Boeing Co.'s (BA) sleek 777 jetliner and Frank Gehry's breathtaking Guggenheim Museum in Bilbao, Spain, share a surprising bit of artistry. Years before the first rivet was punched or scaffold raised, both designs were perfected as exquisitely precise three-dimensional computer models. Computer-aided design (CAD) tools have been around for 30 years, speeding development of everything from microchips to running shoes to autos. But cutting-edge creations such as the 777 have pushed the outer limits of electronic design, leading to a powerful new breed of CAD.
No one is better qualified to expound on the new tools than Stephen M. Ward Jr., IBM's general manager of global industrial services. Ward ran IBM's blockbuster ThinkPad division and has also served as Big Blue's corporate CIO. He believes that the information stored in a manufacturer's CAD systems can be harnessed to do much more than just fuse together virtual parts. By linking the geometric description of a component to related data--its weight, its price, or where it is stored--design teams can make faster, better decisions about what to build, and how. Top-tier manufacturers are drawn to this concept because it promises big productivity gains. For IBM's industrial clients, "design systems will be larger than enterprise management or supply chain systems in the next couple of years," Ward predicts.
As the world's largest software integrator, Big Blue speaks with authority. About 40% of its $88.4 billion in revenues last year came from helping companies install, manage, and link their supply-chain systems, financial accounting packages, and other software systems. What's more, IBM (IBM) is the sole worldwide sales and service partner for Dassault Systemes' industry-leading suite of CAD and computer-aided manufacturing tools. Dassault's CAD program, CATIA, is the tool of choice for the most demanding design jobs--including the 777 and Guggenheim Bilbao.
Ward recently spoke with Industries Editor Adam Aston to flesh out his vision of the digitally designed future.
Q: Can you tell us what kinds of things to expect from the next generation of CAD software?
A: In the next several years, we'll see a higher degree of digital design collaboration. An average new car has about 10,000 parts. In designing that car, there are typically 3,000 parts changes in any week. An engineer will finish a change--perhaps he rearranges the layout of the dashboard--and then he has to relay his changes to the other designers and related suppliers. Imagine using paper or e-mail to do this. You'd have to repeat this process for every one of the hundreds of changes that could go on in one day. Very quickly you see that the time it takes to design a complicated thing like a car or a plane can be driven more by communications than changes to the design.
Q: How do these programs actually address the problem of inefficient collaboration?
A: Put this all together, and you'll find that it can take some auto makers 17 days to figure out the sum of these changes. With fully integrated design systems, we should be able to say that today, of the 10,000 parts, 300 were changed, so the weight increased by four pounds, the cost went down by $18, and we added three suppliers as a result. That's going to require digitalization of all the elements of the product--not just what the physical attributes of the bolt are, but all the supply information that's associated with the part.
Q: What's the payoff of these design advances?
A: Think of it more in terms of quality, rather than cost savings. Consider what a 1980 model car was like. Now think about the features available today: air bags, stability control, and advanced audio and navigation systems, let alone simpler things like power windows and more comfortable seats. These innovations are all a product of computer-aided improvements in design and manufacturing. Also, cars are more reliable now--warranties last longer, and cars aren't serviced as often. And the average car today costs less, as a percentage of average income, than ever before.
Q: Why haven't engineering advances turned into bigger profits for the car makers?
A: It's what I call the Red Queen theory from Alice in Wonderland: You have to run as fast as you can just to stay in the same place. Everyone is adopting these technologies. The companies that don't, fall by the wayside.
Q: How are suppliers being affected by the move toward more integrated design software?
A: Manufacturers are pushing more design and development work down the supply chain. Say you're building a plane. You'd like to have one company completely design all the seats, and have that company feed the seats to someone who is building the whole cockpit. The ability for one supplier to make design changes and then share them electronically [with other suppliers] promises huge new efficiencies. For example, Sikorsky Aircraft Corp.'s S-92 helicopter has leading-edge components, and it's being sought after by both commercial and defense customers. Sikorsky uses CATIA to coordinate five companies from Japan, Brazil, Germany, and Taiwan to design it. With enough computing power--which isn't so far off--designers working at different companies could pause at midnight for five minutes while the whole aircraft is reassembled using the latest version of all the parts.
Q: Integrating all the different software systems sounds fairly complex....
A: You're right. Consider just one aspect of changing a design of a new car. If a designer saves a dollar on a new part, but the company has to add a supplier, you'd like to know whether the price of adding that supplier--communications and manpower costs--is worth it. In the past, engineering software alone didn't recognize that cost. So design software has to be tied to supply-chain software. Gluing together systems like this is difficult. That's why industrial companies now spend fully half of their IT budgets on services--like consulting, installation, and integration of systems--and the rest on hardware and other products.
Q: So far, you've described a link between companies. What about getting more out of the information internally?
A: At any manufacturer, there are scores of people who will take a product's specifications and generate derivatives from that--things like replacement parts lists, instruction manuals, maintenance manuals, or warranty terms. Right now, much of that work reproduces diagrams and data that are already in the design system. So if you had a fast-enough computer, and the data properly integrated, you could generate the technical drawings for the maintenance manual, quickly test for safety, and so on. The base information used to design a product can be used to make the sourcing, manufacturing, use, and service of that product more efficient.
Q: Any examples of how companies integrate these sorts of data?
A: There are a lot of promising developments in the area of maintenance. At John Deere & Co., for example, they're creating Web sites by tapping into their design databases to teach customers how to use and service their equipment. Today a customer can go to a Web site and look for a specific model and find an accurate parts list or see animated service instructions. Those lists and diagrams might in turn be linked to the manufacturer's parts supply system, so a farmer could quickly order a missing component. Or take it another step: the manufacturer's Web site might be integrated with its dealers' inventory lists. No one is all the way there yet. But that's the direction for many industries, including aerospace, airport maintenance, and farm and trucking equipment.
Q: Some critics contend that computer-designed goods have a cold regularity to them. Do you agree?
A: In the past things looked boxier. This is where Dassault Systemes, our partner with CATIA, has led with new features like the ability to design from a sketch, or mix in complex materials or surface features to a design. Think of Nike shoes, Braun home appliances, or cars from Rolls Royce, BMW, and Volkswagen--all designed using CATIA.
Q: Beyond integration, will there be new enhancements that change the way products are designed and built?
A: Today, we can model a car crash in the computer using what we know about the properties of the parts. But currently it takes about two hours to simulate a crash test. With greater computer speed, we'll see crash tests go down to minutes. In consumer products, this would allow an engineer to change the screws on an oven door and immediately see if the hinge becomes weaker.