Elite Factories
By Abrahm Lustgarten

(FORTUNE Magazine) – A common thread connects this year's standouts: They all responded to crisis by radically changing the way they make things. Autoliv reprocessed its factory to fend off stiff competition in airbags; biotech giant Amgen engineered a solution to soaring drug demand; and Raytheon answered a call to arms by quadrupling its missile production.


Learning lean production--and making it stick--took intensive work.

When Toyota Motor Corp.'s Takashi Harada arrived at the Autoliv airbag module factory in April 1998, he made a quick survey of the plant's operations. Then the Japanese production specialist posed a question to production supervisor Bill Webb: "On a scale of one to ten, how strong is the facility?" Webb pondered his answer. Autoliv was a leader in automobile airbag manufacturing. It had a commanding share of the market, and Toyota was a big customer. Autoliv's production system, assembling the airbag modules on linear automated production lines, could stand some improvement, but the company was successful. "A three?" Webb suggested, humbly low-balling his new mentor. "Maybe a minus three," Harada replied.

So Autoliv, based in Ogden, Utah, embarked on its education in the Toyota production system with plenty of room for improvement. It set up a system for soliciting employee suggestions and putting them to work, allowing the creativity of the workforce to drive improvements in efficiency and safety. In inventory management and logistics, Bill Webb's areas, the company set up a pull system to replace parts on the assembly line as they were needed. None of this was new, of course. Toyota had been practicing and teaching kanban inventory control, just-in-time parts delivery, and lean production for a decade. But it wasn't easy to learn. Autoliv had made changes on its own in the past, but they weren't as effective as the company had hoped. With manufacturing defects rising, Autoliv accepted Toyota's offer to help out, and the partnership paid off. Last year Autoliv was awarded Utah State University's Shingo Prize for Excellence in Manufacturing.

Founded in 1956 and originally known as Morton Automotive Safety, it became an early leader in manufacturing airbags after developing a highly technical pyrotechnic device to inflate the bag (the company was acquired by Sweden's Autoliv AB, a leader in seat belt retractors and safety restraint systems, in 1996). However, competitors caught up, and Autoliv, under pressure from its automaker customers to reduce prices, slashed costs to protect its margins.

Until the drive to raise profits, Autoliv had no reason to push for long-term gains in productivity, and efficiency suffered. Webb's inventory operation, for example, held $23 million in parts--seven to ten days' worth--in a giant warehouse serviced by automated guided vehicles (AGVs), but the operation was inflexible and unwieldy. Then it replaced the AGVs with forklifts. Webb pushed mountains of inventory onto the assembly floor, but he was never sure exactly when it was needed, and sometimes he hauled it back into the warehouse at the end of the day. "Autoliv must be a very rich company," Harada told Webb one day. "You have lots of material that is not going anywhere."

Under Harada's tutelage, Webb set some rules for carrying parts to the production floor. He replaced the massive crates with small plastic boxes called totes and parked the huge Cushman forklifts in favor of smaller, golf-cart-sized tugs. That meant operators would make more trips, carrying 55 totes in a 45-minute period, and stay in closer sync with production. Webb also rearranged the inventory room so that it mirrored the production floor and then created software to track parts automatically as they were being used. The data were communicated back to the warehouse so that replenishment orders could be automatically sent to the cells and simultaneously relayed to outside suppliers so that they could ship new stock. Tug operators made their deliveries every 24 minutes, 36 times a day. The numbers are not arbitrary; they are designed to be flexible and easily divisible.

After making the changes urged by Harada, Webb learned he was holding far too much inventory. He cut his stock in half, to the delight of the bookkeepers. "We stopped ordering parts when we made those changes, and the cash went straight to the bank," says plant manager Mark Jenkins.

Walk inside Autoliv's 350,000-square-foot facility today, and you feel like you've stepped inside a clock. You can almost hear the ticking. Eighty-eight compact, U-shaped production cells have replaced assembly lines on the main floor. Each consists of a group of workstations staffed by a handful of employees. A screw is tightened, the finished piece scanned and registered in inventory, and then it's handed off to the next associate, who tags it and drops it into a box to be picked up and shipped. There are cells for driver airbags, passenger airbags, and side-curtain airbags, whose sales have grown 50% annually during the past three years.

Every 24 minutes, loud rock music from the 1970s group Steam blares through the public address system: "Na, na, na, na, hey, hey, hey, goodbye." It signals job change: Each person rotates down the line to a different task, and a tug carrying totes beeps twice and comes roaring out of the inventory room.

Changing the plant to just-in-time was one thing; then Autoliv had to persuade suppliers, some of them 1,500 miles away, to change with them. Autoliv built a transfer dock in Indiana where shipments from suppliers in the East could be collected and loaded onto identically packed semis to make the trip to Ogden seven times a day. The trucks are evenly spaced and arrive like clockwork every 72 minutes, always bringing the same items in the same order at the same time. For every truck that comes in, three tugs deliver to the factory floor. Autoliv only keeps parts buffers amounting to two days' production on-site.

