process. The system demonstrates the feasibility of implementing this paradigm. Its utility for practicing architects is, nonetheless, yet to be tested.

Acknowledgments

The authors would like to thank the Director, Professor Ing. Claudio Cerruti, and all the man-

aging staff of the Progetto Finalizatto Edilizia of CNR (the Italian National Council for Re- searches). without whose support the present work would nqt have been possible; the CARTE- SIANA consortium for their promotion of this initiative; and all the research staff of the la Sapienza University, Rome, the New York State University at Buffalo, and the University of Cali- fornia, Berkeley, who have worked for the last four years on the project.

References

[ I ] H.A. Simon, The Scrrnccs of the Anifictal. MIT Press. Cambridge. Mass. (1969).

[2] M. Minsky (Ed.). Semantrc ~nfonnorron Processmg. MIT Press, Cambridge. Mass. ( 1968).

[3] R.S. Liggett. H'. Mitchell and M. Tan. Mult~level analys~s and optimization of design, in: Y.E. Kalay (Ed.), Ei.aiua- tion and Prediction m Design, Wileg Intersc~ence, New York (1991).

[4] Y.E. Kalay (Ed.). E~.aluation and Predtctron In Dcstgn. Wiley Interscience. New York (1991).

[S] G. Carrara, Y.E. Kalay and G. Novembri, Multi-modal representation of design knowledge, Automat. Conrrruc- tlon, I (2) (1992) 11 1-12?.

[6] L.M. Swerdloff and Y.E: Kalay, A pannership approach to computer-aided design. in: Y.E. Kalay (Ed.). Com- p~ttabilir) of Design, Wileg. New York (1987).

171 J.S. Gero and R.D. Coyne. Knowledge-based planning as a design paradigm. Technical Report. Computer Applica- tion Research Unit. Dept. of Architectural Science. Univ. of Sydney. Australia, 1985.

[8] W.J. Mitchell, The Logrc of Archttecrure. MIT Press. Cambridge. Mass. (1990).

[9] A.M. Kodijat. A computational model for the design of office lighting. Masters Thesis. School of Architecture and Planning, SUNY-Buffalo, 1991.

[lo] R. Subramanian, A computational model for architectural design, Masters Thesis, State University of New York at Buffalo. 1992.

[ I l l Y.E. Kalay, Modeling Objects and En~,ironments, Wiley, New York (1989).

Automation in Construction 3 (1994) 123-133

Case-based design in the SEED system '

Ulrich Flemming * CAD Iniaotire, Technical Unirersify of Denmurk 2880 Lyngby, Denmork

SEED is a software environment to support the early phases in building design currently under development. m o n g the capabilities intended for SEED is the automatic storage of solutions generated with the system as cases and their retrieval for reuse in a similar problem situation. SEED responds to practical demands requiring an approach to case-based design that differs from those known from the literature. The paper specifies the SEED requirements for case-based design, outlines an approach to satisfy the requirements, and presents technical details on the mechanisms under development.

1. Introduction

This paper outlines an approach toward the representation and reuse of past solutions in SEED, a Software Environment to support the Early phases in building Design. The project sponsors are particularly interested in capabilities that support work with recurring building rypes, that is, building types dealt with frequently in a design firm. Such firms, from housing manufacturers to government agencies, accumulate considerable experience with recurring building types. But this experience is located currently mainly in the memory of designers and in institutional

Author's pennanent address: Carnegie-Mellon/Building Industry Computer-Aided Design Consortium (CBCC), Carnegie-Mellon University, Pittsburgh, PA 15213, USA.

Discussion is open until December 1994 (please submit your discussion paper to the Editors on Architecture and Engineer- ing. G. Smeltzer and H. Wagter).

records. Computational support for bringing this knowledge to bear on new projects remains dis- persed and loosely coupled. SEED intends to provide systematic and broad support for both the creation and evaluation of new designs and their reuse in a new, but similar context.

This motivation aligns SEED with current work on case-based design systems [ l ] and, less closely, with prototype-based design systems [2]. But none of the approaches described in the literature matches exactly the requirements and application context of SEED. Section 2 introduces these requirements, starting with a brief overview of SEED. Section 3 outlines an approach toward the reuse of solutions in the SEED context that matches the requirements. Section 4 deals with technical issues that arise from this approach. Readers not familiar with the terminology and techniques of case-based reasoning and design. a subfield of Artificial Intelligence (All, are referred to [ I ] for an introduction.

0926-5805/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved SSDI 0 9 2 6 - 5 8 0 5 ( 9 4 ) 0 0 0 1 7 - H

2s. The .-'ional

d de- at the .nit are

prob- rreat-

:tin? a11

.JI

gners a listed f the

by the I unit

with a

ocate . solu-

L' Fhmnrng / Automatron m

f operation can be used to eliminate functional units that do not exist in the problem specifica-

or that are incorrectly allocated. The edit can be used to adapt the dimensioanl

of a design unit to satisfy certain constraints. At this point, the retrieved solution is more or less equivalent to an initial state with preplaced units. and the normal add and generate operators can be used to allocate or re-alioc- ate functional units that are not yet or no longer allocated.

At any time, a design unit may have no functional unit associated with it, in which case all constraints on its shape and location are relaxed. This suggests an exciting extension of the retrieval procedure' outlined in Section 4.2. If we allow functional units to be comparable - on an optional basis - not only with subclasses, but also with superciasses, we may be able to retrieve layouts based on size and dimensional constraints alone; for example, we may retrieve a narrow, linear scheme from a project dealing with building functions that differ from the ones currently under consideration.

Acknowledgments

The development of SEED is sponsored by Battelle Pacific Northwest Laboratory, the US Army Corps of Engineers Construction Research Laboratory (USACERL) and the Engineering Design Research Center (EDRC) at CMU, an NSF-supported Engineering Design Research Center. The research reported here was supported by grant no. 16-5125-1 OS from the Dan- ish Technical Research Council.

References

[ I ] J.L. Koloairer, Improving human decision making through case-based decision aiding, A1 Mag., I 2 ( 2 ) (1991) 52-68.

[!I J . Gero, Design prototypes: a knowledge representation schema for design. .dl ,\tag.. I 1 (4) ( 1990) 26-36

I31 L'. Flemming. R. C o y e and R. Woodbun. SEED 3

software environmenl to support the earl! phases In building design. in: ARECDA09.3 (1993) 1 1 1-122.

(41 R.F. Coyne. U. Flemm~ng. P. Plela and R. Woodbun. Behavior modeling in design system development. in: L'. Flemmlng and S. Van Wyk (Eds.). C,44D Futrtrrs 'Q.:.

Elsevier. Amsterdam (1993) 335-351. [5] U. Flemming. R. Covne, T. Glavin and M. Rychener. A

generative expert system for the des~gn o l bu~ldrng la!. outs - version 2, in: J . Gero (Ed.). Anrficral Irrtellrgo~ct~ m Engmeenng: Design, Elsevier. Ansterdam ( 1988) 145 - 463.

[6] U. Flemming. More on the representation and generation of loosely packed arrangements of rectangles. En1 I -

ron. Planning 8. Planning and Desrgn. 16 (1989) 32i-359. [7] R.F. Coyne and U, Flemmrng, Planning in design synthe.

sis: abstraction-based LOOS, in: J . Gero (Ed.). .4rtrficral brtelligence m Engineerrng L: Vol. 1: Desrgn. Springer. New York (1990) 91-111.

181 R.F. Coyne. ABLOOS. An evolving hierarchical des~gn framework. PhD Dissertation. Dept. of kchitecture. Carnebie- ello on University. Pittsburgh. PA. 1991.

[9] J. Heisserman. Generat~ve geometric design and bound- a y solid grammars. PhD Dissertat~on. Dept. of Xrchitec- ture. Carnegie-Mellun University. Pittsburgh. PA. 1991.

(101 J . hersserman and R. Woodbur)., Generating languages of solid models. Proc. 2nd A C . V / I E E E Conf on Sohd Modeling and Applicarrons, Montreal. Canada ( 1993).

111) K. Hua, I. Smith. B. Faltings, S. Shih and G. Schmitt. Adaptat~on of spatial design cases. in: J . Gcro (Ed.). Artifirral Intelligence 1t1 Design '92. Kluwer Academ~c. Boston (1992) 559-575.

[12] E.A. Domeshek and J.L. Kolodner, A case-based desrgn aid for architecture. in: J. Gero (Ed.). .4rr!frcral lntelir- gence rn Design '92. Kluwer Academic. Boston (1992) 497-5 16.

1131 M.A. Rosenman. J.S. t iero and R.E. O m a n . What's in a case: the use of case bases. knowledge bases. and databases in design. in: G.N. Schmitt (Ed.). C A 4 D Fu- tures '91, Vieweg, Wiesbaden (1991) 285-300.

[14] T.R. Hinrichs, Problem Solring in Opcn I.!orldc. A Case Study ,in Design, Erlbaum Associates, Hillsdale. N.J. (1992).

[IS] R. Keeney and H. Raiffa, Deririom with ,itdtiple Objer- ii~,es: Preferencrs and Value Tradeoffs, Wile)., New York ( 1976).

:pre- nrdi-

3s thus which roce-

, &move

ELSEVIER

Framework for development of CAD/CAC systems ''

Raghavan Kunigahalli * , Jeffrey S. Russell Deparrmenr of Ci l .11~4 Eni.ironmenra1 Engineering, Unii~ers~fy of Wuconsin. ,Ataditon. H'153705, USA

Abstract

This paper provides a framework for the development of Computer-Aided-Design/ Computer-Aided-Construction (CAD/CAC) systems. A description of the proposed CAD/CAC systems is provided. Thc proposed system contains software tools related to (1) conceptual design, ( 2 ) structural and foundation analysis. (3) design of structural components. (4) routine geometric, mathematical, and optimization functions. ( 5 ) construction management. (6) Computer Aided Process Planning (CAPP), (7) process simulation, and (8) constructability analysis. On-going research work in areas related to CAD/CAC systems are highlighted. Incorporation of Group Technology ( G I ) paradigm for civil engineering structures to enhance the efficiency and effectiveness of the proposed systems is investigated, Issues related to algorithm development are also discussed. Possible extension of Automatically Programmed Tools (APT) language to include construction industry processes are examined. Specifically, the programming level syntax of APT can be extended by adding construction related vocabulav and features to only two out of five statement types and processor-level syntax can be extended by incorporating additional modular features.

K ~ n ~ o r d s : Construction process planning; Computer aided process planning; CAD/CAC; Rectangle adjacency graph (RAG); Automatically programmed tools (APT); Group technology (GT)

1. Introduction

There are three main participants in facility delivery process: (1) an owner, (2) a designer, and (3) a constructor. According to Miyatake, Manag- ing Director of Shimizu, Computer Integrated Construction (CIC) includes not only design and construction issues, but also business-related functions such as marketing and financing of the three main project participants. Computer-

' Discussion is open until August 1995 (please submit your discussion paper to the Editor on Construction Technologies. M.J. Skibnieuski).

' Corresponding author.

Aided-Design/ Computer-Aided-Construction (CAD/CAC) systems proposed by Kunigahalli and Russell [17], is a subset of CIC and focuses upon design and construction issues within the facility delivery process.

The proposed design and construction integrating system requires multi-disciplinary research efforts in, a variety of areas such as (1) Computer-Aided-Design (CAD) and geometric modeling, (2) algorithms and data structures, (3) Artificial Intelligence, (4 ) Computer Numerical Control (CNC) and robotics, ( 5 ) operations research in stochastic and deterministic processes, (6) data communication with Local Area and Wide Area Networks, (7) simulations, (8) Com-

0926-5805/95/S09.50 Q 1995 Elsevier Science B.V. All rights reserved SSDl 0926-5805(94 )00039-5

e s. n e

'17 ce - r - ; )

, , ( 3 ) , i i -

2s

.ial n be

AS

i t ,

' ds .in

..ing ' rut-

'T f ier

4 ef- as TI -

h c . n-

iink- lb-

'l e :ngi- 'nm-

e d , of

:son- 7 tS

: d uage

* h e 21-

a . r l t -

scus- ic

Acknowledgments

The second author wishes t o thank the Na- tional Science Foundat ion for G r a n t No. MSM- 9058092, Presidential Young Investigator Award, for financial support of this effort.

References

[ l ] ANSI. Amencan National Standards for Information Systems-Programming Language-APT. American Na- t~ona l Standards Institute. Inc., New York (1987).

[?I T. Aoki. T . IOmura. K. Momozaki and A. Suzuki, Graph- ical site simulation using an object oriented CAD model. Proc. lOrh Int. Symp, on Automation and Roborrcs in Consrnictlon. Houston. Texas. May 23-26. 1993. pp. 213- 220.

[3] J.L. Bentley. Multidimensional binary search trees used for associative search~ng. Comm. AC.U, 8 (9) (1975) 509- 517.

(41 J.L. Bentley, Multidimensional divide-and-conquer. Comm. ACM, 23 (4 ) (1980) 114-230.

[5] K. Beucke and D. Ranglack. Computing with objects: what does industry hope to gain from it. Proc. 5th Int. Conf. on Computing in Ciri l and Building Etrgineenng, Anaheim. Calif.. June 7-9. 1993, pp. 102-109.

[6] T . Chang, R.A. Wysk and S. Wang, Compurer-Aided .5lanufactunng, Prentice-Hall, Englewood Cliffs, N.J. (1991).

[7] R. Damrath, Object based visualization of physical behavior, Proc. 5th Inr. Conl, on Comprtring in Cii,il and Burlding Engineering. Anaheim, Calif., June 7-9, 1993. pp. 229-236.

[8] M. Fischer, Linking CAD and expert system for constructability reasoning. Proc. 5rh Inr. Conf on Computing in Cil,il and Building Engineering, Anaheim. Calif., June 7-9, 1993. pp. 1563-1570.

[9] R. Fruchter, M J . Clayton, H. Krawinkler. J. Kunz and P. Teicholz. interdisciplinary communication of design cri- tique in the conceptual design stage, Proc. 5th Inr. Conf. on Computing in Ciri l and Building Enginrering, Ana- heim, Calif., June 7-9, 1993, pp. 377-384.

[ lo] H. Fuyama, K.H. Law and H. Krawinkler, Computer assisted conceptual structural design of steel buildings, Prx. 5th In[ . ConJ on Computing in Ciri l and Building Engineenng, Anaheim. Calif., June 7-9, 1993, pp. %9- 976.

[ l l ] C. Harnung and H. Kimura, Three-dimensional FEM analysis with graphical interface, Proc. 5rh lnt. Conf. on Computing in Cil.il and Building Engmeenng, Anaheim, Calif., June 7-9. 1993, pp. 1283-1290.

[I21 D. Hartmann, A. Fischer and P, Holewik. Object oriented modeling of structural systems, Proc. 5th Int. Conf. on Computing in C i d and Building Engineenng, Ana- helm. Calif.. June 7-9. 1993, pp. 78-85.

1131 C. Hendrickson and D.R. Rehak. The w:entlal virtual construction site for automation plannlns and analysis. Proc. 10th Inr. Svmp. A~ctomarron and ~~h~~~~~ in Conslructron. Houston. Texas. May 13-26, 1993. pp, 511-517.

