CAD 78

CAD 78

14-- 16 March 19 78, Metropo/e Centre, Brighton, UK. Organized by Computer-Aided Design Journa/. CAD 78, the third biennial conference and exhibition ...

718KB Sizes 2 Downloads 153 Views

14-- 16 March 19 78, Metropo/e Centre, Brighton, UK. Organized by Computer-Aided Design Journa/. CAD 78, the third biennial conference and exhibition that the journal has organized, was by far the largest and most successful of the series. There were over 500 delegates registered for the lectures, and the associated exhibition attracted 30 companies and another 600 visitors. Further reports on the proceedings of the conference follow below, all are written by the chairmen of sessions. The reports cover the Electrical/ Electronic Engineering and Building Design sessions. The staff of the journal would like to thank the chairmen for their help in making CAD 78 a success; and the sponsors: the CAD Centre, D e c , DoE, Displays Group BCS, and the Institution of Civil Engineers with the cooperation ofACM SIGDA.

ELECTRICAL AND ELECTRONIC ENGINEERING Part 1 Of the six papers presented, two papers covered checking LSI designs by reconstructing the circuit from the artwork, two more papers covered routing solutions for complex printed wiring boards, one paper covered interactive graphics techniques applied to the design of electricity supply networks and the last paper described the considerations given to implementing and managing a production CAD system. There were two highlights in this session. The first was an excellent invited review paper by Henry S Baird of RCA Laboratories in Princeton, NJ. As part of the research work for his PhD thesis, Dr Baird reviewed 30 papers on LSI artwork analysis from 17 different countries. He summarized tha basic operations in performing artwork analysis and

332

pointed out the salient advantages and disadvantages of the various methods. Finally he introduced a new edge-cut algorithm resulting in fast computer run time with a low memory requirements. Although his paper is not published in the proceedings, it is available in the proceedings of the 14th Design Automation Conference. These proceedings are available from either the IEEE or the ACM in the USA. An expanded version of this paper will be published later this year in the Journa/ for

Fault Tolerant Computing. The second hightight of this session was a presentation by Girvan Patterson of Bell-Northern Research in Ottawa. Dr Patterson's paper described considerations given to the problems of implementing and managing a viable CAD system as a production tool. One point that he made was that a successful organization structure consists of a software development organization, supported by an application organization which in turn supports many user groups. Within each user group there is one key individual who isthe most knowledgeable user in the group. The presentation met with considerable interest by the delegates. Some discussion ensued resulting in a request to continue the discussion in a later panel discussion. The topic of CAD organization and management is one to watch as it develops in future conferences. The area of electrical and electronic engineering produces fruitful developments in CAD. Hence, this session is an excellent area for growth for this conference. We look forward to more support for this session at the next conference.

D J Humcke Part 2 It is surprising how, when a conference is actually in progress, things become very evident which were missed during the planning stages, and even when

the papers had all come in. It was not at all evident to me until well into the conference, that most the papers on CAD for electrical and electronic engineering veered mainly towards computer-aided drawing office techniques, rather than computer-aided design of the electrical aspects. This is not a criticism of the papers presented, but a plea for people who are designing electronic and digital system by computer-aided design to submit more papers next time. As an example, I went from this conference to California and attended a symposium on computer architecture in Pale Alto. There was a day-long tutorial on CAD for digital systems with a well known instructor. The only paper in CAD 78 which came near to this aspect was a paper describing CAD for using multiplexers as universal logic elements (presented by MR EI-Karaksy). HM Bayegan presented his paper on CASS, a computer aided schematic system which was an impressive system and well described. The last paper in the session was on the CAD of rotating electrical machines by a Czechoslovak team. It was interesting to see how the use of CAD had quite naturally brought about a range of rotating machines with many integrated features. However, like the paper by CA Lynch and A Brameller on power systems there were probably not enough papers in the same area of rotating machines for this one to get the discussion it deserved. One of the commercial sessions featured Col. Hitch talking about his graphics system and a paper on Gaelic (the Edinburgh LSI layout method). It was very interesting to see that the commercial people were quite reticent about price etc., even though the chairman assured them that it was quite all right, and they gave instead some of the best technical lectures of the conference. However, despite all the efforts of the organizing committee, it cannot be said that the electrical and electronic engineering sessions were well balanced. This may be because they inevitably got mixed up with techniques both of processors and displays. It was interesting to study the list of those present

computer-aided design

and to note that there were very few people at the conference from the major computer companies. As far as I can see there was no-one at the conference from ICL, which poses some interesting questions for the organizing committee of the next conference.

