Just-in-time manufacturing may sound like one of the best and easiest means available to reduce the cost of producing products, but it is generally misunderstood and poorly utilized, says a professor at the Massachusetts Institute of Technology who studies the layout and structure of factories.
"Too many factories are built with insufficient thoughtful application of just-in-time ideas," says Stan Gershwin, senior research scientist in MIT's Department of Mechanical Engineering. "People think in terms of sound bites. Someone says inventory is bad so they design a factory without inventory."
But without having any in-process inventory on hand, a factory will run at far less than its optimum capability. "They get rid of the in-process inventory without realizing what its function is supposed to be, which is as a buffer to reduce disruptions by reducing the propagation and influence of one machine's failure on another machine," says Gershwin.
Hewlett-Packard learned this lesson thanks to the help of Mitchell Burman, one of Gershwin's students in MIT's Leaders for Manufacturing program. Burman analyzed a manufacturing plant being designed by HP and estimated the company would see a throughput that was only 20 percent of what was projected, due in part to the lack of space set aside for in-process inventory. At Burman's urging, the company redesigned the facility to add moderate-sized areas for inventory. As a result of the changes, HP claims that it is saving $10 million or more per year.
"Burman's throughput production index ended up being about right," says Gershwin. "By putting in the buffers, [HP] got twice the production rate they would have had without it." Since doing this work, Burman has started a manufacturing consulting company in Massachusetts specializing in such topics (Analytics, Inc., 617-264-9300).
From Gershwin's perspective, not enough manufacturers have a sound understanding of how just-in-time really works. The way an efficient factory operates is not by eliminating inventory, but by reducing disruptions. "The amount of inventory space has to be comparable to the duration and frequencies of disruptions," says Gershwin. "If you want to reduce inventory, you want to reduce disruptions and their duration. If you do those two things together, then you've got something. If you just reduce inventory without reducing the disruptions, then you will fall far short of your goals. If you reduce your disruptions, you will achieve your goals with excess inventory which you can then reduce." In many cases it might be better to have a redundancy of machines as opposed to a redundancy of inventory, he adds.
The common measure of factory utilization rate also might be flawed when considering the most efficient way to produce products, says Gershwin. If a factory is producing a product that has a variable demand, then having a high utilization rate "probably means the company has a lousy service level or is doing a poor job responding to customers," he says. A better measure may be a factory equivalent of on-time delivery of products.
Gershwin, who heads up MIT's Leaders for Manufacturing's "Design and Operation of Manufacturing Systems" research program, believes there is too little attention paid to how factories are laid out. "I want factory designers and managers to understand factories the same way airplane designers understand airplanes," Gershwin says. "People don't design airplane wings with software in black boxes and quick fixes. The software is a refinement of their intuition. Factory designers and managers just seem to be ... clueless." From Gershwin's perspective, there has simply not been enough research attention paid to the layout of factories --industrial engineering-- as there has been in such areas as physics and mechanical engineering.
Gershwin can be reached at 617-253-2149, e-mail (gershwin@mit.edu). You can take a look at the LFM's Design and Operation of Manufacturing Systems homepage at http://web.mit.edu/manuf-sys/www/LFM5-MAIN.html .
Reprinted with permission of Manufacturing News.