(131 R. Huang and D.W. Halpin. Dynam~c interface stmula- tlon for construction operation (DISCO). Proc. 10th /,li

S\mp. on Auromorion and Robotics it1 Construction. Hous- ton. Texas. May 23-26. 1993, pp. 503-510,

1151 T . Khedro. P.M. Teicholz and M.R. Genesereth. Agent- based technology for facility design software Integration.

Proc. 5rlr Inr. Conf, on Compurrng m C r ~ i l and Brtrldr,rg Engmeenng. Anaheim. Calif., June 7-9. 1993. pp. 385- 392.

(161 R. Kunigahalli. Computer-integrated concrete placement planning and control system. Thes~s presented to the University of Wisconsin. Madison. WI. in partial fulfill- ment of the requirements for the degree of Doctor of Philosophy. 1994.

1171 R. Kunigahalli and J.S. Russell, Towards CAD/CAC: integrated computer-aided design and constructlon. Proc. Elecrncal Power Research Insrrtute 3rd Inr. ConJ on Fossil Plant Cons~rucrron, Palm Beach, Fla.. October 26-20. 1993.

118) R. Kunigahalli. J.S. Russell and M.J. Skibniewski. Mo- tion planning for automated construction, Proc. IOtlr 1111. Spmp, on Aurornarron and Robotics m Consrructton. May 24-26, Houston, Texas. 1993. pp. 407-414.

1191 W. Lipski Jr.. An Otn log n ) Manhattan path algor~thm. Inform. Process. Lerr., 19 (2) (1983) 99-10!.

(201 L.Y. Liu and P.G. Ioannou, Graphical resource-based object-oriented simulation for construction process planning. Proc. 5th. Int. Conf. on Compurrng rn Crlrl and Buildrng Engineenng, Anaheim. Calif.. June 7-9. 1993, pp. 1390- 1397.

[?I] M.T. Neggers and M.A. Mulert. Interference detection with 3-D computer models. Proc. 5th Int. Conf on Coma purrng in Civil and Building Engineerlng. Anaheim. Calif.. June 7-9, 1993, pp. 1179-1282.

[23] G.S. Orenstein. The 3-D approach to design. Proc. 5th Int. Conf. on Comprcring in Cicil and Building Engrneerlng. Anaheim. Calif.. June 7-9. 1993, pp. 51-60.

1231 F.P. Preparata and M.1. Shamos, Compuratronal Geome. try; An Introduction, Springer, New York (1985).

(241 KF. Rcinxhmidt, F.H. Griffs and P.L. Bronner. Inte- grated engineering. design. and constructlon, I. Con- struct. Engrg. Manag., 117 (4) (1991) 756-772.

1.25) J. Reymendt and J.D. Worner, Object-or~ented modellng for concrete structures. Proc. 5rh In[. Conf on Cornputrng in Cii.il and Building Engineenng, Anaheim. Calif.. June 7-9, 1993, pp. 86-93.

1261 U. Ruppel. U. Meissner and B. Moller. 0bject.oriented data exchange for the integration of design processes In structural engineering. Proc. 5th Int. Conf on Cornpitting in Ciri l and Building Engmeerrng. Anaheim. Calif.. June 7-9. 1993. pp. 94-10],

[27] H. Samet and R.E. Webber, The Quadtree and related

R. Kunrgahalli. J.S. R~ltseN/Aurornarron In Consrrucrion 3 (1995) 327-340

h ~ e rarch1c31 data structures. .4CiU Compul. Sun.eys. 16 ( 2 ) (1981) 187-260.

1181 R.J. Schsrer and P. Katranuschkoi. Architecture of an objcct.or~ented product model protor).pe for ~ntegrated b u ~ l d ~ n g d r s ~ p n . Proc. 5th In!. Conf on Compurlng m CII 11 und Blclldmg Engrneenng. Anahetm. Calif.. June 7-9. 1993. pp. 393-400.

1291 R. Stouffs. R . Kr~shnamurti and 1. Oppenheim. Con- structlon process s~mulation with rule-based robot path plann~ng. Proc, IOrh Inr. Symp. on Auromarion and Roborrcs rrl Cot~trrucrion. Houston, Texas, May 23-26. 1993. pp. 195-502.

[30] R.F. Sullivan. lntegratton in design and simulation. Rm. IOrh. In(. Syrnp. on Auromar~on and Robotics in Coturruc.

I! rron, Houston. Texas. May 23-26, 1993. pp. 423-430. t r

1311 V.B. Viscomi. W.D. Michalerya and L.W. Lu, Automated ! 1 construction in the ATLSS integrated building systems. 1.'

I .: Proc, 10th Inr. Symp. on Artrornarron and Robotrcs in , .

, < Conslnicrion. Houston. Texas. May 23-26. 1993. pp. 9- 16. * -

1321 H. Werner. S.M. Holzer, M. Mackert and M. Stark. An a .. object oriented approach to computing with structural components, Roc. 5th Inr. Conf. on Compuring in Cicil and Building Engineenng, Anaheim, Calif., June 7-9.

. 1993, pp. 237-244.

A constraint-system shell to support concurrent engineering approaches to design*

Steffen hl. Fohn,' Arthur R. creel," Robert E. Younga & Peter J. O'Gradya "Grolrp for Ir~relligent S!,stclti~ in Drsign atid .\lanlrjart~rring, Dep.~rtt~ient oflndustrial Etrginrering. .Yorrlr Carolrnu Srore l'tii\.crsr!\

Raleigh. .Yorth Carolina 276957906, I.'SA 'IB.\I Corporotron. Infon,iarion techno log^. o ~ d Strateg,r. Personal Conrpurer Co. Aniericas, Sotjiers. ,Ye,( )'ark Inj85, ('3.4

This paper describes \.4TLRN. a powerful constraint-system shell that pro\.ides kno\vledge integration and reasoning features ideally suited to supporting concurrent engineering approaches to design. SATURN is a logic-based constraint modeling sbstem. tightly coupled with a relational database. and supported by a truth n~aintenanse system. This paper prescnts S.4TURN.s architecture and its modeling and manipulation teihniques. An example appl~cation in automated-storage-a~d-rctrieval-s)stem design is provided 10

illustrate how a S.ATURN constraint-based model is constructed by a concurrent engineering deslgn team. which then can tk: used by a product designer lo create a feasible design Unique features of this constraint-system shell incluie its rich knowledge rcpresentat~on fasil~ties. its seamless integration of forward. backward and relational database inferencing, and its engineer-oriented user interface.

Kej. ,c,ords: concurrent engineering. product design, constraint satisfaction, logic, logic programming, relational databases. automated storage and retrieval s! stems.

Today's con~pet i t ive market reqaires that engineering #:ompanics rcast quickly to volatii;. market demands and ?rowing prodllct complexity thr ,ugh efficient design processes rhat ensure a product's quality. co~npet i t ive price and prompt availability t o consumers. Success stories' have shown that this can be achieved by !clopring a design philosophy ter1nc.d concurrent engii;eer~ng (CE). This philosophy encourages the s i~nultancous consideration of all aspects of a product's I,!'c cycle at the design stage. It stresses a par;?:lel ;rpproach t o design that is different from the more traditional approach where products are designed in isolation and only then considered in terms of their r~ianufacturability, testability, quality. serviceability, e t ~ . ~ - ~ o n c e ~ t u a l l ~ , the CE approach to dcsign scems plain. Practically, however, when considering the life cycle of e\ en thc simplest prod~lct , the number of aspects to be considered at the design >ta?ge is quite formidable.

T o counter this problem, conipanies usually set u p design teamss-' hose members, drawn from all stages

'This ivork \\as funded in part by the National Science Foundarion, ;rant numbcr UDI\i-R9i4200.

of the product's life-cycle. are responsible for ensuring that their design rules a re satisfied (or at least considered) at the time of design. Design rules are derived from many sources including: tables found in manuals, standards established locall). nationally and internationall!, equations found in teutbooks. and heuristics based on expcrience and elpertise. St111 others a re based on the results of ~echniqlles like simulrition and proccss planning. Each of the \.arious rules cserts a constrairiing influence on the c\.ol\.ing design (our interpretation of d e s ~ g n rules is thus s jnonymous ~ . i t h that of c,onstraints) and scrtes to guide a desigller in makin? fcasiblc Jcsign decisions. As ~ : e ~ v information surfaces during escli step of the design the consrquences of the design dccisinn are chtcked against the design constraints t o enbure coi;forn~ity. If a dccision results in some o f the constraints no \ being met. the team members interacti\.cly co l labora~c to reso1i.e or minimize the potential do\\.nstrea~ii nlnnufacturing conflicts. This essentially results in an interactive constraint satisfaction p;,~~:css that i, best pc.~.foimcd if the team members t l ~ c ~ : ! ~ c l \ ~ e s are rcspi . sible for building and maintaining the constraint-b,\>.d models. As the design e\,olvc.s. nc\\, deqign-rulcs. procedures a ~ i d manuals are written so as to record nc\+ a i ~ d dr ~ t p d d t t d

using the cell references bl?=blO'b9'b8). Using the Justification window, a designer or an application b'uilder can trace through the complex interconnections among constraints and values.

Another window into the ATMS is the Violation window. Itsepurpose is to show which constraints are violated and which assumptions (designer assignments) caused the violation. For example, suppose a designer changes the value for the number of unit loads to store (cell b13. named nuls) from 15 360 to 16 000. This causes the constraint located in cell k2? to be violated since the number of unit loads to store now exceeds the storage capacity (cell hl5, named masnlrls) of 15 360 unit loads. The Violation window is shown at the bottom of Fig. 11. The violated constraint is shown as k22 in the Consrrainrs box and its form is shown in the Constraint box. The values that have caused the violation and the cells in which they are located are listed in the Assumptions box. A designer can use this information to 'fix' any occurring violations. A feasible design is obtained once all of the logic constraints have been satisfied. This is reflected in the constraint-sheet by the symbol, #TRUE.

7 CURRENT LIMITATIONS

Although SATURN was developed as a tool to support CE. its flexibility in knowledge representation and reasoning makes it quite generic in applicability je.g. see Ref. 2). The largest application built in our laboratory to date has c. 500 constraints and is used for printed wiring board (PWB) design. In the current version of SATURN the total number of variables and constraints for a model is limited by the constraint-sheet size of 2500 constraint-sheet cells. SATURN. however, allows multiple constraint-sheets to be opened simulta- neously and in the future, communication facilitated between the sheets will increase the number of allowable variables and constraints. Although the PWB application runs with good performance on a 486 machine, it is doubtful that significantly larger applications would execute in a reasonable amount of time on the same platform. Lastly, in this version of SATURN, the equation rewriting system in the Constraint Engine limits equations to a linear form.

8 CONCLUSIONS

A constraint-system shell, named SATURN, has been developed as a tool to support CE approaches to design. The architecture, the modeling and the manipulation techniques that form the basis of this tool have been reviewed. SATURN is a logic-based constraint' modeling system, tightly coupled with a'relational database, and supported by a truth maintenance system.

SATURN incorporates rich representation formalisms that accommodate both the large volume and the wide variety of knowledge required for large scale CE systems. Its' constraint engine is based on constraint propagation and constraint satisfaction algorithms that permit a designer to;approach a design from any number of perspectives. Additionally, it provides support for interactive constraint satisfaction and constraint' violation resolution during the design process. Integrating these features into the familiar spreadsheet-like user interface has significantly simplified model building. allowing developers to focus more on the model building process and less on AI-oriented. programming techniques. Together these qualities provide for powerful knowledge integration and the means with which to support the building and application of constraint-based ,models for CE. This makes SATURN an attractive support tool for engineers engaged in CE in design.

ACKNOWLEDGMENTS

The authors would like to acknowledge Ramdus Murali for his work in implementing the assumption-based truth maintenance used in SATURN, Steve Minnaar and KAre Christiansen for extensively Beta testing SATURN, and the reviewers for their valuable suggestions in improving this paper.

REFERENCES

I. Wheeler, R.. Burnett, R. W. & Rosenblatt, A., Concurrent engineering: success stories in instrumentation, commu- nications. IEEE Spectrum, 28(7) (1991) 32-7.

2. O'Grady, P., Young, R. E., Greef, A. & Smith. L., An advice system for concurrent engineering. Inr. J. Cotnpurer Inregrated Manul., 4 2 ) (1991) 63-70.

3. O'Grady, P. & Young, R. E., Issues in concurrent engineering systems. J. Design and Manu/., 1 ( 1 991 ) 17- 34.

4. Young, R. E., Greef. A. & O'Grady, P.. An artificial intelligence constraint network system for concurrent engineering. Inr. J. Production Res.. 30(7) (1992) 1715-35.

5. Chiesi, B., Parasida, T. & Walliser, E., Concurrent engineering at Boeing Helicopters. 47th Annual Forum Proc. of rhe American Helicopter Societja. Part 2, May 6-8 1991, pp. 921-30.

6. Meyer, S. A., Integrated design environment-aircraft (idea). An approach to concurrent engineering. In 46rh Annual Forum Proc. of the American Helicopter Societ!,. Part I (of ?), 21-23 May 1990. pp. 509-22.

7. Reddy, R., Wood, R. 7 . & Cleetus. K. J., The Darpa initiative: encouraging new industrial practices. lEEE Spectrum, 28(7) (1991) 32-7.

8. Adler. R. E. & Ishii, K.. DAISIE: Designer's aid for simultaneous engineering. In Conlputers in Engineering 1989. Proc. of the 1989 ASME Inr. Computers in Engineering Conf. and E.rposition. July 30-Aug. 3 1989, Anaheim, CA. Vol. 1, pp. 19-26.

9. Gopalakrishnan, B. & Pandiarajan. V.. Product design for

n ~ . ~ n u l ' , ~ c ~ u r ~ n y : the use of knowledge based systems in concurrent engineering. In I E E E Int. Conj, on Sj'ste~ns, .tftni (I I I(/ C.i~honleli(:f. Los Angeles, 4-7 Nov 1990, pp. 566-8.

10. Kott. A., Cederquist. A. & Kollar C., Product modelling as a foundation for a concurrent engineering system. In Proc. 2lsr Annual Pifrsburgh Conc. University of Pitts- burgh, Part 3, 3-4 May 1990, pp. 1289-95.

. 1 1 . Cutkosky, M. R. & Tenenbaum, J. M., Toward a computational framework for concurrent engineering. In Proc, of the 16th Annual Con/. of I E E E Industrial Electronics Societ! - IECOIV '90, Pacific Grove, CA, 27-30 Nov. 1990, pp. 700-6.

12. Serrano. D. & Gossard. D.. Constraint management in MCAE. In Arf$cial lnrelligence in Engineering Desigo. ed. J . S. Gero. Elsevier. Amsterdam, 1988, pp. 217-39.

13. Bowen, J. & Bahler. D.. Frames, quantification. perspectives. and negotiation in constraint networks for life-cycle engineering. Arrij. lnrell. in Engng, 7 (1992) 199-226.

14. Fisher. D. C. & Nguyen, T. D., Design and analysis aid for evaluating aircraft structures. Proc. 3rd In [ . Con/, on Itrdurrial and Engineering Applications of Artificial lntelligence and Expert System, I E A I A I E , Charleston, 1989, pp. 469-72.