Professor F G Heath

B U I L D I N G DESIGN Part 1 J Lansdown opened the first session with a review giving a spirited response to those critics of CAD who accuse it of a lack of imagination. He presented a cascade of slides on computer art, recited poetry, played music and performed a play all written with the aid of computers. He drew parallels between computer art and CAD, and suggested it was time CAD accepted aesthetics as a central concern. He thought a theory of computer-aided architectural design might now be possible. More should be done to improve the ease of using computers, and criticisms of the benefits of CAD answered by publishing case studies. Not unnaturally the conference was sympathetic, but it would be interesting to see the response of a more sceptical audience of design practitioners. In the same session, two papers describing the evaluation of computer programs were presented. T Reynolds used some simple tests to compare the performance of four drawing systems and the manual equivalent. Interactive input techniques showed savings in time but higher computer costs. The widely differing output made comparisons difficult, however, and it was agreed that office organization, design procedures and type of project were of special significance in the application of drawing systems. S Bensasson of the DOC (Design Office Consortium) concentrated on the methods necessary to make the results of evaluation applicable to a wide range of computer users. He described how computer application areas were chosen, the testing methods and the effects of obsolescence on evaluation results. A large number of

volume 10 number 5 september 1978

programs shown as available on program lists turned out on closer examination to be unusable in practice. The level of documentation on programs was often poor and important facts might only be discovered after use or on detailed examination. He concluded that several different levels of evaluation were necessary, ranging from short descriptions to full-scale tests of the type carried out by DOC on programs for perspectives and continuous reinforced concrete beams.

Part 2 This session contained two papers describing university research and one describing the development of a research program. The session began with a presentation by J Gero, who showed the use of dynamic programming in dimensioning building designs while satisfying constraints on their planning. One of the problems in dynamic programming is making the technique accessable to the user and this had been tackled by designing a command language and using a graphical display. A small, single storey house was used as an example, and this was not sufficiently indicative of the type and scale of real-life problems to convince the audience of the utility of the method. O Akin gave an introduction to the techniques he was using to design a system to interpret drawings. He described methods of analysing how an architect reads drawings and builds the results in the form of heuristic rules into a program. As the work was not complete, no detailed results were shown, and the audience was left to speculate on the insight which such a system might provide into the design process, if successful. J Burdett then gave an able presentation of work at BRE on a program for optimal design of district heating systems. Information on the site is entered using a refreshed graphics display, a tree layout is generated and iteratively improved by examining different methods of connecting the tree and calculating pipe sizes and hence costs. The program is used as part of a research project at BRE, but several questions were asked about the detailed methods used and its availability for use on

layouts for other services as well as district heating.

/ Chalmers

Part 3 Energy and thermal analysis This session contained an interesting group of papers relating to energy analysis, with particular reference to shadowing and the design of shading devices. Peter Smith of Atkins Research and Development described KOSHAD, a preprocessor for the ATKOOL energy program, using perspective views to calculate sunlit areas. Don Greenberg and Richard Rogers of Cornell University gave a brief, joint description of their computer graphics facilities and their method of inputting modules of walling with window and shading devices, for analysis and colour graphic output of shadows and solar gains. Edna Shaviv's paper from the Technion in Israel contained a very thorough analysis of shading requirements for windows with different orientations at different times of the year and the day. A design for shading devices on a proposed building was shown as a convincing example of the program's application. The last paper was on energy flows from all sources as modelled by the Environmental Systems Performance program from ABACUS at Strathclyde University. It was presented by Joe Clarke and showed that solar gain was just one of a series of complex interactions presented in the ESP program by graphs The discussion which following each paper, and more generally at the end of the session, was comparatively lively, although the most heated discussion was on the relative severity of the weathe'~ in eastern USA and IsraN. What level of accuracy is worth pursuing in relation to the needs of practice? The sophisticated analyses presented were very nice but, as one building services consultant said, he does not get paid to do this nor do his clients always appre/:iate it. It was pointed out that the paper by Edna Shaviv represented a rare case of an analysis in which all the possible

333

alternatives can be explored. The form of input for ESP was being improved and Dr Clarke envied the form of graphical input available at Cornell.

