conference report: cad 84

2
CAD 84 Sixth international conference and exhibition on computers in design engineering Metropole Hotel, Brighton, England 3rd-5th April 1984 CAD 84 is the latest in a long-running series of international 'state-of-the-art' conferences held in the UK, started in 1974 and organised every two years by the journal Computer-Aided Design. Although Brighton in early April is not everyone's cup of tea, this conference has attracted a steady following and is now something of a firm tradition among the growing CADCAM frater- nity. About 450 attended this year. The associated exhibition has grown rapidly in size and this year included over 80 exhibitors — half as many again as in 1982. At over 7000, the attendance re- flects not only the rapidly growing market for CADCAM systems but also the fact that the real focus of develop- ment has moved from academia to the system vendors. There is a unique flavour about these CAD conferences, with an odd mix of academic and industrial interests. The associated journal Computer-Aided Design could be permitted the label 'learned' and the conference has similar aspirations, so that the papers tend on the whole to deal with the more de- tailed technological and mathematical aspects of CADCAM. Add to that the enormous breadth of industrial applica- tions and we see that most of the pres- entations must be comprehensible to only a small fraction of the audience, although a different fraction for each session. Two parallel streams were run, it being notable that engineering appli- cations were relegated to the less pres- tigious Clarence Room! The conference The conference proceedings are pub- lished as a bound volume by Butter- worths. The contents are divided into areas of application, which suffer from the usual problems of definition and the need to balance the number of papers in each session. With over 60 papers in the proceedings this review will attempt simply to pick out the highlights. The opening presentation by Pierre Bezier, a founder member of the inter- national computational geometry frater- nity, signalled the fact that no less than ten papers were to be concerned with geometric modelling in one form or another. Given the fundamental impor- tance of such techniques perhaps this is not surprising, although the nonspe- cialist becomes somewhat bemused by the apparently endless range of diffi- culties presented by the many methods of geometric modelling. The current state of the art is represented by the combination of surface modelling with solids modelling as described by Jared and Varady, although it remains to be seen whether such extensions of gener- ality improve or worsen the ease of use. The session on drafting systems was a mixed bag, mainly for the software system designer. Ishigami's paper on automatic drawing and cost estimation for timber-framed house construction showed how easy it is to produce an 'expert' system when engineering stan- dards are well defined. Whether the system will be widely used by small builders, even in Japan, is another matter. The topic of expert systems merited its own session. Cero and Coyne gave a nice potted review of expert systems terminology in the context of architec- tural design, pointing out why pro- cedural languages are inappropriate. The next paper by Bowen showed how a satisfactory expert system (for con- figuring microcomputers) has been writ- ten in a procedural language! As always, it would appear that worrying about nonideal tools can slow up the progress. Expert systems in the context of CAD are aimed at total design, involving op- timisation between a possibly large number of probably conflicting criteria. The structures session was entirely con- cerned with optimisation using mathe- matical techniques: Topping and Robinson applying sequential linear programming to design three- dimensional timber-frame structures; and Frangopol investigating nonlinear programming techniques to produce reliability-based optimisations for plastic-framed structures. It is inter- esting to contrast these structural engi- neering approaches with the paper by Radford, Hung and Gero on acquiring new rules of thumb for architectural structural design in order to build expert systems. Optimisation of struc- tural design appeared again in the design techniques session, the paper by McCafferty showing how sensitivity analysis allows interactive qualitative assessment of the results of design changes which would be completely uneconomical if the structure had to be completely re-analysed each time. Electronics applications were thinly represented. Harris, Davidmann and Musgrave described the hardware description language for HILO-2, a com- mercially available logic simulator allowing mixed modelling at system and device levels. Litovski and Petkovic de- scribed simulation of time-domain tem- perature dependence in small-scale MOS electronic circuits. And Swan and Eades described CADIC, which is appar- ently faster than GAELIC in carrying out design rule checking when laying out integrated circuits, provided that the window size can be limited. In large-scale industry the possibility occurs for correspondingly large-scale integrated design systems. Systems of this kind were reported from ICI for load-bearing steel work in process plant; from Czechoslovakia for design of induction motors; and from General Electric Company (USA) for the design of steam turbines. All three have in common the notion that 'detailed drawings and manufacturing data are produced automatically as a byproduct of an engineering design process', as distinct from the use of interactive drafting. In each case the design rules are well established, although the result is most impressive in the context of complex steam turbine design. The ICI application is ISOSTEEL, which in turn links to PDMS for plant and piping layout and to DOGS for detail drawings. Here we glimpse the value of having underlying standards of data- base and language facilities, a theme 116 Computer-Aided Engineering Journal April 1984

Upload: brain

Post on 20-Sep-2016

214 views

Category:

Documents


1 download

TRANSCRIPT

CAD 84Sixth international conference and exhibitionon computers in design engineering

Metropole Hotel, Brighton, England3rd-5th April 1984

CAD 84 is the latest in a long-runningseries of international 'state-of-the-art'conferences held in the UK, started in1974 and organised every two years bythe journal Computer-Aided Design.