15. Silverstein. E., Kaminski, J. S., Farenci. R. & Kumaran, A,. A knowledge acquisition tool for design engineering applications. Proc. 3rd Int. Conk on Industrial and Engineering Applications of Art$cial Intelligence and Expert S.rstenls. 1E64iAIE, Charleston, 1990, pp. 1167-72.

16. Greef, A,. Fohn. S. & Young. R., Implementation of a logic-based support system for concurrent engineering. LISDEM Tech. Rep., North Carolina State University, 1993.

17. Kim. C.. O'Grady. P. & Young. R., TEST: A design for testability system for printed wiring boards. J. Electronics .\lanuf (submitted). Kim. I.-Y.. O'Grady, P. & Young. R., The architecture of a concurrent engineering system lor rotational parts. Int. J. Srstems Auronration (submitted).

19. Oh. J . S., O'Grady. P. & Young. R. E., An artificial intelligence constraint network approach to design for assembly. I I E Trans. (submitted).

20, Colson. R. H.. Helman, D. H.. Gomes, B. A. & Cihula, J., An application of artificial intelligence in costing spread- sheets. In ?tld Int . Conf. on Industrial & Engineering Applications of Artificial Intelligence & Expert Systems, I E A A l E - 89. University of Tennessee Space Institute (UTSI), Tullahoma. TN, Vol. 11, 6-9 June 1989, pp. 558- 67.

$ I . Fohn. S. M., Liau. J . S.. Greef. A. R., Young, R. E. & O'Grady. P. J.. Configuring computer systems through constraint-based modeling and interactive constraint satisfaction. LISDEM Tech. Rep., North Carolina State University, 1993.

22. deKleer, J., An assumption-based TMS. Art$ Inrell.. 2.8 (1986) 127-62.

23. Murali, R., A nonmonotonic support system for an interactive constraint satisfaction tool used in concurrent en@ncering. Masters thesis. Department of Industrial Engineering, North Carolina State University, 1 992.

24. Reiter, R., Nonmonotonic reasoing. Ann. Rev. Compurer Sci., 2 (1987) 147-86.

25. Oberschelp, A., Order sorted predicate logic. In Lecture Notes in Rrfificial Intelligence: Sorfs and Tspes in Artificial lnrelligence, ed. J. Siekmann. K. H. Blasius. U. Hedtstuck & C.-R. Rollinger. Springer-Verlag. New York, 1989. p. 418.

26. Gibbins. P., Logic with Prolog. Oxford Lniversity Press, New York, 1988.

27. Enderton, H. 0.. A Mathemarical Introduction ro Logic. Academic Press. San Diego, CA. 1972.

28. Lloyd, J . W., Foundations of Logic Progran~nting. Springer- Verlag, New York, 1987.

29. Pfenning, F., Types in Logic Progranrming. MIT Press, Cambridge, MA, 1992.

30. Stickle, M. E., Resolution theorem proving. Ann. Rev. Computer Sci., 3 (1988) 285-316.

31. Genesereth, M. & Nilsson, N., Logical Foundarions of Artificial Intelligence. Morgan Kaufmann. Palo Alto, CA, 1988.

32. MacAllester, D. A., An outlook on truth maintenance. Rep. AIM-551, Artificial Intelligence Laboratory, MIT. Cambridge. MA, 1980.

33. Reiter, R.. Towards a logical reconstruction of relational database theory. In On Conceprual Modelling, ed. M. Brodie, J. Mylopoulos & J. Schmidt. Springer-Verlag, New York, 1984.

34. Gochet, P., Gregoire, E., Gribomont. P., Hulin. G.. Pirotte. A.. Roelants, D., Snyers. D.. Thayse. A.. Vauclair, M. & Wolper, P.. From dfodal Logic to Deductive Databases, ed. A. Thayse. John Wiley, West Sussex, UK, 1989.

35. Minnaar, S. & Christiansen, K.. SATURN users guide and reference manual. Tech, rep., North Carolina State University, 1992.

36. Tompkins, J. A. & White, J. A.. Facililies Plannitlg, John Wiley, New York, 1984.

37. AS/RS Assoc. (Automated Storage Retrieval Systems Association), Considerations for Planning and Installing an Automated Storage/RetrievaI S ~ s t e m . Material Hand- ling Institute, 1977.

38. Bozer. Y., A minimum cost design for an automated warehouse. Masters Thesis, Department of Industrial Engineering, Georgia Institute of Technology, GA. 1978.

39. Zollinger, H. A., Do-it-yourself guide to costing stacker systems. Automation, 1974 (Sept.).

: l rr i f~cra/ Inrzl l rgencejor Engrneerrng Desrpn, Ana l~s rs und . i l o n u . f ~ ~ l ~ r r n g (1995), 9, 161-IP! Prlnred In the LSA C o p l r ~ g h t 1 1995 Cambr~dge L n ~ ~ e r s l r ) Press 0890.0604195 S t 1 00 + 10

A framework for case-based reasoning in engineering design

PA-

C .

H . S H I i ' A KULIAR AND C.S . KRISHNASIOORTHY mi---T- -095 Department of CIVII Eng~neerlng, Indian I n s t ~ t u ~ e of Technolog), Sladras, 600 036, l n d ~ a E/~ 'z ' , :;;,,biG RECEIVED J l a ) 19, 1994; ACCEPTED October 18, 1994 LgI,.ansy

Abstract

.Although the czse-based reasoning (CBR) process is domain dependent, certain aspects of i t can readil!. be cap- tured into a gc:~eric frameuork uhish in turn can be applied ro various engineering domains. One such exercise that has been Lavied out is described hcre. In this paper, He present the notion that CBR can be formalized and applied in a specialized frameuork in an integrated knouledge-based entironment. We rirst analyze rhe CBR proce5s to abstract the steps iniolied in the de\ flopment of a CBR system. H'e then propose a framework in which most of these steps are formalized so that the) can be applied in a domain-independent manner. The salient features of this frdmework, called CXSETOOL (CASE-based reasoning TOOL-kit), are then described. The high- light of this approach is the use of a concept called design criticism in the CBR process. The versatility of the tool is demonstrated through an applica!ion froni the bridge engineering domain.

K e > ~ o r d s : Car--Based Reasoning; Engineering Design; KBS D.:\elopment Tool; Bridge Design

Engineering design involves wide usage o f domain- specific knowledge and considerable problem sol\,ing skills to come up ~ i t h a clear, concise, and unambiguous specification of the artifact being designed (Dym & Levitt, 1991). In engineering domains, the solution generation pr .xess is experience-based and highly iterative. Hence, de~ig i i experts hake recommcnded a study of similar cases for obtaining solutions at I ~ r i o u s stages (Heins & Lawrie, 1984). According to them, the study of similar cases gives a n approsimately good estimate of design parameter values to start with. hloreover, various feasible solutions can be generated at different stages of problcm solving (>la- Lcr, 198t , 1990). T o select a possiblc best solution, thcse feasible iolutions are then critiqued from ~ a r i o u s aspects (Shi1.a K i ~ m a r et al., 1394). However, this process is time consunling and i t is possible t o avoid going through the same escrcise if \ye know the history of past problem solving episodes. All these factors point toward the applicability of past cases in engineering design domains.

Analogical reasoning is a very efficient way t o use past experitnee. Corbonell suggested that analogical rcason-

ing is a cetltral injerence tnerhod in human cog~lrrion, and defined it as:

transferring knowledge from past problem ;ol\,ing episodes to new problems that share significant aspects with corresponding past experience and using the trans- ferred knowledge to construct solutions to the new problem (Corbonell, 1983).

Depending upon the problem definitiorl of past and current situations, either rransjort~~ational o~tolog~~ (Corbonell, 1983) o r deril-arional unalogy (Corbonell, 1996) can be used for solving the current problem. /lnalogical reasoning and design dependent reasoning ( C ~ a i g Howard er al., 1989) can be used in the models mentioned above to guide the sdarch through shorr and apprc\p! inte paths in the search space. Rcasoninp based on a . i n ~ i l ~ ~ . pas! problem- solving experience helps the designer to exploit tile useful details specific t o a particular similar case. This problem- solving strategy is termed as case-bawd reasoning (CBR) (Schank, 1982; Kolodner, 1985; Schank, 1956; Riesbeck & Schank, 1989) and is based o n the premise that human reasoning processes a re founded on specific experience rather than a set of general guidelines. It rclatcs a current

- - - - . - - - -- - . . - situation t o the closest specific experience in memory and Reprint requests to: Dr. C.S. Krishnamoorthp, Dcpartrncnl of Civil

Engineering Indian Institute of Te~ \~ . lo logy . Madras, 600 036, India. uses that esperience t o solve the problcm at hzrqd. Thus, Fax: 01 1.91 u-235-0509. ( h e central idea of CBR technime in problem sohing is

H.S. Klrrtlar and C.S. Krishnutuooriiy

(1993). Krloulcdgs based sbstcn with generic rools for s:ructural engineering de,ign Strucrural Enartleering Rerrew S(2). 121-131

Lehncrt, b'. (1937) Case-bascd prohlem sol~inp ui th a Isye knouledds bare of icdrncd cater. Pruc, h'a110na/ Con/. on .d l . 301-306.

Jlahcr, J1.L (1957). Enginec,ing design s!nthcsis: .A domain 1ndepc.n. dent ~mplen~cnrarion A1 EDA.\I, / ( I ! , 207-?I!.

Jlahcr. 31.L.. (1990). HI .RISt and bejond: D i re i t~on i for expert sys- in dc;i<n. Cotripir!rr Aided Desrgn, 17/91, 420-127.

3lahc.r. 31.L.. & Zhang. D.JI (1991). CADSYS: Using c a m and de- compoiii on knot+ ledge for design synthesis. Artrjicral InreNrgzcce rn De5i:n (Gero. J.S., Ed ), pp. 137-150. Butrer\borth-blcinemann, O\forJ

Jlaher, X I L . , & Zhang, D.31. (1993). CADSl 'S : X case-bad de,ign pr0ce.i model. 41 ED. i . \ I . 7/2/. 97-1 10.

S a \ ~ n s h a ~ : d r ~ , P (1998). cab^ + a d reasor 73 In CYCLOPS, !,:sign p r o b l r : so! ..:. Proc. D.4Rt!4 U orkshop orl Cuce.B,. ..,I t?<../jon. ing. !So-301

Pu. t ' (1993). Inrroductior~: Issues on ca5c based tecign s!jtem, 41 t l1 .4 I f , 7/21, 79-86.

Ric!beci. C.K. . b. Sshank. R.C. (1989). Inside Case.B~scd Rzasonrr: L au.tri ie Erlbauni. Hillsdale, h J .

Rod.::: ~ n . S . , S Tsa~soulis, C (1993). P.4SD.A: .A ~3se.based system- to 41 nob i z designr:i. .dl ED.4, i f . -/?/. 123-133.

Rosc:,nlan. \I .A, , Gero. J.S., 8: O m a n . R.E. (IW!). \\'hat's in a caje: The use of cast-bases, knouledge.b3~cs, and daraba~es In des~gn. C.4.4D Fici.~res '91, (Sshm~tt, G.N. , Ed.), pp. 285-300. \icweg, \\'ierbadcn.

S~hank, R.C. ( 19S2). P!natnic .~iettror.v: ..I Theor! o/Learnrng In Com- puters and Peopli Carnbridgc L'nrbrrsir) Press. Ncu Yo r l .

SihsnL. R.C. (1986). E.rplu~rarron Parrerns: L'nderstondrn~ .Jlechunr. ral!, and Crearrri!i Laurence Erlbaum. Hillsdale, XJ.

S h ~ i a Kumar. H.. Sureth. S.. Krishnamoorth!. C.S., Fenies. S.J.. ii Rajeeb. S. (1994). GENCRIT: .4 tool for knouledge-bared critiqu. lng in engineering des~gn. A l ED.S.ft, 8/11, 239-29

Simon, H. 4 (1973). The structure o f ill-structured problems, Arrd i . o o l Intellrgmce. 4. 18 1-201

Zhang, D.J1., & Jlahrr, h1.L. (1991). The transformational process in case-bad reasoning in design. Proc. I JCAI - I2 W'orkshop on .-I/ rn

Design (Gcro. J.S. Sudueeks, F.. Eds.), pp. 32 1-336. Sydne). .\us. tralia: Dcpartrnenr o T ~ r c h i t ~ c t u r a l and D t y n Scien~e, L'ni\c.-,i[) o f Sydney.

Zhao, F., 8: \ lahtr, M . L . (1988). Lrsing analogical rcaioni;; ro d:jign buildir~gr. Eng~necr rn~ wl:h Compri!ers, 4, 107-1 19.

ti. Shiva Kuniar is currently a Ph.D. scholar in the Civil Enginccring Dep,;rtment at the India11 Institut: of Tech- nolog) , Madras. His research interests incluLl debign criticism, case-bascd reasoning and integrated kno\rlc.dgc- bared s!j[r.ms in general. He received his J!E degrec i n Struct~;ral Engineering !'rom Bharatiar University, Coirn- batore in 1990 a n t a E Tech degree in Civil Engineering from Jawaharlal 'ichru Technological L'niversity, Hydc- abad in 1958.

C.S. Krisl::~~moorthy is Professor of E:ructur~l Enyineer- ing in the Department of Civil Engincuing and Dean o f Academic Research at the Indian Institute of Technology, hladras . H e received his Ph.D. degree from the Univer- sity of London in 1972. H e obtained his B.E ( H m ) and hl.Sc. (Engineering) degrees from the Univeri ty of Ma- dras. His current research interests include Adaptive Fi- nite Element Analysis, Structural Optimization and K n o ~ d e d g e Based Expert Systems. H e is a fellou of the Institution of Structural Engineers, London, Fellou of the Institute of Engineers, India and Fellou of rhr' Indian National Academy of Engineering.

A parametric design assistant for concurrent engineering

- - . - . .. . . . - -- - - . . - - - - .