Assessing CAAD systems Dr Nigel Cross from the Open University was introduced as the author of The automated architect and a recent article in the R/BA Journal Both of these has roused some controversy and he agreed to be brief in his presentation to allow for discussion. He started by saying that it was not his aim to be controversial, but that there was little published evidence for the success of CAD systems. He referred to several examples in his written paper and to the graphs of experiments comparing human, human-machine and machine performance. He mentioned the social problems, discussed by Mike Cooley at CAD 76, which CAD would eventually cause. In starting the discussion, the chairman pointed out that Dr Cross was using a narrow definition of CAD: the use of computers to aid or automate the early design stages. His remarks did not necessarily apply to programs used at all stages of the design process. John Lansdown said that there was evidence of CAD benefits and quoted a number of tasks in his office which had been reduced from 2h to 10rain by using CAD. With regard to the social effects, he invited Dr Cross to ask his staff how they used the spare time resulting from these savings. John Chalmers also had evidence of benefits from the evaluation of CEDAR on the design of steel frames. He felt that Dr Cross filled a useful role as the iconoclast of CAD. Professor Tom Maver referred to the invention of working drawings 5000 years ago, and how the master builders were sceptical of their acceptance. Did CAD need thousands of years before it would be accepted? In reluctantly concluding a lively discussion, one chairman referred potential users of CAD systems to the checklist at the end of Dr Cross' paper. He had raised important points which needed to be discussed at a conference like this.

334

Too many people, he said, had been waiting for hardware developments to solve their problems instead of The author of the first paper, Dr getting on with the job. The impliRobin Th'ng, was unable to attend, and this left another interesting cation was that they had been getting comparison of three integrated on in developing and applying systems for building design and RUCAPS in the design office, but documentation. Unfortunately the there was little information on how papers which came after lunch sufRUCAPS worked. That it does work fered from a dwindling audience. was clearly illustrated by J RydeShould the programme sacrifice Richardson, a partner of Gollins unique presentations for more Melvin Ward and Partners. He showed central themes? out.put from a large'university comThe three systems presented were plex in Saudi Arabia designed jointly OXSYS from Applied Research of with American consultants, and a Cambridge by Paul Richens, RUCAPS smaller office project in London. from Gollins Melvin Ward and He said that GMW's intentions were misunderstood. Their aim was to Partners by Dr John Davison and Julian even out the peaks and troughs in Ryder-Richardson and CEDAR3 from their work rather than to aid design. the DoE Property Services Agency by They did not wish to prevent intuitive Dr Brian Thompson. The first was a 'leap-frogging'. The trials of a sketch fully integrated system originally design aid in the office had not been developed for the Oxford Regional very convincing. Health Authority and allows comBrian Thompson illustrated CEDAR parative evaluation of alternative designs 3 with the results of a recent workas well as producing very full proshop session to examine alternative duction documents. The second was layouts for a Government building. two-dimensional and concentrated on As with OXSYS, CEDAR had resorted producing drawings, since these are the main concerns in the major projects to perspective drawings to arouse the interest of clients and audience. to which the system was being applied. The third was for appraising sketch Neither of these systems used such designs with regard to cost, environdevices in their earlier, idealistic mental performance and visualization. days. The more fundamental outPaul Richens gave a well illustrated put for comparing performance was of talk with a series of perspective views heat losses, daylight levels and outline and output from projects by various costs. Graphics was also used for conusers of OXSYS. He described the firming data visually. database on which the system operates The expected discussion did not and the method used for storing 3D materialize. The supporters of one descriptions of components in the system refrained from criticizing the codex. He explained the level of others and the inexperienced appeared generality in the system with BOS, a to be so impressed by the facilities basic operating system independent of of these systems that they found application, BDS, the building design difficulty in relating them to their system independent of particular forms tentative explorations in CAD. Should of construction, and DDS, the detail conferences try to impress the unindesign system, of which OXSYS was volved with the most advanced applione example based on the Oxford cations, or should they encourage Method of construction. In an answer sta.tements of user requirements? The to a question on applications outside nextconference might contain a system building, he replied that such session.on the subject: 'What .would systems were unlikely to be design practice like the computer to economical where there was not a do'. high level of repetition of components. The chairman concluded the. Dr J A Davison asked some fundasession by saying that he felt that, mental questions about CAD in his in comparison with presentations of presentation from GMW Computers. these same systems a¢ previous conHe referred back to the optimism of ferences, they had been reduced in the York Conference in 1972 and scope, but the tasks carried out the expectations expressed for 1980. were done better and hadnow been Current application

computer-aided design

tested in practice. After lunch a few people were present to hear a presentation by R C Waterfall of the Department of Electrical Engineering and Electronics at UNIST. This was novel and was largely concerned with logging movements of pedestrians in circulation areas with the eventual aim of designing corridors more economically. While being an original idea, it suffered from being carried Out without architects and the discussion pointed out some more practical applications of the work in the design of escape routes, or mass circulation areas in the underground, for example. A Sarkar of the Professional Computing Branch in DoE Property Services Agency, described the programs available on terminals in PSA regional offices. Their present budget was £¼ million for the year and this was spent on commercial bureaux. A list of programs was supplied and these could be made available by PSA or through the Design Office Consortium.

Rob Howard Computer aids for building layout design Over the past 15 years, considerable effort has been expended on the

CAM 78 20-22June 1978. NEL, East Kilbride, UK. The joint NEL/University of Strathdyde CAM 78 conference on CAM and CAD/ CAM was attended by 140 delegates from ten countries. 28 papers were presented and these ranged over topics as wide as review papers on computeraided manufacture world-wide and the social implications of these new techniques to specialist papers on topics such as CAD/CAM related to die sink~ ing for hot extrusion, and manufacturing compressor and turbine blades for aero engines. More general-purpose topics on CAD/CAM systems currently available, both commercially and in research laboratories and

volume 10 number 5 september 1978

building layout problem with little apparent effect on the practising designer. This failure to contribute to practice has been explained, largely, by criticizing the single measure used in the objective of the problem. This measure has almost always been related to movement costs between spaces which seems to bear little relation to the various, often conflicting, objectives the designer is aware of. The three papers in this session addressed this problem in various ways. A Bridges, of the University of Strathclyde, presented a procedure which aims to manipulate complex multivariate problems to produce an architect's 'bubble diagram'. His is a new approach based on synthetising various existing techniques, drawn from other fields, to manipulate multivariate data. The discussion following his paper concentrated on the large amount of da~ta required by the procedure. J Lansdown, of Turner, Lansdown, Holt and Paterson, made the comment that there is not enough data available in practice. D Campion, of Cusdin, Burdin and Howitt, pointed out that the problem in practice was to build up and record the data not to manipulate it. H Neuckermans, of the University

of Leuven, presented a paper comprised of three parts: an outline of classical location-allocation methods; an architectural critique of many of the layout methods used in architecture; and a new algorithm designed to circumvent some of the difficulties described. Because of time limitations, he was only able to present the first two parts. There was general agreement with his critique. Edna Shaviv, of Technion Institute, Israel, presented a decom position-recom position approach to evaluating the layout of existing designs. The process makes extensive use of cluster analysis and graph theoretic techniques. A comparison between the clusters produced by the process and the design of a 130activity building was made. The presentation drew considerable discussion which fell into three categories: those concerned with the methodology involved; those commenting on the relation between the problem and practice, and finally, those viewing the work as a positive contribution to bridging research and practice.

methods of manipulating Bezier patches were also covered. There was still an interest in NC programming techniques, with four papers dealing with this topic and one on verifying the accuracy of control tapes using a computer. There were papers on developments of the minicomputer/microcomputer invasion of the NC control system and a paper on the automatic reading of identification labels of materials in a hostile environment. This conference saw an increase in the number of papers devoted to computer-aided planning and scheduling from Holland, Norway, FDR and the USA. There were also papers from Norway on the use of robots in a numerically controlled manufacturing cell, and from the USA on simulat-

ing the operation of cells of this type on a computer. There was an interesting paper from Heriot-Watt University describing some work they had done with a local firm in combining the design, planning and estimating functions with those of the production of drawings, manufacturing data and NC tapes. Because the work is designed around a family of parts of similar characteristics whose particular function, size and shape vary from each customer, they were able to reduce greatly the effort required to produce accurate quotations and to follow these up (if the order resulted) with economical manufacture. Although tailored around one firm's product, the authors claimed that as they design the program in a modular fashion, the work could be repeated

Professor J S Gero

335