HARVARD GAZETTE ARCHIVES
By William J. Cromie
People have been trying to solve the problem since they began making clothes. How do you lay out pattern pieces for a shirt or pair of pants so that you waste the least amount of cloth?
If you are a manufacturer of intimate ladies' apparel, for instance, the problem burgeons to the best layout for 508 pieces on a bolt of cloth 62 inches wide.
"The possibilities are staggering," says Fred Abernathy, Abbott and James Lawrence Professor of Engineering. "And, of course, you have to come up with a good solution in a limited amount of time. Then when sizes or styles change, you must do it all over again. People around the world have spent a lot of time and money on this problem."
While doing a study of the U.S. apparel industry, Abernathy and colleagues at the Harvard Center for Textile and Apparel Research decided this was a task naturally tailored for computers. They teamed up with computer scientists and invented an automatic system for quickly putting pieces in place.
The University has licensed the system to Gerber Garment Technology Inc., which, fittingly, is a subsidiary of a company that invented the first automatic machine for cutting cloth. The layout program produces numbers that control the cutting movements of these machines. Gerber sells the layout systems to clothes makers such as Bali, Haggar, and Wrangler.
"That's only one application for the technology," Abernathy points out. "It also can be used for leather, wood, or metal. One manufacturer is interested in it for cutting carpets, another sees possibilities for efficient layout of automobile body parts and upholstery."
Sum of the Parts
The project began in 1990 when the Alfred P. Sloan Foundation gave the University a grant to research the apparel industry and come up with specific recommendations for improving performance and competitiveness. The investigators included Abernathy; John Dunlop, Lamont University Professor Emeritus; Professor Janice Hammond of the Business School; and David Weil of Boston University School of Management.
"We found the primary driving force of the industry to be product proliferation," said Abernathy. In other words, the number of different clothing items carried by stores almost doubled from 1989 to 1992.
"To fill customer demands, a typical shirt manufacturer may make 25,000 different types of shirts in a year when you consider different neck and sleeve sizes, cuff and collar styles, and fabrics. A jeans manufacturer may turn out 40,000 different items, including tight or baggy varieties, fabric styles like stone-washed and various waist and length sizes. Each item requires a change or a new pattern layout.
To help address pattern proliferation, the researchers sought out Victor Milenkovic '81, then an assistant professor of computer science. Now at the University of Miami, he is an expert in computational geometry, a field that involves finding rapid solutions to tasks such as positioning straight-sided figures in a restricted space.
"When they came to me with the problem, I thought, 'Beautiful! This suits both my experience and my temperament,' " Milenkovic recalled. "I wanted to use what I knew about mathematical reasoning and computer-aided design to solve a real-world problem, and do it soon."
Milenkovic, with the help of graduate students Karen Daniels and Zhenyu Li, developed an "automatic marker-making tool kit," computer software that lays out markers, or patterns, for pants, blouses, jackets, or any other type of garment.
"It can simultaneously move all the pattern pieces to generate a more densely packed layout," notes Daniels, now a postdoctoral fellow. "It also eliminates overlaps, which is important for handling style changes, such as larger shoulder pads. And it fits items into a given space, or tells users that the items won't fit."
Abernathy compares such activities to "solving a huge jigsaw puzzle by moving hundreds of pieces in and out of thousands of combinations."
For some layouts, the compaction feature of the automatic marker maker cuts waste by almost 2 percent. That multiplies to a huge savings when a marker is used to make thousands of items.
"We found that the software could lay out 108 pieces, or 12 pair of pants, in seven minutes versus an hour for a human," says Daniels, who did her Ph.D. work on the system.
Harvard copyrighted the marker maker software and licensed it to Gerber in 1993. Depending on the computer hardware used with it, a system costs between $50,000 and $100,000.
"Facilitating product proliferation this way contributes to changing the style of retailing in the apparel industry," Abernathy notes. "Specifically, it minimizes inventory and makes retailers more responsive to consumers."
Previously, retailers placed orders for apparel six months to a year in advance. That involved a large element of uncertainty. What was not sold had to be discounted. Today, bar-code readers keep track of which items people are buying. The information in retailers' computers then can be sent to manufacturers' computers.
"In this way, manufacturers run the stores' inventory replacement," Abernathy says. "What's selling is what's replaced. It's very difficult to service that kind of retailing from 5,000 miles away, giving U.S. manufacturers a terrific advantage."
In China, Korea, and other less developed countries, patterns are laid out by hand. The cloth is then stacked and cut with vertical knives much like huge band saws. These and other tasks are done by people working for wages much lower than those paid in this country.
"Automatic marker making helps the U.S. to be more competitive in this kind of market," Abernathy says.
Harvard's license to Gerber covers so-called "sewn products." Besides apparel, such products include carpets. At least one carpet-maker is talking to the University about automating its marker making.
Harvard's Office for Technology and Trademark Licensing is also pursuing the possibility of additional contracts with automobile manufacturers, whose layout problems include metal body parts and cloth or plastic seat covers.
"Companies had been trying to solve the layout problem for 25 to 30 years," Abernathy points out. "Other computer systems were developed that can do some, but not all, of what ours does. For example, none is as effective at compaction, squeezing more pieces into a given space. One hundred to 200 items can be compacted in one or two minutes. The system nicely demonstrates how private funding of university research can result in new technology that makes U.S. industry more competitive in a global market."
Copyright 1998 President and Fellows of Harvard College