D.-ISIEL KUOKK.4, ' S T A N L E Y J E F F E R S O N , LEE BARFORD ASD FELIX FRd4l'\1.4?J Heuleri.Pa;kard Laborarories. 3500 Deer Creek Road. Palo 4110. C.A 9430.1. C.S.4

( R E C E I I E D 313) 2 , 1994, A C C E P T E D S o \ e m b e r 1 1 , 1991)

Abstract

The E\plorer parametric dcsign assi1:ant. an inreracti\e tool that proiidec intelligent support for searihing concurrent-CI.; ,.erring trade-rpazcs under multiple, conflicting objciti\es, is described. The s).stcln proiidcs a conienient n:c,inc for speiif!ing c . ~ltiple, cross-di*;iplinar! constrainrs in terms of ~ables , formulas, and logi;al c c n ! ~ ~ n < ~ s . B a v J cl11 these data, thc j! stem perfo, :'r ~nteracti\e constraint checking, iomputes feasible designs, and proiidec praphiial anal!.sis facilirics, alloi\ing users ro colnpare designs based on multiple criteria. .As a first application. E\pl< rer has been used a. a printed circuit board (PCB) construcrion design assistant. In initial tests, Elplorer ha! he!;-'.d users to find desrgn ~onfigurar ion~ nor pre~iourl! considered that yield comparable perlor- niaiiie and cost hile offering better manufazturabilit! and reliabilr~!. The capabilities and use of the E\plorer s!,sicm are desirlbed in detail, the underlying techilclopies are ou~lincd, and an e\.aluatron of the protot!pe s!,s- tern is p!t,en~ed.

lie!uords: Concurrent Eng~ncciing; Constraint Sa:isfaction; Human Computer Interaction; Intr.llipcnt .4ssistanr; Trade Spaces

1. IhTRODL'C'TIOS neither possible nor sufficient. Engineers \I itb specific

\\'hen the entire life cycle of a product is considered, man! differ:nt, often conflicting, factors must be included in its design. Not only must the product satisfy [!pica1 design requirements such as heing fast, small, and light, i t must also be eff~sicntl!' manurac t~ l rab le , reliable, repairable, and disposable. Unfor tu~~ate l ! , each of these factors is, in itself. a n entire discipline, a n d typical design etigineers simply cannot be expe,ti.d t o have all the requircd expertise. This has led t o t?ic ad\,ent of Loncurrent-c~iginceririg approaches such as Integrated Product Dc\ cloprnent ( IPD) realns. These focus on bring- i n t a su:'fici~~ritl!~ diverse set of .;nginr.crs togcther earl!, in tlie desi: . phase so that all I~fe-cbcle issues can be

knoi4ledge a re of ten not available \vlicn needed, and the interactions a m o n g the disciplines lead to a trade-space that no single engineer understands. Both of these problems can be addressed by creating a tool that captures aspects of each engineer's knowledge and combines that kno\vledge into a single trade-space search assistant.

\\'e have created a tool , called Explorer. that sup- pons intelligent, highly intcracti\,e searchcs of \ \\ncurrent- engineering trade-spaces bnder conflicling it!)jecti\es. Elplorer combines the use of automated c L ~ r m ~ t r s i n t sat- sfa action algorithms and a flesible graphic :! L.i;:~;an interface to yield an intelligcnt assiztant. Sprtcifi\ 'I I ! , E\pIcxr'r supports the follo\\ing major capabilities:

considered. 0 .4 declarati\e Lno\\ledge and Ja ta format In tc.~rns o f

E \ ~ I I though I P D is a n important step in inipro\ ing \4 hich engineers from ~ a r i o u s d~sciplines i an .i)c'clfy

product do<- lopnien t , there a r e m a n y cases \$here i t is general constraints and metrics for thelr do9:ialn.

- -- - - . . - - - - - - - - - -- - This forllis the static Lnonledge base of a specific

R c l ) r ~ n l requesrs lo Dr l ) ?n~e l R EuoLLa, I.oclhr.rd Palo 4110 Re- E\ ylorer application (e.g., for tlie dorilain of p~ ~ n t e d ccarch 1 abs. 0 96.10. B ?:-IF. 3251 Hanoier Srrcel. Palo 41to. C.4 clr~*uit board desien). - . 91304, U . S . A . Phone: -ll5-!c1.5?9l; fax: 1l.c.354.5235; E-mai l : An interacti\.e interface by \vliich an iridi\ idual de- d r k @ aic loclhccd corn

I Crrrrd~i! a f i i b ~ ~ ~ o r i : I.oiLhted Palo A l i ~ l Rcsca~sh Labs , 0z.96-20. signer can enter p robkm-spcc i f . i~ rcquirc~nc.nts and II 254F. 32.21 H d n m c r SIIXI. Palo A I W , CX 91301, U . S . . 4 . properties (e.g., for the spccific circuir b u r d treing

ceptunl dejigr~ of satr.I!::es and bicycles have sho1r.n promise as E ~ p l o r t i applications, h!oreo\er, since Explorer is, in effect, a gciicric spreadsheet tool, its applicability goes b q o n d engintering. Howevcr, the benefit t o concurrent enginecring is tangible and clcdr. Explorer permits designers to gain a better unclsrstanding of multidisciplinnry trade-spaces, and it c r~codes aspects of each I P D team member kno\ i ledgt so individual designers can bcnefit froni the teani's e\pr.rtise even when separated b), space and time. This allo\r.s thc conlples and often conflictin; clenler.;.; of rriultiplc cnginccrs' expertise to be integratid into a ;ingle, uell-inforn:td design deciiion.

\\'e u o u l d l ike to a;k rjoiiledge the suggestions and commenrs .. o f Norrnan Chang, c.,pc.cially his board construction dorndin analysis and data f rom IPDX. I n addit ion, ue thank Jani.c Bradford for her ;ontrihutions to the B C A + project and her usabilit! stud) o f Explorer, 'BC,4+. Finally, !\e are grateful for the support and asslsranse o f B i l l Brown, Scott Conradson, Rand! Coterstone, jhane Fazzio, Sreven Greenbaum, Terr i l Hurst , B i l l Shrebe, and Bill Tani.

Bouchard. E. (1991). Concepts for a future airiraft desrgn enklronment. Pror . .Aerospuce Desrgn Con/. Number .41.AA-92.1 I S P . .American Ins~~rute of .Aeronautics and hstronaut~s,, IVashington. DC.

Chang. N., Chang. K.. Leo. J.. Lee, K., 8: Oh. S. (1991,. IPD,A: High- 3pc:d lnterionnect performance design assrstant. Proc. He\c.I~.r:. P lL .kord Design Teci ino iog~ Conjerence (Cornpan! ionlidcntiaii Hc\rlett-Pailaid, Palo Alto, C.4.

Cllne. T.. Frayrnan, F., Fong, W., 8 Katz, E. (1990). A 1001 for derrgr. for manufaaurabiiity of printed circuit boards. Proc. H r ~ l e r r - P ~ l k a r d Desrgn Techno1og.v Conjerence (Cornpan! confidrntral). Hculi i t-Paclard. Palo Alto, CA.

Fr~!man. F.. 6 \ l i t ! . t l , S. (1987). Cossack: ,4 constraints-based expert 3)jtr.m ior iontiyuratldn tasks. In Knowledge Based E.rprrr S,rsre~rrs ln Enytre.ering: Piturning and Dargn (Sriram, D.. and Aicy. R.. Eds.). Computat~onal Jlechanics, Inc., Billerlsa. JIA.

Holdcn. H.. Tam, B., Rosers, T., Henderson, P., 8: Parrish, C. (1989). Thhz DFIl . t lan~rul, Proro!)pe 4 . (Company confidential.) Technical Rcporr. Heuleti-Packard Printed Circuit Dikision, Palo Alto, CA.

Kolb. 11. 11989). In i ts t~gat~on of consrrainr-based component modeling for knoulcdge representation in computer aided conceptual des~gn. Techn1ca1 Report. Dcpanment of Aeronautics and Astronau- tics. Jlajiachujctts Institute of Technology. Cambridge, M A .

Koualsk~, R. ( 19iY). Logic j o r Probletn Sol r ing. Else\ ier, North Holland.

KuoLLa. D.R. I 1990). The deliberative intesration o f planning, evesu- iron. and learnrng Ph.D. D~ssertation. School of Computer Science. Carntgie Jlellon L'niversity, Pittsburgh. PA.

Kuokla, D., 8: L~\eze), B. (1994). A collaborati~e parametric design agent. .Sarionai Con/. on Arrficrul lntellrgence .A441 Press. Jlenlo Park, CA, pp. 387-393.

Kuolka. D., Jefferson, S.. Frayman. F.. Conradson. S.. Chang, N., & Bradford. J. (1991). Toward a design information framework: A de. slgn optimization assistant. Proc. Hew1er1-Packord Desrgn Tech- nologr Conference (Company confidential). Heulerr.Packard, Palo Alto. CA

\ la i i \ tor th, A. (1992). Consrraint satisf3irion. In E r i c ~ c l o p t ~ f ~ a o j

.-1rrrjicrul lnrelligence (Shaprro. S.. Ed.). Val. I. Johq \\I]<! Sonc, Kz\r York, pp. 285-293.

Ullniin. J.D. (1982). Pritic,i: :a of DarclbuseSjsrctfrs, 2nd Fd. Cornpursr Sc~enie Press. Ro ikb i l l~ , \ID.

Dan l iuobka is a Rcjearch Scientist and Project Jlaria:cr at the Lockheed Artificial Intelligence Center in Palo Alto, CA, wherc' he leads projects investigating intelligent, agent-based a rc l~ i tc i tu res for d i~r r ibu ted ~ o r k group. (SH.ADE) and interaitive visual interfaces for kno\\lcdpi. based sp ten ls . Prior t o joining Loikhecd , hrb \ \ 3 j a Re- search Scientist at Hcbi le t t -Packa~~l Labcratorit5, \( here he worked on portable i n f o r m x i o n a p p l ! ~ n s e s 21iJ tntel- ligent, inttractive assistants for engineering. He holds a Ph.D. degree in coniputer science and a n 51.S. Je21i.e in electrical engineering from Carnegie 5lellon L'nibe~,ity, and B.S, degrees in both computer science and cllcctrical enginetring from the Unii.er>ity of N'ashington.

Stanlr) Jefferson is a member of the Technical Staff at Heulett-Packard Laboratories. He holds a B.S. and XI..A. degrees in mathenlatics from the Unhersit!. of California, Davis, and a Ph.D. degree in computer science from the University of Illinois at Urbana-Champaign. His research interests include programming language semantics, visual programming, computational reflection, and intcract i~e, automated reasoning.

Lee Barford tarried a n A.B. degree in computer science in 1982 from Temple Uniiersit}, a n J1.S. dc.:ree in cornputer science in 1985 from Cornell L'nhersity, ~ n 3 a Ph.1). degree in 1987 in computer science from Corliell Uniier- sity. H e has been a hlember of the Technical StaZf at Hewlett-Packard I aboratories, Palo Alto, C A , since 1987.

Felix FraJrnnn is a member of the Technic,~l Staff at Heiflett-Paikard Labora:~ries , whcre htconducts research in the areas of design for manufacturability, economics of manufacturing testing strategies, constraint-based reasoning, simulation modeling, and compleuity theory. Prior t o joining H P he worked on knowledge-based systems applications for configuration and design at Xerox PARC. In addition, he has extensive consulting esperience building application soft\vare systems for the industry. H e received the equivalent of a B.S. degree in applied mathematics in 1978 from Yaroslavl State Uniiersity in Yaroslavl, Russia, a n M.S. degree in computer sciences from DePaul University, and Ph.D. degree in computer sciences from Northwestern University. H e is the H P representative on the ANSI C o m m o n Lisp standardiza- tion committee a s well as a member of the AAAI, I E E E , a n d ACbI, and is a member of the editorial board o f A1 EDAIL!.

Knowledge based floor plan design by space partitioning: A logic programming approach

L. B. Kovacs

Department of Computer Science, Universip of Copenhagen, Universitetsparken I , DK- 2 100 Copenhagen 0 Denmark

This paper describes the incremental development of a knowledge based system for supponing floor plan design. A method of gradual space partitioning is developed which allows large flexibility in expressing demand, including highly incomplete descriptions. A simplified logic programming language is used to provide a readable and precise description of tasks, concepts, methods, regulations and other knowledge. A variety of architectural constraints are used to express demand, regulate solvability and improve efficiency of reasoning. A new lexicographic method is embedded in the system to increase time efficiency and convenience of use. The s!,stem is implemented in Prolog. Detailed experimentation with reasonably realistic design tasks are shown.

Key Words: logic programming, knowledge representation, concept modeling, demand modeling, space partitioning, lexicographic rules, combinatorial explosion, efficiency of reasoning, reasoning on constraints, rectangular design, architectural space planning.

Recent advances in knowledge based systems technology promise new hopes of complex support in areas like architectural and engineering design, requiring deep knowledge and experience in a variety of subjects, high skills and strong reasoning with the - often inexact - conflicting goals. It is no more only the raw computer power, which can take over annoying routine parts of the job, but also an intelligent assistance in all kinds of human mental activities such as reasoning concept formation, knowledge interpretation, creation of completely new methods, modeling and solving ill- defined problems. The degree of automation of the design process however should be controlled by the human designer according to the task, the present phase and hislher confidence in the present system action. Open system, open minded human designer, good exchange of views (explanation) and learning from good results and mistakes both automatically and consciously are important assets of such a fruitful 'cooperation'.

Floor plan design is an architectural space planning activity determining subspaces of a given space according to various more or less defined or implicitly understood requirements and conflicting criteria. Typical problems of this category are, the floor plan of an apaRmen1, a family house or one storey of an office building. A facility layout in an industrial environment may lead to a similar problem. The requirements may include bounds for the sizes or restrictions on the shapes and locations of rooms, adjacency of rooms, orientation of windows, lighting conditions, convenient internal traffic. etc. At first, the floor and the rooms will be considered as rectangles. This condition will be relaxed, the

( 1991 Elhev~er S c ~ r n c t Publ~shers Ltd

floor and the room will be compositions of adjacent rectangles.

The family of rectangular'design problems is quite large. I t includes various cutting or placement type of applications with specific constraints, such as the cutting of metal, glass, wood, textile sheets into smaller rectangles. Numerous Guillotine6." and non-Guillotine cuttingJ algorithms have been developed, including one of a network type7 and a general flexible tree search method26. There are many different approaches to the automatic generation of floor plans as well: cataloguing of small rectangular plans for further processing'. various graph r e p r e s e n t a t i ~ n ~ . ~ ~ , ~ , exhaustive generation and abstracti~n". '~, optimization in focus3' and many others". These systems have been implemented using imperative (procedural) languages like Fonran or Pascal.

Logic programming plays an important role in intelligent systems and also in the present paper. I t can be viewed as a language, which can be used to reflect our meaning, intention, knowledge in a readable, yet sufficiently precise form to lay the basis of a computer system specification. Furthermore, the clear logic can support various algorithmic interpretations, and is thus very flexible in use. Fortunately the computer language Prolog inherits this double (logic and procedural) interpretation and contains additional features for controlling execution and other functions like internal and external communication. Many books have been written on logic programming and P r~ log~~~ '~ '~ . " . " .

Floor plan design by logic programming has been carried out24 using some rules for matching predefined rooms to form a floor. A series of papers by Gero and coyne 10, I 1.20 also use predefined rooms and certain

162 Artificial lt~telligence in Etlgineerirrg, 1991, Vol. 6. No. 4

iser =igure A.l) the

,, maximum shaking targeted value. The :es.

t e the input link. The

~ i l ab le in Smith and ialnic analysis are : synthesis involves

*e requirements are

Enq Opr . 1994, Val 3 , pp 25-46 C 1994 Gordon and Breach Sclcncc Puhl~rhrrs S A Repr~nts aballablt d~rectl\ from the publlshcr Pr~ntcd ~n htala! ,IJ

Photocopjlng prmltted b) Ilcensc only

A METHOD FOR FORMUL.ATION OF THE CONCEPTUAL DESIGN OF NON-FOSSIL

FUEL FIRED POWER PLANTS

E n e r y ! Erigineering it~sritute. Sat1 Diego Slurt> L'nirersity, Snn Diryo, CA 92182, U.S.A.