Although Brighton in early April is noteveryone's cup of tea, this conferencehas attracted a steady following and isnow something of a firm traditionamong the growing CADCAM frater-nity. About 450 attended this year. Theassociated exhibition has grown rapidlyin size and this year included over 80exhibitors — half as many again as in1982. At over 7000, the attendance re-flects not only the rapidly growingmarket for CADCAM systems but alsothe fact that the real focus of develop-ment has moved from academia to thesystem vendors.

There is a unique flavour about theseCAD conferences, with an odd mix ofacademic and industrial interests. Theassociated journal Computer-AidedDesign could be permitted the label'learned' and the conference has similaraspirations, so that the papers tend onthe whole to deal with the more de-tailed technological and mathematicalaspects of CADCAM. Add to that theenormous breadth of industrial applica-tions and we see that most of the pres-entations must be comprehensible toonly a small fraction of the audience,although a different fraction for eachsession. Two parallel streams were run,it being notable that engineering appli-cations were relegated to the less pres-tigious Clarence Room!

The conference

The conference proceedings are pub-lished as a bound volume by Butter-worths. The contents are divided intoareas of application, which suffer fromthe usual problems of definition and theneed to balance the number of papersin each session. With over 60 papers inthe proceedings this review will attemptsimply to pick out the highlights.

The opening presentation by PierreBezier, a founder member of the inter-national computational geometry frater-nity, signalled the fact that no less thanten papers were to be concerned withgeometric modelling in one form oranother. Given the fundamental impor-tance of such techniques perhaps this isnot surprising, although the nonspe-cialist becomes somewhat bemused bythe apparently endless range of diffi-culties presented by the many methodsof geometric modelling. The currentstate of the art is represented by thecombination of surface modelling withsolids modelling as described by Jaredand Varady, although it remains to beseen whether such extensions of gener-ality improve or worsen the ease of use.

The session on drafting systems was amixed bag, mainly for the softwaresystem designer. Ishigami's paper onautomatic drawing and cost estimationfor timber-framed house constructionshowed how easy it is to produce an'expert' system when engineering stan-dards are well defined. Whether thesystem will be widely used by smallbuilders, even in Japan, is anothermatter.

The topic of expert systems meritedits own session. Cero and Coyne gave anice potted review of expert systemsterminology in the context of architec-tural design, pointing out why pro-cedural languages are inappropriate.The next paper by Bowen showed howa satisfactory expert system (for con-figuring microcomputers) has been writ-ten in a procedural language! Asalways, it would appear that worryingabout nonideal tools can slow up theprogress.

Expert systems in the context of CADare aimed at total design, involving op-timisation between a possibly largenumber of probably conflicting criteria.The structures session was entirely con-cerned with optimisation using mathe-matical techniques: Topping andRobinson applying sequential linear

programming to design three-dimensional timber-frame structures;and Frangopol investigating nonlinearprogramming techniques to producereliability-based optimisations forplastic-framed structures. It is inter-esting to contrast these structural engi-neering approaches with the paper byRadford, Hung and Gero on acquiringnew rules of thumb for architecturalstructural design in order to buildexpert systems. Optimisation of struc-tural design appeared again in thedesign techniques session, the paper byMcCafferty showing how sensitivityanalysis allows interactive qualitativeassessment of the results of designchanges which would be completelyuneconomical if the structure had to becompletely re-analysed each time.

Electronics applications were thinlyrepresented. Harris, Davidmann andMusgrave described the hardwaredescription language for HILO-2, a com-mercially available logic simulatorallowing mixed modelling at system anddevice levels. Litovski and Petkovic de-scribed simulation of time-domain tem-perature dependence in small-scaleMOS electronic circuits. And Swan andEades described CADIC, which is appar-ently faster than GAELIC in carrying outdesign rule checking when laying outintegrated circuits, provided that thewindow size can be limited.

In large-scale industry the possibilityoccurs for correspondingly large-scaleintegrated design systems. Systems ofthis kind were reported from ICI forload-bearing steel work in processplant; from Czechoslovakia for designof induction motors; and from GeneralElectric Company (USA) for the designof steam turbines. All three have incommon the notion that 'detaileddrawings and manufacturing data areproduced automatically as a byproductof an engineering design process', asdistinct from the use of interactivedrafting. In each case the design rulesare well established, although the resultis most impressive in the context ofcomplex steam turbine design.

