If you need help
EUCLID: the Beginnings of Computer Aided Design
For almost 20 years, the EUCLID 3D CAD-system has been used for designs of CERN's accelerators and detectors. CERN was a pioneer using a 3D CAD-system when the few CAD-tools in the marketplace were mainly 2D drafting tools. The choice made in 1982 was a good one, the vendor survived until recently and EUCLID has been the workhorse for design of LEP and LHC.
While CERN continued to develop software specific to our needs, the rest of the world was beginning to produce sophisticated general-purpose computer tools. At around the same time (1982) both Oracle and Euclid came to CERN. Each was then fairly new and each was just one of a set of competing products in a new domain.
Selection of a CAD-system for LEPIn the case of Euclid the the study group to select a Computer Aided Design (CAD) system for the design of LEP worked in close collaboration with the future users. Claude Hauviller, Hans Horisberger, Detmar Wiskott ... had to learn all about the problems of CAD - then in its infancy and with not much more than principles and promises to recommend it. They had to shepherd the thought processes of the study group to the best choice and defend that choice right through to the final decision.
At the time the CAD market consisted of a few packages for Computer Assisted Drafting and Drawing, that were little more than computerised drawing boards, and more advanced systems that allowed for design of 3D surfaces but were very complex and targeted the aerospace and car industries. The EUCLID system was one of a few tools that worked with true 3D solid geometry, knowing the precise volume, shape and position of each element. It thus was able to fit pieces together, detecting interference if an element is moved among others, as for example in a tunnel. Of course an assembly of elements can be projected on to any plane to make up a conventional drawing. Other, graphics-oriented, systems which simply remembered and reproduced the formal collection of lines which constitute a drawing could merely be ancillary and were no match for Euclid.
Another quality of Euclid was that it had a true database that allowed users to share components in a design at a time where other CAD systems worked on individual files. This allowed all designers and engineers to access the same versions of parts in an assembly and keep it coherent, a major advantage for dealing with the very large assemblies made at CERN.
Computing Power and Resources
The computing power needed was underestimated. The power available for 100 kCHF was miserable by today's standards. The VAX had a cycle time of 1 microsecond and was equipped with only 12 Mbytes of memory. The first displays were on Tektronics 14" storage tubes and each terminal cost about 100 kCHF. To make it acceptable to the directorate, Euclid ought to have served electronics applications as well! To get best value out of the large investment, the design office officially extended normal CERN working hours to allow people to work from 7am till 8pm. New money came hard too; memories linger of a day spent running around user divisions to get 4*10 kCHF to buy a 400 Mbytes disk.
Early users were not merely early risers. Indeed, without the dedication and commitment of a small number of designers who put up with extremely difficult conditions, CAD introduction at CERN would have been delayed for years. It is important to recall their working conditions in 1982-3. Whole days were spent in darkness behind windows covered with black curtains. There was no mouse - the cursor had to be moved with 2 knobs. There was no documentation and little or no training. Paper copies of drawings produced in building 513 were delivered by van to a local RIOS (Remote Input Output Station). Of course these pioneers had to make reservation in advance for one of the few terminals available ...
The directorate was extremely sceptical and wanted to put all available money into LEP construction instead of "computer games". However, in 1988 when LEP was well under way, Professor Schopper acknowledged that it had become clear that LEP had very much benefited from CAD, indeed some designs were only possible because of the 3D representations.
When Detmar Wiskott left, Roland Messerli took charge of the computing, interfacing ... aspects for the maturing years of Euclid, to oversee its use expand at CERN and transfer CERN's knowhow elsewhere in Europe through his involvement with the EUCLID user group.
Many key contributions were made in that innovative phase. For instance, Marcel Mottier made the new software his tool and the first convincing results were delivered largely due to his commitment. He wrote the programs ASMAC and LEGO that produced, overnight, complex layout drawings of complete LEP accelerator sections in the tunnel.
Serge Oliger used Euclid and Oracle for the installation of electrical equipment and cable routing in the LEP tunnel. This led to improvements in installation, maintenance, ordering and contract follow-up.
Another outspoken and convincing protagonist was Claude Hauviller, founder and longtime chairman of CAEC, a committee firmly in control of mechanical CAD at CERN. But throughout the acquisition, introduction and deployment it was teamwork and setting up and running the institutions of teamwork - such as CAEC's "Euclid Correspondants" subcommittee, under Roland Messerli, voicing grass-roots opinion - which made it all possible.
We cannot mention everyone who contributed to the success of EUCLID at CERN in a short article, but a major factor behind this success was effective coordination and central support, be it in providing adequate computing resources or disciplined working methods in the drawing offices.
ConclusionIt is a tribute to the care and foresight of those involved that EUCLID, of all the systems proposed at the time, was the only one destined to carry through a product cycle of unbroken development. At CERN the number of experts has grown from about ten in the early days to more than a hundred today. Each user has at his disposal something approaching the power of a computer centre of twenty years ago. One of the few CAD systems still supported after 20 years, Euclid itself is still highly capable and is the central workhorse in the design of LHC and its experiments. The more than 100 000 3D models and drawings in its database attest to the scale of Euclid's contribution to CERN.
But there is no miracle, Euclid is fundamentally a twenty-plus-year-old Fortran program and has fallen behind today's CAD capabilities. Although Euclid's parent company Matra Datavision now focuses on services and sells Catia V5 from the former competitor Dassault Systemes, EUCLID is still supported during LHC construction. So Euclid and CERN-Euclid knowhow will live on to finish the bulk of LHC design work. Today's CAD market is being studied by the CAD 2000 project to find Euclid's successor which will gradually supplant the ancestor before the fatal date, end of 2005.
About the author(s): Based on input from people who were there at the time, the article has been edited by members of the Mechanical CAE section of the Computing for Engineering support group.