I Rrceiced j u l \ 2 I993 i r~finalf i~rnl .\o~<nrher 16 1 9 9 3 )

I n th~s paper crlicrla for efficient and optlmal des~gn of~hermal s!s!ems are prcscntcd The maln cnlphascs of the paper IS In the problem set-upand In the structunngof the lhermal s)stem design ~nforrnatcon dur~np thc conceptual deslgn phase A two-levcl dengn structure (macro-leiel and m~cro-leiel) 1s ~ntroduced to enable thc dcs~pncr to a d ~ p l h~s deslgn to the aiallable lor ~cqu~rcd~technolog~ leiel. I)pe ofapplica~~on,econorn~c fdctors, and O & V requlrcments At the macro-leiel of deslgn. eccirnmlc leaslbcllty (buslncss) dec~s~ons are made. uhlle at the m~cro-level oidesrgn, tcchnlcal feas~bcl~ty ienpncerlng) dec~s~ons are made A 50 hl\5 uood burnlng pouer pldnt In Northern Callforn~a 1s presented as an cllustrat~ve case stud) All the lnlormallon proildtd In the paper 1s from real uorld pouer plant dectgn. and the problem 1s presented In a form to allow other researchers to try dlfrerent optcnl17atlon techn~ques to solie the problem

KEY WORDS. Poucr plant. optlmlzatlon, system design, conceptual deslgn

I. BACKGROUND

h'on-fossil renewable energy sources (e.g. wood. refuse. trash, wood residues. pulp, etc.) offer flexible, affordable, locally available, and sustainable alternatives to fossil fuels' '. Wood, a renewable non-fossil fuel (commonl!. referred to as a bio-mass energy source). is one of the most widely used energy sources around the world. Throughout history. ~vood-based fuels have been the principal source of energy utilized by man9. In the past. most of the wood residues generated from the forest products industry have been either burned without energy recovery or disposed of as solid waste. However, in the recent past. there has been an increasing interest in utilizing these residues to gcne- rate poNer, especially in the Pacific Northwest N here forest residucs are abundant9.

Globally, an average of 25 percent of the wood entering the timber industry is available for conversion to energy, for industrial. commercial and utility applica- t i ~ n s ' ~ - " . In the United States the percentage is higher at about 33 percent. Wood- fired electricity generating facilities can provide significant contributions to the power supply. Since 1983 four wood-burning power plants have been built in the U.S., with 121 M W installed capacity8. In the development of a 11ood (or any non-fossil fuel) fired small power plant (5 to 100 M W) using a "standardized" power plant design will not be appropriate, because of the uniqueness of the fuel. and the environmental effects of the fuel before and after processing. Moreover, plant locatlon and system configuration are very much fuel dependent. Therefore, an efficient design method is needed for small power plant design and implementation.

sures as the basis. The following improvements to the model can make the design process more thorough:

Including other subsystenis (e.g. condenser) into the optimization; this will allow .for studying a more complex interaction between the subsystems. Including a more sophisticated heat balance program (e.g. Ref. [34]) that can handle technical data for the subsystems, and the interactions between them. Incorporating electricity sales price, ROI models, and life cycle costing (LCC) into the economic analysis.

Work is underway at the San Diego State University (SDSU) Energy Engineering Institute, to implement these improvements as well as to increase the current knowl- edse of interactions, and couplings, among thermodynamic simulation, engineering decisions, and business decisions in power plant design.

SDSU's overall research effort is aimed at developing design/optimization methods for optimal design of thermal energy systems. In particular, methods combining thermal simulation models with mathematical optimization techniques, and a unified (generic) designioptimization perspective applicable to any thermal system, ranging from as small as a residential air conditioner to a large power plant, are being developed. The universal objective function is made up of maximum efficiency and minimum cost (initial, life-cycle, etc.).

Design trade-off studies for various types of thermal systems (e.g, cogeneration, thermal energy storage, HVAC, passive solar, parabolic trough concentrating collec- tor, small power plants, etc.) are carried out in developing the universal (generic) thermal system design/optimization perspective.

The author is thankful to Mr. Donald Lark. Managing Director of SDSU Energy Engineering Institute, for h ~ s help in the project. and for the countless hours he spent with Mr. Javccd Mohamed who worked on a 51. S. thes~s on the subject. Also, thanks are due to Dr. Farrokh Mistree ofGeorgia Tech for his continued insp~rational support of the author's research program.

1 . Probert. S. D., Kerr. K. and Brown, J. (1987) Harnessing energy from domestic, municipal and industrial refuse. Applied Eneryy, 27,89- 168.

2 Klass. D, and Colleen, T . S. (1987) Energy from waste. Chemical Engineering Progress, 46-50. 3. Alter, H. (1987) Material andenergy from refuse: Trendsin the United States. U.S. ChumberofCommerce.

I (ash~nyron, D.C., Elsevier Science Publishers, 29-38. 4. Pollack, C. ( 1987) Recycling: A man-made energy source. Sun World, 11, (3). 76-80. 5 Phillips. V. and Takahashi, P. (1989) Renewable energy development: It has many environmental

benefits. Enrlronment. Science. Technoloy!. 23, (I), 10- 13. 6. Jacobs, S. (1988) Municipal waste-to-energy marketplace to e x m d S2O billion during next ten years.

Srrarryic Planning and Energ). .\lanagemtsnr, Spnng edition, 55-57. 7. Editorial: Energy from refuse. (1987) Applied Eneryy, Elsevier Applied Science publishers Ltd, England,

83-87. 8. Shea, C. P. (1988) The promise of renewable energy, Part 11. Sunworld, 12, (2). 38-45. 9. Smith. K . R. (1987)The blofuel transitton. PaciJjc cmd Asian Journalof Energy ,Tata Maraw-Hill Pub. CO. Ltd. 13-32.

10. J ~ l a n , R. K. ( 1986) Energy: The biomass option. Chemical Age of India. 37, (S), 479-481.

I I . Goodman. G. T. ( 12. Frankena. F. (198

electric power. B ~ c 13. Lorber, K. E. (198'

Consert~arion, 14, : IJ. Hofer, P, and Hal

range covering all 15. Adam, P. J , and t

ul i / i :a~ ion conferen 16. Eckart, L. E. and t

CliRs. NJ. 17. Li. K. W. and Prid, 18. Tnbus, .M er al . ( 1 $

from sea water. Re1 19. Frangopoulos, C. P

Improvement of ct Georgia.

20. Von Spakovsky, M deslgn and improvc Atlanta, Georgia.

21. Dadkhan.N~koo,A deslgn of a combin The Winter Annual 1986.65-70.

22. Guven, H. M.(1986 systems. Compurer- hleeting of the Arne

23. Guven, H. M., Mistr for developing coun

21. Guven, H. M., Mis: troughs for direrent 60-66.

25. Bascaran. E. (1990) designing for concc Houston, Houston,

26. Struble. C.. Bascarar cal design olspacecr San Diego, CA.

27. Bascaran, E.. Bannc conceptual design. L

28. Bascaran. E.. Mistr multi-object~ve then

29. Muster, D. and Mis l r~rernarional J O U ~ I I L

30. Neurnan. V. and Gu Journal of the ASSOC.

31. Kuppuraju, N., Ittin Design Srltdies. 6. (21.

32. Mohamcd. J. (1991) hl. S. Thesls, bfecha

33. "STEAM BAL", Thc 31. "THERhl". Asea Brl 35. Boehm. R . F. (1987) 36. El-Wakil. M. (1984) 37. Kok. M . (1987) Ene

Policy, 10, ( I), 2 1 -40 38. Ahmad, S. and Linnt

Computer A ~ d e d Enq hlechanical Engineer

39. Rhyner, C. R. and H blunuyrmmr and Re:

ake the design

.ti-on; this will allow tems. Ref. [34]) that can s between them. 2 costing(LCC) into

Znergy Engineering : the current knowl-

nulation, engineering

Jtimization methods methods combining

iques, and a unified nal system, ranging

der plant, are being .;mum efficiency and

s (e.g. cogeneration, concentrating collec-

universal (generic)

y Engineering Institute, for Ltohamed who worked on

la Tech for his continued

:. mun~ctpal and industrial

76-80. has many environmental

In during next ten years.

:r publishers Ltd, England,

, 3 8 4 5 Tata XIcCirau-HIII Pub.

CONCEPTUAL DESIGN O F POWER PLANTS

I I. Goodman. G. T. (19871 B~omass energy in dcveloplnp countries: Problems and challenges. ~ ~ b ~ ~ . 4, I ? . Frankena, F. (19871 Rethinking the scale of b~omass energ!. conversion faci l~t le~: The tax of hood-

electric power. Biomuss, Elsevier Applicd Sctence Publ~shers Ltd. England. 149- 171. 13. Lorber. K . E. (1987) Analys~s of household waste and measurement ofthermal emlsslons. Resourcesand

Conserr.arion. 14. 205-223. 14. Hofer. P. and Haussmann, 0 . (1987) Thermal power plants from Brown Bovcri-A comprehensive

range covering all needs. Brom.n B o ~ e r ~ Rel,tew, ( I ) , 4- 1 1. 15. Adam, P. J. and Hollrah, R. 1. (1984) Plant sizes for the future. 7th lnrernarional coal, lignire and

uri1i:arion con{ermre. Houston. Nov 13-15. 1984. 1, 33-55. 16. Eckart. L. E. and Weisman, J. (1985) hfoder~~ Power Pluirr Engineering. Prentice-Hall. Inc Englcwood

Cliffs. NJ. 17. Li. K. W. and Priddy, P. (1985) Power PIattt Sysretn Design. John Wiley and Sons. 18. Tnbus. M. er a/. (1985) Thermodynam~c and economic considerat~ons In the preparation of frcsh water

from sea water. Report N. 56- 16, UCLA. 19. Frangopoulos, C. A. (1983) Thcrmoeconomic funct~onal analysis: A method for the optlmal des~gn or

improvement of complex thermal systerns. Ph.D Thes~s. Georgia Institute of technology, Atlanta, Georgia.

20. Von Spakovsky. M. R. (1986) A practical generalized anal!.sisapproach to the optimal thermoeconomic design and improvement of real-world thermal sjstems. Ph.D Thests, Georgia Institute of Technology, Atlanta, Georgta.

21. Dadkhan-Nikoo, A., Ranasinghe. J., Bushnell, D. J. and Rcistad,G. hl. (1986)Computer simulation and design of a combined-cjcle wood-fueled power plant. Co~~rpurer.Aidrd Enginerring of energj S ~ s ~ t n t s . The Winter Annual hfecting ofthe American Society of hleihanical Engineers, Anaheim, CA, Dec 7- 12. 1986.65-70.

22. Guven, H. M.(1986) A multi-level design structure for computer-basedoptimal design of thermal energy sjstems. Conrpurer-Aidrd Enyinerring of Energy Systems. I'olurne I-Opfintr:orion, The Winter Annual Meeting of the American Society of Mechanical Engineers, Anahelm. CA.

23. Guven. H. M.. Mistree. F. and Bannerot, R. (1984) Design synthesis of parabolic trough solar collectors for developing countries. Engineering Oprin~irarion, 1, (3). 173- 194.

24. Gwen. H. M., Mistree. F. and Banncrot. R. (1986) A conceptual basis for the design of parabol~c troughs for diferent design environments. JournulofSolur Et~eryj Engineering. AShlE Transucrion, 108, 60-66.

25. Bascaran. E. (1990) A conceptual model for the design of thermal systems: Concurrent decisions in des~pning lor concept. Ph.D. Dissertation. Depanmen~ of Mechanical Ehgineering, Univers~ty of Houston. Houston, Texas.

26. Struble. C.. Bascaran, E., Bannerot. R. and hlistree, F. (1989) Computer-based multiobject~ve hierarchical dcsign of spacecraft thermal control systems. I9rh Infersocier)~Con/prencr on Encironn~enralSysrrms, San Diego, CA.

27. Bascaran, E., Bannerot. R. and Mistree. F. (1989) Hierarchical selection decision support problems in conccptual design. Enyineering Opr~mi:ar~on. 14, pp 207-238.

28, Bascsran. E., hltstree, F. and Bannerot. R. (1987) Cornpromise: An erective approach for solitng multi-objcct~ve thermal design problems. Engiitcering Oprintiration. 12, (3). 175-189.

29. Muster, D, and Mistree. F. (1988) The decision suppon problem technique in engineering design. The It~fc~rnarional lourr~ul of Applied Enginrering Educarion. 4. (1). 23-33.

30. Neuman, V. and Guven. H. (1988) A simple ranking mcthod for choosing HVAC systerns. Engineenng, Jolrrnal offhe Associarion of Energj Enyinerrs, 85, (2).

31. Kuppuraju, N., Ittimakin, P. and Mistrec. F. (1985) Design through selection: A method that works. Drsign Sfudres, 6. (2). 91 - 106.

32. Mohamed. J. (1991) Design optimization of non-loss11 fuel fired small to medium sized pouer plants. hl. S. Thesis, Mechanical Engineering Department, San Diego State Univers~ty. San Diego, CA.

33. "STEAMBAL", Thcrmal Analysis Systems Company, 725 Parkview Cir., Elk Grove Village, lL 60007. 31. "TIIERM". Asea Brown Boveri (ABB) Power Automa~~on, Connecticut. 35. Boehm, R. F. (1987) Drs~gn Analysis of Thrrrnal Sjstems. John Wiley 8: Sons. Inc. 36. El-Wakil. M. (1984) Pouer Plant Technoloyy, hlcGraw-Hill. 37. Kok. M. (1987) Energy modeling with multiple objectives and multiple actors. Energy Sysrums and

Policj. 10, ( I ) . 2 1-40. 38. Ahmad, S. and Linnhof, B. (1986) Supertargetinp: O p ~ ~ m u m synthesis 01 energy management systerns.

Con~puter Aided Engineering of Energj S~.srems. The Winter Annual hfeetlng of Amencan Society of hlechanical Engineers. Anaheim. CA, 1-1 3.

39. Rhyner, C. R. and Wenger, R. B. (1986) Capital costs of resource recovery facilities in the USA. Ij'usre hfanagrmenr and Re~earch. 321-326.

APPENDIX A: OPTIMIZATION STRATEGY: COMPROMISE THROUGH SELECTION DSP

A.1 Ocerciea

In order to select the optimum cycle design, a selection DSP, which uses the quantitat- ive (hard) information from Table 3, along with somequalitative(soft) information (e.g. reliability), must be developed. The formulation of such a DSP is presented below. Full details of a selection DSP can be found in Refs. [24,29-311. Only a brief overview of the particulars of the model are presented here.

A.2 Fon~iu lnt ing a selection D S P

The steps for such a selection DSP are:

r Generating alternatives, r Screening the alternatives, r Identifying feasible alternatives, r Weighing the attributes, r Giving attribute ratings to the alternatives, r Evaluating the merit function, r Ranking the alternatives, and r Post solution analysis.

Generarlng nlrerttorices. The list of alternatives for the case study plant were based on altering the extraction pressures and temperatures at the turbine, and on the number of feedwatrr heaters in each configuration. As a result fifteen configurations of the power plant were evaluated. These are listed in Table 3 . Scree~iitig the nlrenlnrires. Any infeasible alternatives selected at the previous step, are deleted from evaluation. Identi/:riny frczsihle alrernarires. This step involves furnishing information and specification to all alternatives. This step involves assigning weights to attributes in the order of their importance. The criteria (attributes) used in this model are shown in Table A.1, along with their relative importance. The four atributes are: initial cost, operation and maintenance cost, reliability and efficiency. The ranking is done from the most important to the least important in ascending order. Assiyni,ig a t t r ih t~re r a t i ~ ~ y s to alrernarices. Rating values for each alternative can be derived using different types of scales, viz. ratio, to interval and ordinal scales. An ordinal scale can be used when the attributes are qualitative in nalure.

Table A.l Relative Importance of A~tributes.

.irrriburrs Rlmkiny Heiyhrs h'ormalrrc'd I\'rltyl~rs - - ~~

In~tial Cost 2 3 0.3 0 % M Cost 3 7 0.2 Rcllabilitg 4 1 0. I Efficiency I 4 0.4 Sum 10 I

When using : of the samc mag

When a smaller

I Tables A.2 and I alternatives. 1 Ernluoring the r

evaluate the mt attributes toget frequently used

where, m = nurr n = nurr

'Ij = rela Rij = ratir

M F i = valu Ranking the alre

' after the merit f l

one ( I ) corresp unsatifactory s) optimal system Posr solution at validation and

Table A.2

AH P BHP CH P DH P EHP AMP BStP CMP DXI P EXlP ALP BLP CLP DLP ELP

' Rellabll~l) lnd~cate gre. rel l~hie In I r

:h uses the quantitat- f t ) information (e.g. esented below. Full

I brief overview of the

elant were based on .id on the number of

lrations of the power

~ v i o u s step, are

).nation and specifi- ributes in the order

- shown in Table A.1, : I cost, operation and

ne from the most

h alternative can be ' ordinal scales. A n

ure.

When using an ordinal scale, the attributes are normalized, so that all attributes are of the same magnitude. When a larger value of an attribute rating represents preference,

Ri j = (A,, - ATi")/( A y ' - AYi") (A . 1 ) When a smaller value of an attribute rating represents perference,

Tables A.2 and A.!, show the attribute ratings and normalized ratings for the fjl,e

alternatives. Ecaluaring the r~~erir/ur~ction. The next step in solving a selection DSP is to define and evaluate the merit function. A merit function combines all the individual ratings of attributes together using proper weights or importances defined earlier. The most frequently used model in evaluating the merit function is a linear model,

where, rn = number of alternatives n =number of attributes

I , = relative importance ofjth attribute Rij = rating for alternative i, attribute j

,If F i = value of merit function for alternative i Rollking r l~e alterr~trrices. The alternatives, are ranked from highest value to the lowest, after the merit functions are evaluated. Ranking is shown in Table A.4. A merit score of one (1) corresponds to a perfect system and a score of zero (0) corresponds to an unsatifactory system. According to Table A.4, configuration B with low T & P is the optimal system design. Post solurion anol~s i s2 ' . Post solution analysis of the selection process consists of validation and sensitivity analysis. The latter involves sensitivity of the solution to

Table A.2 A ~ ~ r i h u t e s Rating lor [he Alternatives'

A1rcrnat1r.e Coprral Cost. 0 & .W Cost, R e l l a h i l ~ t j Efic~enc) .Jf.\fS M M S Rating O / ' 0

AH P 17.53 0.3506 5 29.72 B H P 18.01 0.3602 4 32.62 CH P 18.56 0.37 12 3 32.90 D H P 19.04 0.3808 ? 33.18 E H P 19.60 0.3920 1 33.45 A M P 16.61 0.3322 10 28.30 B M P 17.11 0.3422 9 3 1.00 C M P 17.63 0.3526 8 3 1 .23 D M P 18.01 0.3KJ2 7 3 1.46 E h l P 18.62 0.3724 6 31.61 ALP 15.82 0.3164 I5 27.37 BLP 16.25 0.3250 14 30.90 C L P 16.75 0.3350 13 30.13 D L P 17.18 0.3436 12 30.3 1 ELP 17.69 0.3538 - 7- 30.36

' Reh~bl l l t ) rallnps arc bawd on turblne throttle pressure and tempcralurc of thc c>cle Lower ratings lndlcalc greater r c l ~ ~ b ~ l l l ) It IS assumed that cycle operaline a! lower Tk P will lnhcrentl) be more rellable In 11s long term o p r a u o n

Table A.3 Normalized Attribute Rat in~s ., .4lrernarrce Coprlul Cosr, 0 & At Cost. Rrliabilrty Eficiency

MMS MMS % AHP 0.55 0.55 0.29 0.39 BHP CH P DH P EHP AMP BMP CSl P DSIP ESIP .ALP BLP CLP DLP ELP

Table A.4 Merit values for the Alfernatives.

Alrrrntlrive ~ktrr ir Rank

AHP 0.502 12 BH P 0.579 6 CH P 0.5 I8 I 1 DHP EHP AMP BMP CMP DXlP EhlP ALP BLP CLP DLP ELP

changes in the attribute weights and sensitivity ofthe solution to change in the attribute ratings. Post solution analysis is described in detail in Refs. [29-311. Post solution analysis is very important due to the nature and quality (hard and soft) of the informntion being used. This step involves two parts, namely, the validation of the solution and determining the effects on the solution of small changes in the relative importance attributes and also the changes in the attribute ratings. The goal of validation is to ensure that thedesigner feels comfortable with the computer solution. If there are any subjective misgivings these need to be resolved by modifying the ratings and, or altering the formulation.

Eny. Opl. . 1994, Vc Reprints available Photocop)lng per:

SCH

CHUNG

' Acer .

This paper deals w lines, each of whic

I bottleneck, slnce 11s supplied by the bot (or setup) tlme for : make as few change processlng~obs for. ~dle Thus. a sched~ busy w ~ t h few chap at~onofthe lofa1 prl the opttmal schedu

K E Y WORDS: Sc

INTRODUCT

Consider the sc Each of the n 11 common facilit: capacities of thc be re-designed ! the latter stater common facilit; a whole.

When the bor a non-product: supplied by the i t is desirable to However, if the inventory at the to determine a s

I* Ship Definition System in Computer Integrated

1 Design and Manufacturing

DelmrLa~leuut of Naval Arc:Li trc:llrrr ~ l r r l Orcnu E~~giuecriug Faculty of. Eugivrering, Uvivrrsi LY ol Tol~a, . ,

I Introduction

Evcrlr t.lrort8l1 C:AI)/CA.\I s!-strnis Ilat*r 111~dr n rrma~*l;lrl~lc pro8rms ill tllc s t ~ ~ r a/

(Ir.*igt\ a ~ i d rr innt~hct~~rirrg ill rr II i t rd~~s~r im, i t is o i~r t r snit1 t l ~a l~~c la l~a rn till& cr~git~ccri tr~ works arc nrcdcrl t.a managr ('Ai)/CA,CI systcna.

111 r c r r~~ t . years, the t~rc~ssi ty or r0111p11tt-r i~tt#cgraLrd ~ ~ \ a ~ ~ i ~ ~ a e t t ~ r i t \ h gystrttr I \m 1w.n rcco~lrixcd in .Japarme ship!-ards. ' f ' l~c Japarrnc sl~ipljttiltlitrg inrll~stry orgnnirctl A j)mjer.r to makc fcasihility s t ~ ~ d y of ronlpukr integrated rlmign m r l nranlrfarr~~rilig in I!lMi. I'rod- ucl modr?lit i~ a11 ri cxpcrt a!*stc~~rr rvcrc! ta kc11 as tlrc 11rt~t i ~ i l p a r k ~ \ t tltclnca ill tilt prajcct in order tfi ir\tcgrat$ C.CAI)I(!Ai\.I. syaten\,

, . I bs ailn of tl3; papo is to pmpov onr r x r a p l r of prnrlltrl i l r l i~t i t inn rptrlit or r tltilr (1. 2, 31. Ol>jcrt ori~trkcd cat~rrp l is ltrrcrl to rcltrmt-~It sl~il, slr~lrtttrm. . 11r Llw dmigtt a t r p , two typm of a l r j r rk arc i~~trocltlrrrl. Otlcr is parr aljjcct u.lrirll s ~ a t m . . gmmctrical propertin and its sttrilrt~tm. I lrc otlrcrr is rotincrtion a l t j n t \vlricl\ states cou~rect~iou rclatiolr at part olrjrct A and H. Flllirtio~ts OC r l ~ t t i t ~ g II rlmignrrl S~~III-tlrrc i i r . .

to pieces alrcl f t ~ ~ ~ c t i o ~ i s of asae~iiling tllmc ~)iwc?a 'to ntiiadt~lc arc! also i t ~ t r w l ~ ~ m l l~rrc. 111 tlie fitage of a.trc~irlrling. two otlrcr oljjcrtas arr i~rtradctrrd. Facillity ahjrrr i s one

of them. And operation ol>jccr is tlrc atllcr. h r facility aljjrcl. lift-niagt~rt.. ra trrnsfcr a sniall strttctl~rrr, cranes n r ~ l \vclrl i~~g ccl~~il)mcrrta arc ro~izirlcrrrl hrrr. I f the s j ~ t r t ~ r lias data 011 cal\acitica of tllrrcc farilitics. tl\c systrrlt sllawa Its flow 1na11y Iio~trs i k takr+tn rlo tlic opcrnt.iotr. 111 this st.ngr. Iwacrns l ) Ia~ i~~ i t~ ; a~rd s c l i c d ~ ~ l i l t ~ ACC rrisn tiiarlr a t tllr same

- ' ..l: :.c.., . . . . . ,. ,. : <,' : 1. . . . . . . . . System Design of CIM for shipbuilding

i j i

, , . . ,

! K Doi. H Hot% hL ~ a b a i , Y. Nyue, K. much i ,

, . . . . .

m.......... 8 ' .

I

ship d Oxm'foundarion, Technical Cornmince on CIM for: jfi#wil&nl ' .-.stian .tccrs+; . . . 15- 16, Tomomon, 1 -Chomc, Mb~llt+hl. TOkyO, 10s; J i p . . - r c : ... c ~qlud* u s . . a . . .? a \

Abstract ' :r*dncously anc J:'.,:: : 'Ihc Ship dr Ocm Foundation hu c&ndusdng feasibiil&b&Wa 'ob: b d uublLckiag 1

concep&b-or CTM for shipbuilding. Now we 28d finished h i p i n g 1 -4

. concept to develop the pilot model which consists of pmducr model system. $&but ; s e a c3n s ~ s m n s and user interface. and have been pmtotypin~,Lhcm to sNdy fhc by --- ; uchnologies in h e rquircmcnu for rho CIM for s h i p b u J 6 u ~ l h h pw in~&uccs &

'. ..... , . '- : , . '

, findings obtained b u g h kc processes study *Am 2 . I ....... . ,, I ' l .

\ In = o a t y e w . there is r gowing h a n d for compu& 1 n t e g k d h u f a c ~ g fm : shipbuilding (CIM for shipbuilding) to enhance the productivity and to hprove the Lbor. . ,

intensive krdusuy [I], : Under rhc circumswccs. the Shipbuilding'Rescush Association of Japan made r

' fundamental survey on he method &d possibiliry to eonrrmct the CIM for shipbuilding in its 'Study on New Generation Technologies for Design. Roducuon and Maintenance of Ships' (SR2101, and proposed that i f was. cffcnivt to study the key technologies by

, developing a pilot modcl phor to the comprrhensive dtvcloprnear (2). Bawd on the SR210 proposal. the Ship k Ocean Foundttionorpanized Technical

Committee on CIM for Shipbuilding" in 1989 and is now promoting a three-year pmjcn on the research and development of I pilot model prior t o the development of CIM for shipbuilding. By studying the following mrjor~te;hnical subjects lhrough the nseuch and &vclopment of the pilot model. the Cornmirtee aims 10 offer fundamental materials to rhipyards with which hey can decide whelhcr they should sm Bc development.of CIM for shipbuilding.

The Cornminee complctcd chc design of the pilot model in 1989 [3], developed a d evaluated asprototype of pilot model based on the derign documents in 1990 141, and is funhn promoan'p research and development ro make a comprehensive evriuation this ycu. This paper. inuoduies the findings obuincd through the processes ro s~udy the subjects (product model system, data b w system, upm sysom, user interfact and object-oriented/ protoryping) in the key ttchnologies.

4 . .. . . ' . . I

Scheduling System of CIM for Shipbuilding Appped Roduct Model, Process-Equipment Model and Operation-Resource Model '

muumua Nakmms, Kuutoshi H o ~ ~ U C ~ L - . T i k a b Minemum, Technical Committee on CIM for Shipbuilding, Ship k O c m Fwndrrion,

, , . # - . - . , I . , , ,: :,,.. f',.? . ; l a , < - "-, .; ; . Q

Abstract . . ". 'it it vy Nacdt to consauct a cdmputct system in the lhipbuilding indusay when rll

work commcnca upon the ieceipf of an order for r ship. In the pilot modcl of h e Computer- Inregrated Manufaauring Syswrn for shipbuilding, a -, which is r focus of our recent project, we have construclcd subsystemr which E c n r c p u u -.- i n r n a r i i n .. ind &ermine. -.. .. work sequence. Cambbgi~hof h e P e l in which the dur dtveloped by B e - subsysttmr is packed with a&e~~quipent -----. madtl witb UI o~cntion-rtjource model alIows for timely scheduling with computer. The Uata fmm the - . scheduling system -- is lnregnted so dtar it can be uriliud contirtenrly down to the final process coneofin Ihc Lorkshop. The prototype scheduling and procut conwl sysum which incorparaw above functions i s inuoduced hue.

1. Introduction

At the present time, thc manufacturers that produce a variety of products in small quanrirics have made in heir possession dl the relevant informadon (process planning, work load, etc.) before commencing to mmufactrrre newly developed producr They can with wc make long-term schcdults by csdmating the number of the producrs, and make working schedule accordingly,

In the shipbuilding i n d u q , however, when produco an ofun uilor-made, almost no infonnadon exists at the inilia) nage d design because infonnrdon is developed for each individual order, Even at r)lc i n m e d i a t e - m ~ , b e informatton-adablc is still incomplete, so pmeu scheduling has to be handled by veeran staffs. For thcse rwons. shipbuilders have encountered many obstacles in introducing the computer-aided scheduling method. 1)

The ship and Ocean Foundation has been engaged in a thne-year project that staned in fiscal year 1989 involving rhe research and dtvclopmen~ of r pilot model for CIMS.

For the pilot model of CPUIS, the subsystems to determine the paxu and work squencc were consaucrcd. The data that is output from these subsystems, shows each individual operaaon and process u wcU as all opquions in the process m g e d in a hiennhy. Then by using the data the scheduling system enablu one to establish the optimal work load wd I d dme for each unit, intarrnediats prcduct and final product by process-equipment model. 2)

Moreover. the sirnuladon using the operaaon-resource model which identifies &tailed work throughput. the number of workcrs.,&d rht equipment capacity in a pmcus, including

cem#rr n Olr A u w W 01

Sh1py;lrd ~ r a o o n 8 d YIq Ckrrp, fl

c- - -. Ovrrp I(uo ICU&I

Process Planning System of CIM. for Shipbugdin* 1 1

Technical Cornmirace on CIM for Shipbuifdin& Ship Br OcunlFoun&cian, Senpku Sh& ~ldg. , 1.5- id T a a m i o n , M L U ~ O ~ , TOIC~O, ILU, ~lprn .

. , , < 1 ; 1 ,c ;psy;,;,,::.!,.! ,, <; ..-j .

Abstract Thc'ppr proposcs rhr pmccss pluming system for shipbuilding. It &es out pmcus

planning design by use of both the d a i @ information depcndtnt on the product and the factory infomalion independent of it. Therefore i n u r n o n of &sign and production ..'- *-a. - ' .;.-,,.,,. - -. , . . , i n f ~ q n ~ i i ~ ncccssuy for genendon of wable pr#css plrantng. , ..-

The rim of chis paper is to propose both how b e infannation should k represcared and how the knowledge for proccu plannbg,duiga should be described systcmaidy.

Regarding h e p p o d . i n f m s d o n rcprrsenmrioa as r rnodd, the - model -,.. ... .... i n c l l n g - ,, dl . .,. infomarion for rh~roduc t ion ( wba'i' ir ul lcd' '~rodytm& ) is proposed f m m ihc - - . -----... viewpoint of process planning. It can k flexibly applied to the itvision of rhe prudun design (conccprud design, preliminary design, and derail design) even u procers piaaaing stage.

This papa describes the ESt a n of t h e ~ - J y ~ - c s ; - ~ l a n n i n ~ .---... .I- e g s u ~ system with infmnce engine and vuificodon of iq tffccdvenesr md pncdcabiljry. . .-

1. Introduction

Proctrs planning means desi,ening the entire producdon pmesses, such as production , merhods, production sequencu, the types of machines to be used, quanrily and pmccduns,

based on the conceptual knowledge from the product design to producrhn nrge concerning . products, pans, and materials.

In [he manufacturing indusmer, in gcnerd, machining tools md application ttchnologics haw been signir~crntly improving, most mess planning relies on the experience and p c p u o n of process planners and its evaluation iums arc various such u qualify, cost, delivery, etc. Therefore, a study on process planning chat allows cornprchtnsive evaluation hu been lagged behind. ?his i s also me for the shipbuilding industry.

In the recent shipbuilding indusuy, CIM for shipbuilding (CIMS) hu been cxpccnd as one of the approaches to be taken to uckle subjects such u reduction in lead cirnc, productivity imptovernent, and quality assurance. CXMS is also expected sr a means of solving various problems on the product stags such as a dccnue in the number of skilled workers, ~d a ftlative decrease in labor force popqlrtjon. And a research on CIMS has been

A NEW APPROACH TO DESIGN SYSTEM FOR SHIPBUILDING - A kralodqo-h8.d kru-Dmriqn S y a t . ~ Lor Yull kmccuza -

ib. Ship L P o u d ~ t i o a h s k.a -crkrg f v r j h i l l t y -dim. aa. ud o r t r b l i r b h g & concoptud , f w d Lor O H f o t 8bipbuUdhg. Ikv p m j ~ hr flai-d baiqoing a to -&*.lop tb e. *+ - r i m 08 irr JLO-d . y . t ~ e ~ ~ ~ ~ t l d l y . , & has b.d --q a& t r a u l a i s~ctk.l

hriqa s y r ~ u i r on. # p k m d r . ThLa prpr -pa- tha sub-ryatra having m a fvrretian t o k bpluord en rb. #trues& dealon r y n u . t h o r y r t u i r , ro dl&, L h .rp.rs ay-v M d *-tor up.rr hwldg. tht . m t r u 8 t i a a oa sh ip l r darignmd trrdng -0 dosign conorpr from L h m a U a t urd typLul conrtruertos. TUs r y g u .opliad t o rb. t n n m r r a * a d o n h bnld put of oil . . =&a: =C -Aa - s 3 t n k&kz:a -A% ' 2 2 Cstts f a rmlicrblr t o S.te:arr zk* dosign f a f o m t i e n ovm u r U o t .nd output f h a l n fo ra r t ron n e c m ~ r r y t o t h e

1. z?rmmCTIQR t o b o t r en & h a l t o f mxiat ing CAD/CUt In recant ymars, rbotel i m a growirrg

damn4 for . Conpot a t . I n t a g r a t e d Il.nufaccuring f o r SJdpbulldLng (CTLI: f o r Shipbui ld inp) to a c h i e r - h i q b r r producr iv i ty and t o ' i r p r o r m thm labor- , i n t a r i n h d u w 111 .

Ondmr the re c i rcor runemr , tho Shlp r Ocorn Foundation foraod tbm Technical bdttn on QH fo r .hi--, &d i s nov oanatructtng p i lor m d d f.r m for Shipbui ld ing t o r m r i f y t h a a p p l i e r b l o tochn&eal subjeeta .ad t o emrabli8b tho frumvork f o r -8 concepe t&ougb r year remaarch .nd dmlopnnt p r s j e e t21.

One technical aubjrcr of tbf8 project i r t o h r a d n e tho p o a 8 U t y ol M o p i n g

OyWtur but a.0 wOpa8 jwr & 8 i & n q md g.enrazor fb. doriqa Liorutioa -.a

mar l l ax than t t L a . Thr projmct m i l l eva lua t a uhotbar r y 6 t u natuork crs ba a o a s t m d to prorid. *aUam plraahg i n f o - t i o a Cot r c h a d u i h g , proomsr pl-9, aad Il.tuiJ R . e o k r r n r Planninq (IIAPI, wbieA is not p o s a l b l r uaing c o n r a n t i o a r 1 CRP/CAn r y r t o u . md co ganorato h tagr r tmd M e r v c l a n availabla Cot fm h i @ r pndboaocr aotLvlty.

h +h. d n c sib, ubon a d u - u u i l l make drau ing o f a p r f i o f h u l l aaanntcrlorr. & n h r it b r m f d g rh* ~ ~ l l - r l ~ b 0 ~ 8 t 0 d . a i d l a r and typAcrL con8eruorian. Zh. lthd 1. m i l d narai-

r s t ruc tura l &sign r y r t r that not m l y dosign, nou middy umod Lor huU otmctuxrl mk.8 &awing# m d gonaratmr n 6 2 i a dariqn uork. thr y r t o m b rhicb t h i s f o r WC c u t t i n g u c h i a r o . n d / o t umldlng mmthod i s lnco rpo ra tod can rsvma

SYSTEM INTEGRATION OF MANUFACTURING 'MANAGEMENT FOR SHIPBUILDING

The ship md Ocmm Touabtion baa k o n conducEhg f u 8 L b U t y muator on, U B ~

. . . r U s b h g a e~c.prurL.frnmuark fo r CIK for shiphrUdiaq.. . . - .. . . . nai r papu diaauwme8 a htoqxat lnq rhm r y r t w u concernhq ruaufac turhq b

C M f o t sNphuilaLog - R ~ n m 8 s R L d g , S o h d d b g a d D i 8 p a t c b b g / ~ t o r i a q uirh efffcfmntly toaurer iur iag ma& othmx by the uao of uoiqum product m&1 - r o pr~vidm a p t W m d coarirrurt Lrfanvrlon and dm~. i l .d hmtructionm tor o p a r i o n , and a lso t o u u b l e r d f ud rod- t o r u i f t l y rmrpoad t o d i r r h c . by mxtrnul ruaoa .

tn rh ipbu i ld fnq indurt y , one s h i p

should be b u i l t and & l i v e d h ao lmrr thur one year aftax r p m d f i u l l y orduad. Jn addit ion, f t facluckr w n y pzocearor

rmgiaq from u c h now produet desigaing, p ~ d u c r i o n plurninp. f abricatiorr procar 8

from mrtrrbl, t o tha amsadly of huge and onormous prodac ta . This a r p e e t of r h i p b u i l d i n g i n d u r t y make# t t f a i r l y dl: f i c u l r t o improvm tho product ion p lanninq and con+ra l , and f o r c e r ahipbuilderr t o depwd on #ki l led k r k e t a ( c h o i r axperimncmr and know-how) t o rupplommnt inmfi lcimnt &La.

Soma mlabora ts and sompreh~n8lve manufacturing aanagmmmnt, t h a t , is, production p h n n h q a d control, however, i r v i t a l t o t u n h e r davmlopmnt o t r h i r

i n d u r t y i f it rhou ld h aaaomplirhad wi thout h i g h e r c o n t r i b u t i o a i t o r i t a rlrll lod *et)rc-. b . ~ ~ # o t& l.n&aty l a cbractrrizrd by i t s bimr f l u l b i l l t y in d r u i l e d w n r t m o n op.raclon8 uich broad r o r t r i e t i s n r imposed by l a r g o - r e a l e f a c i l i t i e r a s dookr and r o on, t b m

offoct ive production pluming and conttol mat bm worked our qu i t8 d i f f e r e n t from e t h e r i ndu r t r ima auoh aa au tsmebi la u n u f ae turmra whore rho product ion p ~ ~ ~ a d u ~ * ~ . a m t .LtLy d t ~ l b d i n to a nuabmr of apecia1 and a m r e n t r u p a .

Wm r p r t LLll f b m qual i ty of prOcbction planning, ror8 decailmd lnrtxuarlonr f o r opmratiaa fn .stud wrkplacmr, l t r r u l f t responsa with eon8 l s t . ny to any Ppbco In deripir, which La unavoidable in p u a l l a l work of hmigr and p.oduetion pkanlng o r t o rhm change of product ion planninq

A Proposal for the NextGeneration Shipbuilding Systems

* ship r 0ea.n ToundrLiOa , Taabaiul -tw cp OM ior S U W d & g

:, $ (., . - . I

' , 4. .I y

xay vord r for WpbuUding, r u . t - h t r t i a ~ ~ . , ,++a, ,. *1 - % . A .

d l &phth hgurm. u r w u r d . r ~ l 8 t ~ d '

The W p 6 &ern Ooundatiocr has hm eondurtCry fuuW+r atudiaa en, m d errablirbfng r c m u p t U f ~ ~ w o r k for CIH f o r .htpWAaLog. P# out p r o j u t hr f i a r d d daaiqnkrq the wneopt ro h a l o p tha p i lo t u U & , , c ~ s l r r o t iu ex

kamaZ 8ystaor f u n o t t o m f ~ y , and h a b-a p t a t o t y p h g . @a@ ,fruoel;. Ulc f a t 'I

. , SbiphUdLng, ' re uU CW, ,.!a d 8 o a , cmoopr . i o r n u r t v m u a t - .bLphild&q 8yrt.am. fbat h a r m powarful functions t o ruppsrr. UrU8 p h y s i d y 4 u n r r l l y . Thir paper proporer r concept o t u8.r-oriantad ay r toa f o r CIW, .ad umaful tunczionr rucb ysrmn n d .

I .

Stnca AprAl 1989, t h e rbip 1 Ocean F o U n d r t i ~ h&S kU! g w h g f u 8 u t y study t o l ap l rnan t rha coacrpt of CIM (Coapotat t n r eg r r rod b u f actuxiaq f o r

UJpbuilding), and L rrp& to 8-c r f h l propoaal by 1992. [ l j , [ l ]

CXMS r a f a r r to .n a u t o u t a d drai9n

mymtaa urad t o e f f i c i m t l y bui ld rhipa. Thi8 8y r tm ht.p;nrcor vuimaa l d o n r t i o n of daaiqn m d .production vlng corpora? notworkr t o obtdn fbe required qurnt i ty .nd quality of dau rhmavar nrceawry.[ lJ ,

I2 1

c r e a t i n g damiqno, davia ing p l a a r and impAernnting mbnaqemnt, the systom inca rpo raca r l a i t i & A d a t a qene?&rlnq f u n c t i o n and zaLi.8 on varAous mxpext ,

8y.r- t o 9.rut.t. h i g n dau u ld t o a u q i n r t ~ c t l o m m . am we l l a r aehedullnq Inf or s r c ion . It, h o u e v ~ r , would bo d i f f i c u l t to complata product8 witbouf cocapiling w e d t o UII initi.1 au b; d humso 01ytJwi ty m d a r d l fo-f, bacauoa 8hipboiLdots would a t i l l m k o b s i g a r a d produetion pluu f o r u c h .hip f o r ooch or*, mmn o n b r tin r y r t r la mls.

Tho rhfpbui ld iap a y r t a u , theref or*. rhould bm urer-es1mr.d by placfng g r a a t u iqaorranca eu improvinq productiwity by f u l l y eonriderlnq e v o r a l l 8.a. 'of uaot intorfaem and r y r r u opararioa. CIUS l a u l t i n t d y domigad f o r uaor-axienurion, whiab i r c ~ n s l d r r o d rho maat i=porr .nt c r i t a r i o a i n eon8r tucc ing t h a mart- qonoration mhipbuildlng s y a t u .

AN EXAMPLE OF STEPS TO MODERN SHIP PRODUCllON BASED ON PRODUCT MODEL TECHNOLOGIES I . L 0.

, . . a . . ... Fosrichi Kauyunr, Kmji'Dol . . . ... ...

! a ' - 7 : . ::

Hiushi Zoxn Corporation, Ariakc Work, 86941, Nyriu, lapa "

Takanori Itoh Hitachi Zoscn Corporation, H u d Office, P.O. BOX 7,530-91, Oukr, Japan

A B S T R A ~ Recently highly advanced Computer Lucgnted M u r u f u n r r i n g ( ~ yam has corn to be regarded u essential, in o d e to erublirh competitive shipyud, intepucd QM wvi- ronmcnt i s essential. L u k of intc8mtion of product infomr- ouru r lol of problm taday. Considering caanurt problemr, ruthon di#rur two practical roluriont: integration of product information distiibutcd in vcnicrl and hofizod.

Vvrical integration is inteption of 3D CAD dru thmughwt h e p r ~ from r concept design phase to r production phue in order m eliminru double work between adjacent two phase . CAD/CAM ere effective tools for this purpose.

Horizontal integration is integration of human rctivitiu ia vuious dcpamncna, c.5, hull/outfitting designers, hullloudiuing plmncn. Communication between people is the nrain issue, especially between dcti jncn md production plrnnen in ofdu to climinnte rc-work doc to late urivol of the other people's rcqucrL Ploduct infomaiio~l flow is in various ways Lilce a network, Product Model is 8 most probable tool.

Authors definition of Product Model is a product informtion rnmgcr who cul provide meaningful data for pwple in design and productiun deputmcnu. For rhs dewlopmcnt of Product Modd, object daubuc, 3D g n p h k and expcn syrm arc key technologiu

Stepwire development or promtype development u more efficient than wattifdl type development. Authors believe in piototypingwith objcu-orienlcd p u d i g m Moduluiiy should be considered in system uEhiteuurc. Applkuion systems should be dependent on Producl Model inrcrficc, but independent of drubuc. Mditionrl rppliuuons can be developed without giving suiaur iafluense on, existing ipplieuiarrr,

~ntcr-pracesz ccmmunicrtion is c f f d v e for development of appliulion systems, but not yer confirmed in practical use.

, . .

AN IMPLEMEN'X'ATION OF A PLIDDUa DEFDWXIOIV SYSTEM IN COMPUTER XNTECUTED DESIGN AND MANUFACKJRING ' ,

., i , a , .