The ICI application is ISOSTEEL, whichin turn links to PDMS for plant andpiping layout and to DOGS for detaildrawings. Here we glimpse the value ofhaving underlying standards of data-base and language facilities, a theme

116 Computer-Aided Engineering Journal April 1984

well promoted by the NorwegianCADCAM software industry.

The exhibition

In many ways the exhibition was moreimpressive than the conference.CADCAM is now an industry in thebillion-dollar league; the competition isintense and the rate of progress, at leastin hardware, is accelerating. With over80 exhibitors a review must be selective,although we can pick out a number ofclear trends.

One is the proliferation of low-costdrafting systems based on 'personal'microcomputers such as Sirius, the IBMPC and a flurry of Motorola 68000-based machines, including the HP9816.Software tailoring companies such asAdmel and Norrie Hill are creating arange of systems at different levels ofcost, some overtly aimed at training,others capable of real drawing officework. About 15 suppliers in this classwere on hand.

At the next level of price the battlelines are being drawn between Apollo,PERQ and System 80 (and others notpresent) for the networked single-userworkstation market. Software vendorssuch as Alper, CADCentre and PAFEChave so far been hedging their bets,putting their software on severalmachines. ICL chose this exhibition tolaunch PERQ 2 with a 'proper7 screenand a decent array of CADCAM soft-ware. With most of the teething prob-lems overcome, the ICL marketingmuscle will now begin to have aneffect. Apollo relies on a growing rangeof models and a large software referralcatalogue, plus one or two traditionalCADCAM vendors such as Auto-troland Calma that have chosen Apollo fora rejigged attack on the market.

Some of the better known 'turnkey'vendors were notable by their absence,but the British big three were there: CIS,Ferranti Cetec and ARC. IBM was alsothere, for the first time, reflecting a newresolve to extend its CADCAM market.They introduced FASTDRAFT to the UK,a (for them) low-cost drafting systemthat fills in at the smaller-user end ofthe market where CADCAM is uneco-nomical; and also the new 5081 high-resolution raster display.

Among the British (or quasi-British)CIS had an impressive array of new soft-ware, demonstrating a trend to buildapplications on top of its basic Medusasystem. Ferranti Cetec (now operatingas Ferranti Infographics) is about to

The Prime 2250-based Mini-GDS system exhibited by Applied Research of Cambridge

have the three-dimensional version ofGNC and seems to be prospering. ARCshowed a single-user Prime 2250 optionthat may be Prime-netted, togetherwith quite a number of software en-hancements.

Software is becoming the one stableelement in a changing scene, particular-ly as the difficult demands of true com-puter integrated manufacture begin toimpress themselves on the vendors.CADCentre's GNC3D package forgraphical NC programming was on fivevendors' stands, and a previous gener-ation of CADCentre NC software iseven being reimported from the USAvia Keen Computers (Cadlinc Division).PAFEC is the arch exponent of portablesoftware, its DOGS drafting systembeing similarly widespread about theexhibition. The Hanratty AD2000/ANVIL4000 software continues to emerge asthe basis of systems from the late-comers to the market, Gould SEL, onthe other hand, exploiting CADAMthrough an arrangement with the Lock-heed subsidiary CADAM Inc., offeringan alternative to IBM. Good CADCAMsoftware seems to need considerablematurity — most of these systems havebeen under development for 10-15years!

Displays and plotters continue to im-prove, with colour the big selling point.Smaller and cheaper plotters abound,and colour ink-jet technology is a work-ing proposition, at least for businessgraphics applications.

Alternatives are photoslide or Pola-roid output for more realistic colour, Di-

comed claiming great expertise in thisfield.

Versatec offers the most advancedcolour-plotting-on-paper equipmentand introduced interactive scanningand editing, which is worth deeper in-vestigation than we could give it on thestand.

As to graphics terminals, a wholereview could be devoted to the state ofthe art. With 18 suppliers represented,there was something for everyone,ranging from the newly launchedTopaz system from Primagraphics,which supplies custom tailored high-resolution multiplane colour displays tosystem builders, to Graphic Data Prod-ucts with an eight-colour 14in displayfor the economy minded. Westwardcontinues to look good for cost-effective nicely designed CADCAMworkstations with the highest resolutionin the business.

For the rest, there were about adozen CADCAM services companies,ranging in application from tool and diemaking to land surveying, five publi-shers and two exhibition organisers, in-cluding CAD 84! Presumably they wereselling the next CAD exhibition, which,breaking the tradition of biannualevents, will be CAD 85.

From our observations it looks asthough the users of CADCAM go to theexhibition and the makers of softwaretalk to each other at the conference. Isthat the right formula? In our view themakers and the users need more ge-nuine contact.

Brian Gott

Brian Cott is with CADCentre Consultants, High Cross, Madingley Road, Cambridge CB30HB, England

Computer-Aided Engineering Journal April 1984 117