Today the company continues to make changes under what it calls the Autoliv production system. Defects per million in module parts have been reduced dramatically from more than 1,100 in 1998 to just 16 last year. "Their plants are as good as any in the world today," says Ross Robson, administrator of the Shingo Prize. Last year Autoliv reported $1 billion in profits on $5.3 billion in revenues.

One of the lessons Webb learned from Harada is that the journey to lean production is an endless one. When Harada was leaving the Autoliv plant to return to Japan in 2001, he posed his question to Webb again. "How strong is the module facility?" Webb replied, "Maybe now it's a three?" "No," said Harada. "Now it is zero." In the religion of continuous improvement, there always is room to grow.


Coping with surging drug demand meant a sharp ramp-up in protein production.

It's not rocket science--it's actually more complex. The biological expertise and technical skill required to produce Enbrel, Amgen's blockbuster recombinant protein drug for the treatment of rheumatoid arthritis ($1.3 billion in sales last year), makes it one of the most complex manufacturing processes in the world. Amgen not only mastered that process against steep odds in its current West Greenwich, R.I., facility, but has improved it in a groundbreaking new factory that is due to open next year.

When Amgen acquired Immunex, the maker of Enbrel, in 2002, 40,000 patients were waiting for the drug, but Immunex hadn't finished construction of a facility to produce it. Amgen had to take over production and get Enbrel to market in the U.S. quickly. It did so by retrofitting a tired Rhode Island bio lab into a state-of-the-art protein plant. The plant, named BioNow, was renovated, approved for production by the Food and Drug Administration, and put into operation in a speedy four months. Meanwhile Amgen was applying lessons from BioNow to the construction of a new plant next door called BioNext, before BioNow reached full production. Even for the world's largest biotech company, with 13,000 employees and $10 billion in revenues, it was an enormous task.

For Amgen, manufacturing Enbrel means raising a family of complex organisms in larger quantities than ever attempted. Mammalian protein production --the cultivation of proteins from engineered cells from mammals--used to be done in amounts measured in kilograms. Now, says Fabrizio Bonanni, Amgen's vice president of manufacturing, the company wants the ability to produce proteins in tons. Enbrel sales rose 400% last year and continue to climb following the FDA's approval of the drug to treat psoriasis.

Soon after acquiring Immunex, Amgen poured $500 million into the renovation of the 250,000-square-foot BioNow facility. With that under its belt, Amgen immediately undertook an expansion, building the 500,000-square-foot BioNext facility adjacently. Constructed from scratch with bioreactor tanks ten times the size of the industry norm, BioNext will be the world's largest and perhaps most efficient recombinant protein factory when it is scheduled to start up in 2005.

Recombinant proteins come from biologically engineered cloned cells (Enbrel is derived from a hamster) that are predisposed to maturing and secreting the specific protein. While Amgen keeps the master gene line--its ticket to the $1.7 billion Enbrel market--frozen in an undisclosed location, qualified clones are farmed at West Greenwich. Keeping the cells healthy while they multiply requires an absolutely sterile environment. The proteins must be fed and cared for while they multiply, until Amgen stops the growth, catalyzes the protein production, and tosses out the leftover cells.

Executing that process once is an achievement; repeating it in a factory environment is what Amgen excels at. "As large molecules go [Enbrel is roughly 2,000 times the size of a chemical molecular drug like aspirin], Enbrel is one of the most complicated to make, and so it is also extraordinarily expensive to produce," says Bain consultant Charles Farkas, a health-care manufacturing expert. "It is a cooking process, and it has the potential to come out a little bit differently every time because it is a living organism."

The core of the BioNow plant, about 50,000 square feet, is at least 100,000-class clean--meaning everyone wears head-to-toe white bunny suits to control dust contamination and keep the air-particle count within bounds. Each time the cells are transferred between reactors, they flow through a series of pipes and valves made up of hundreds of connections and O-ring seals. "It only takes one," says senior quality director Kimball Hall, referring to a molecule of bacteria from human skin or breath that might penetrate a faulty seal and ruin an entire batch. "If you breach that environment, that's it."

The new plant shows a maturity of design that evolved from the lessons learned across the company's construction-strewn courtyard. The process and even most of the hardware are identical to those at BioNow. But BioNext's layout, organization, and accessibility incorporate a well-thought-out plan to cut costs (they won't say by how much) through enhanced efficiencies. Amgen is reluctant to admit that cost and productivity pressures weigh heavily on the industry. Still, many of the improvements expected to come online in the new factory will mean a 300% increase in capacity (22 batches will run simultaneously in BioNext, while BioNow continues to run eight to ten) and lower operating costs from energy bills to cleaning.

Here, small things make a difference. In BioNext, for instance, gravity-fed reactors are built to flow from the top floor down, eliminating the piping and pumping, and associated energy and maintenance costs. Unlike BioNow, where everything is contained within sterile spaces and requires extensive cleaning, large reactors at BioNext are constructed behind walls, with access portals in the clean rooms. That places massive amounts of hardware and huge unused spaces outside the clean rooms, which are extremely expensive to maintain. Amgen estimates capital savings of up to $400 per square foot, as well as savings from the elimination of energy-intensive, high-efficiency particulate-air filters and refrigeration. