Data across the ages

Working with multiple CAD platforms is rarely easy and ideally to be avoided, but for many of us it's a fact of engineering life. Few companies have the financial muscle of the automotive majors, which seem to have a policy of encouraging their suppliers to use the same CAD systems and releases as they do to minimise data exchange errors. But whether you're caught up in that problem or not, most of us come across the difficulties, if only with legacy systems. Interestingly, even the automotive OEMs have problems. Stephen Hooper of Autodesk told us that in one, 3D models of engine blocks produced using SDRC software were used to generate hundreds of 2D sections and views – so that engineers still working in 2D could use them to produce detail drawings! Philip Shade, CAD manager and senior product designer at Fulham-based non automotive design consultancy PDD, says: "PDD is fortunate, or perhaps unfortunate, to be using six of the most respected CAD platforms. Although managing a project in a single package for the duration has clear benefits, careful selection of the use of these, perhaps in combination, is essential to ensure complete success." The issue for him is securing design intent and maintaining robust control of specification, which is the focus of much of PDD's PLM (product lifecycle management) and CAD work. "This becomes more of a challenge when a number of packages are used and design intent is in non-CAD data." His solution: "PDD's primary objectives in its definition for a universal PLM solution are to be able to manage multi-source data, retain traceability and change control without constraining creativity and versatility, and to provide effective communication through a common interface." Alec Cassells, who worked for IBM before CAD was invented, goes to the heart of the problem: "You lose a lot of design intent and embedded knowledge in round tripping between systems." Referring to IBM's own operations over a decade ago, he says that much money and effort were saved by rationalising to a single system. IBM now has only one CAD system, Catia, one PLM system, Enovia, and one electronic CAD system, Cadence – and he counsels caution in any other approach unless legacy systems are in place and dealing with separate business units' requirements. Turning to the problem of legacy system data, Cassells cites the example of the Canberra Bomber, brought into service in May 1951, and with the RAF's last five PR9 machines scheduled to remain in service until at least 2006 – and similarly the Nimrod, which started as the DeHavilland Comet 4 in 1958, the last three of which are not expected to retire until 2012. The original Canberra drawings were on parchment – likely to remain readable for the foreseeable future, but if someone needed to make replacement parts to keep a machine flying, the lowest cost, quickest route might be to re-enter data into the incumbent CAD system. Then they could perform modifications and FEA, automatically generate CNC machining instructions and so on. Scanning the drawings wouldn't generate a solid model, and "paper doesn't know if a circle is a hole, a disk, or a representation of a round solid." A similar, but more serious problem might be that of recovering nuclear reactor design and test data, particularly the results of early experiments on reactor stability that nobody would dare repeat post-Chernobyl. Sources tell us this data is still held on computer tape, with no machine now capable or reading it. One solution to that and other old data maintenance and recovery issues – far fetched though it seems – is to set up a computer museum. To recover BoM (bill of materials) data on a IBM System 38, for example, would require interfacing to an IBM eServer iSeries (AS/400 in legacy speak) – an expensive idea. And what about accessing Eurofighter design data in 25 years time? IGES and STEP have their places, but looking back over the dead technologies of the past 25 years, one has to wonder if any of the established CAD data exchange standards will still be around. Cassells' view: microfilmed images and ASCII text files should still be readable, but for the rest, who knows? Best solution is probably to transfer important data through successive systems, provided it is worth the expense of doing so, and provided it is maintained in a form that can still be read and understood.