T o r h i h Nomoto d Aoy- . . . . , ' . I . , ' warncat of Naval A r s h l w m dr Ocean Enghauing, a ~ & t y d

' " ' ' "' '

Univcniy of Tokyo, 7-El. H q o , Bunltyeku, . . . T w , 113, , J l p a I '

HPriymto I ! t i 0 ' , ' I: ,

Aguncy for Assessment rod Application of Tcshnoloe QPP T h l o g i ) , PDT-PPI Kclwtui, l2flwr, JLMH Thmrin No. 8, JJurtr, & h h , #

'

ABSTRACT In the previous meeting which w u held in Rio dc Jrncim in IWl, rht authors rrponcd a papa 'Ship Defrnilion Systun in Compukr IntcgraPcd Duign md Mllruf~sturing" . In hat paper, r product Jctinition s y s m for shipbuilding u d i w s d fmm t&e stand point of Lhree functions. 'Ihese ue rlcsipn function, cuuing funuion aid usembling function.

In this paper, a deep considcnrion is mainly paid for dczign function Ia order to makc i t pasibk to design my kinds of ship svucturcs in three dimensional space, new con- cepu of 'Roomn and "Walln is introduced. Roam division funcrion k rlso introduced just as cutting function wu introduced for plalu in f ie previous pap, And-it can 4 applied to any typw of room wirh rurroundcd wdb. In rddicion o 'Room' and 'Wall' concept, design function of inlunar ctructulu such as urarvcm n n p . flm and l id- trs ue 1156 considered, By using rhc above menuoned function, ii is poYibk 10 dcaign any kinds of ship. I

Just like the previous prpet, this r y t l ~ l h a aha cutting f u n c h n uld u~cmbling function for platts and p h k suucuru. 'Ilucfore, a designed ship svuculn cm k cut inur smrUcr elemenu By using assembling function, simulrtion of rnrnufaciunng or ship smctures can bc cur id Some of mul l s which rrc implernenlicj u an oppliadotr system will k rhown. One of the mast imporunt thing k hat Utwe application systems, aze fully inkgrated with hq product model which L d&bd in this paper.

INTRODUCTION Even though a g ~ t Jml of elcant haw been cunied fa impbmcnring computer riclcd design and manufacturing sysum in shipbuilding indusuics with advanscmcn~ of computer technology, engineers in this field seun to bc not suisfkd wilh their existing

EXPERT PROCESS PLANNING SYSTEM OF CIM FOR SHIPBUILDING

Hiroshi Nalrayrmr NKK TSU W o w , 1 Kokan-Cho Kumozu, TSU-Clty, , Mie , Rt , J & p ... ,' ,

ARSTRACC . , ... . " , Q i , ' ! , . . ,

Process pluming plays not only an iategml rnk in r m u l r i . a p p W W ~ ~ d ClhQ but Jso it s e w 8s the active i n f d a n link the ddgn dqWmmt and the production dtpmrnent in shipbuilding. Ernbodyin8 the properdu of can cum^ an- gineuing, procesa plrnning supports such infonnuiospr#csriq Mivities. On the OM hand, this paper proposes knowledge bucd ptaesu pluming system fordecrrmining the sequtnce of ship block assembly and p i p fitting, On thc otJw hmd, i t synrhcairu factory information W r i n g Group Technology (m which ir thc unduly@ muwfac- turing t o n u p l In this paper, factory information (production lina, workshops, wwk facilities etc,) is clstdfied according to rhe sinrllvitier of design and ptoductlon chum acuaisucs u well u rirniluitier of zpecific pm. By uring wcput p w knowled80 and Group Technology, redundant procerr pluming work cm b,. rlimiartsd which hu been performed mmudly, Optimal acqueace af production k deacrmintd, md productian flow can be 8tIdyzed quickly. This paper describes the verification of iu practicability and tf2cctivemv by exemplifying rhc nurdud MoJc wcrnbly sequence, utilizing graphical user interfaa: which enablb the user to correct plluvling rcsulrs interactively.

INTRODUCITON Shipbuilding must abandon iu vdilioad Irbinrcnrivc chulster, and dm enhance the productivity with chc reduction in lead cim. 'Ihc Ship & Ocun Founduian, Technical Commicw on CIM for shipbuilding has llro been developing and researching CIM. Shipbuildin8 is norrnllly ideatifid with 'so-called" job-order production. Tho fmum uf shipbuilding can be ntmmuitd u fdlom:

enormous qurg~rty of infonrution for various producu which we have lermtd intennedirte ptoducu continual ongoing n~odifiution by customer from specific contract to delivery during design md production.

* simu1tmtaur'~rnMing for hull pans and outfiaing pans I

accuracy wunnce to bt needed (which heat dktonion by weldinn demands) concurrent engineering between the derian dcprnmcnl and the production . depamurt

. . . . . . . .

SCHEDULZNG MODEL OF CIM FOR SHIPBUILDING .

T. Minunun , .A. - . CIM-Projeq Koyrgi chnmuctiar Dcputwlt. . . ., . Miuu bihi Heavy Ldunrtu, LuL , ~agurid 'em.& Ba* Workr ., ,,; '' * # - : : ~ a . * . * , . . . '. 180, Koyrgi-mrehi, N*@-guo, N4asJGlrqfi143, Japan - .",,:i!:~ ,,

L-:L * ,: ,;.;.,-, .. e ! ! ; , . , : , . : : _. , . I . !

. *,. . Vrrr?sti-\ ,+- -'

AisTRACr p:,, ~!;!:!:~: , /

fbr spmaliution of cchdulin8 ic wantid to luliv QM for ihpkildii. -7, the scheduling work in shipbuildin# Iugely tdu err the pemnd t t p c h c c d 8 pluvrcr rhrr is not suilsble for wide applica~im lie systemuizrtion of dwdulinp an not k a s u r u l y uecutcd. Thh paper iatroducu tk. d y d r d hc ddu l ly w k in hi* J ,,

building for rulir. tbe M u l i n g 8yNm in Ch( for chipbuildhy. . ,

" ' :" " .+c . ', ' , e.9 !

' ! 3 f ,,

PPTRODUCZlON CIM for ~ p b u i l d a conrLD of a pmdua modrl and cppliu~on cyrormr'ib.~rPPll~ '

I P' " tion system M h c d e d t~ ~ p p o r ~ design wQ(JU, pmdycth woks d o r h a & h a , system will gcncmb infannuion d f i t i d y by its funstion d ruromrric information generatian and that of uvble gnphid wr M a c e 11-21. On tha ofherhad. rhc prodmi model will quickly and rccurrlcly communicate rnrC adjut i d d o n rmong the rp- plication system, By in~oduchq CIM, tbc following ianovuioa on rchcdulinl un be expccwl. - Information neccuuy for design and ptoarr plrnniq ir M y to be q u i d u

rhe~ysup.Tk~n,thsvolumofworisladcrak:yuC#dsuly. .: . - Possible unang tk scheduling for different tenru, ruch u that between the long term tchcdulin'j d middk turn m u l i a g , can k djuszai. - ScMuliny un k mom resuntely executed 'P iu~DoXkk' cycle in control* ling work perfonnrnccr

CIM for shipbuildins quim (1) thc yucma& of rhaduliw ud (r) he modeling. of informdon for rchduliu~ for th control in tb product modcL The scheduling in dipbuilding, u pnoent n l y on the upah of the upricaEed plm . n m in charp. Howmr, most ofthe expeniff of those plumen ue m a d wirh some p d c u l u shipyudt, and, haafore, not ruilrblo fur wide rpplicuion. In chir ngud, I)# systcmnlisnlim of rheduling un not be recunlcly crccuced (3 J. In d d i h , m po).. lem including thc following, cm be also pointed out - The qullity of Ihc & d u l i n ~ deeply dcpenb on tk sku of the pluum, . ,

Up jrading the quality of the schcdulin~ L diflicult , . In order to improve thue tiuuioru md realize CIM fos shipbuildin& thr W u l i n g

THE ROLE OF COhIPUTER INTEGRATED M A i i A C l T J R I N G FOR THE FUTURE SHIP BUILDING

. C ' . - Takto Koyrmr I - t d

Drpanmcnt of Navd k h i r c c n u c r ~ d O c w Engineering : - . ., 'PF ----,

University of Tokyo 7-3-1 Hongo. Bunkycyaku, Tokyo, Japan

c t b l Cc.,:,.; f 3 ULL&. ,

Key words: CLM, Shipbuilding, CADICAM. Rqr t inmcnt r . ,. - , 1::-2 r 'JC.'- ---

ABSIRUC~ , ,-: 51 ... ... : * ' 7 .A\,-,\

a i , ? - s;,; >:d&

The role of CIM for the future shipbuildiag i s reviewed. Ship and , , -<: : 7 Ocean foundation is now conducting a 3 years joint project for the : - protory~inC of shipbuilding CIM &ice '1989. Thc d m of h e project is to figure out h e features of rbe shipbuilding CIM and to review the recent software environment through prototyping from the view point. of large system development. Review is made by full time members ~c:uaIIy ming thc most advanced systems.

VL:- ,,,,:, in;;sdl;;cs ibr ex-6T-;cnzi ob;&acd L-- & b u l l 8 .L- u b r ---:-a. y.uJcrr - - A we-

point our the importan~ aspects for the shipbuilding CIM.

1. Wlia! we exoect for C I m

CIM(Cornputer Integrated Manufacturing) is expected to be the key ttchnology for any rypcs of industry now, Many people believe that they c ~ n n o t survive in the future industrial society without CIM. It is so much expected that even package programs for CIhI are sold in the market with reasonable price, Howovcr, it may be roo optimirtic to believe t h a ~ if you purchase a CIM package program. then you will survive happily i n the global shipbuilding marker in the future.

CIM is considered to bc an integrated iniormation proccasing syslcm for the production act iv i ty and consis ted of following fo~tr funcrions:

1. Business functions 2. Product design 3. Manufacturing planning 4. Manuhcruring control

The most common CIM system. which is used in mass production Lype industrics, mainly focuses on the inkct ion of business functions and producuon control. In other word, the commonly used Clhi is on i nkgra ted sales and production conlrol systettl.

In mass production type industries. product design and manuiacturing planning are made extremely in detail beforehand and the resulu are stored in a daubasc. So [hat when they want to makt s product. they can

Appendix C: Selected Software Abstracts

Computer Select June 1995 * Data Sources Repon COPYRIGHT 1995 Information Acccss C o m p q ..............................

Architectural Power Tools

Eclipse Sohare. Inc. (WA) 301 W. Holly. Ste. U4 Beflingham WA 982254328 800-758-6779; 3604766 175 FAX. 360676492 1 Tech support: Use main no.

Categoy Sohare. Applications

Engincering/Scienlific

S ~ 1 c a t i o n s Mfr. suggested list price: EU)O Number sold: 1.000 Release date: 1985 Application: Civil Engineering/Architecture Compatible with: PC-MS/DOS: Apple Macintosh; Sun

SPARCstationJSunOS Disk storage required: 2 MB Additional hardwardsohare required: AutoC AD Source language: AutoLISP; C Customer support: Free phone support Site licensing available: Yes

S m AutoC AD architectural system. Pmides parametric commands, integra led 3D routines, 3 D qmbol libraries, i n d m standard kqnoting and flexible layering system Includes hatch paaern and automatic elmation generators. mmie generator, COG0 routines, unlimited scheduling capabilities, full-screen text editor. source code and dynamic space reporting.

Other Tenns architecture; source code; computer-aided design (CAD); scheduling: editor

Computer Select June 1995 . Data Sourccs Repon COPYRIGHT 1995 Information Access Company

.............................. PASS

Dala One. Inc. 5420 Soulhcrn Ave.. Ste. 106 In&anapolis. IN 4624 1 8004324664; 3 17-244-2999 FAX: 3 17-240-6255 Tech suppon: Use toll-free no.

Categop S o h are. Applications

Engineering/Scienufic

Spec~fications Mfr. suggested list price: $5,400 Number sold: 300 Release date: 1987 Application: Cibil Engineering/Architecture Compatible ~ i t h : PC-MS/DOS; SunlSunOS Mnimum R4M required: 8 ME Disk storage required: 15 ME Additional hardwardsoftware required: AutoCAD Source language: C Customer support: Maint fee $1.350 per IT.; phone suppon Site licensing nailable: No

S v Designing tool. Manages facilities layout product @cation. product re-order and physical assets. Generates and stores furniture and equipment configurations, automatically generates elaations. counts and prices components. compiles repons and compares project databases. Automatically generates 3D models.

Other Terms comparison. computer-aided design (CAD)

.........................

Compuw Select June 1995 * Data Sources Rcpon COPYRIGHT 1995 Information Access Company ..............................

Allegro Common LISP (V.J.2)

Franz. lac. 1995 Univenip Ave. Berkclq. CA 94704 800-333-7260; 5 10-548-3600 FAX: 5 10-548-8253 Tech support: Use main no.

Categop Sohare. Applications

Artificial InleIligenceEspen Systems

Specifications Mfr. suggested list price: $4,500 Release date: 1994 Application: W i c i a l Intelligence/E.upen Synem Cornwile with: Sun SPARCstatiodSunOS: Silicon Graphics; HP/HP-UX; IBM RS/6000/AIX; DEC DECstationnrLTRIX; AT&T UMX System V

Minimum RAM required: 12 hlB Disk storage required: 50 MB Source language: LISP Source code price: $2,000 and up Customer support: Maint. fee 25% of purchase price per )T. aAer

firsl )T. Site licensing available: Yes

S v Features CLOS for objectoriented programming. Implementation of Common LISP meeting all Common LISP speciftcations. Produces highly optmmd code. Includes lexically scoped interpreter and compiler, debugging aids, bitlfield manipulation. userdefined error handler, LISP Library and parserhash table facility.

Full Descriphon An implementation of draft ANSI &dard Common Lisp OBen X Winclo~r support and integration uith the UNM operating system. Programmers work in an interactive environment with incremental compilation automatic memory management and W n g tools. Code written in other languages. like C or Fortra~, can be integrated with LISP programs using the foreign function interface. Includes the Allegro Presto dclivery system for creating runtime applications. which can be used with Allegro Runtime for lowcost licensing.

Features a Common LISP Object System (CLOS). the objectoriented standard for LISP. Includes ~ n a m i c binding method combinatioq generic functions and tbe metaobject protocol. Used to build anifkial intelligence applications. such as language processing and e.xpert qstems, and to incorporate A1 technolo@ into other fields. such as computer-aided design and manufacturing.

Computer Select. June 1995 * Data Sources Report COPYRlGHT 1995 Information Access Company ..............................

QuckEST/CAD for Windows

ConstrucWe Computing Co., Inc. 5600 Inland Dr. Kansas City. KS 66106 800-156-21 13; 913-596-2113 FAX: 913-287-7652

Category S o h are, Applications

EngineeringIScientific

Speclfimtions Mfr , suggested list price: S 1,495 Releasc date: 1986 Application: Civil Engnecring/Architecture Compatible mith: Window 3.X Minimum RAM required: 5 12 KB Additional hardwardsofhare required: Windows 3 .X Source language: BASIC

Summaq System for e.wacting dimensions, areas and quantities from drawings prepared AuloCAD or other CAD systems utilizing DXF files. Transferring this information to QudrEST qsems to sped up p r o m of preparing estimate from CAD drawings.

Other Terms computer-aided design (CAD)

Attachment 3

Software Questionnaire Responses

Evaluation of Computer Aided Software as a Space Analysis ...

Documents

Transcript of Evaluation of Computer Aided Software as a Space Analysis ...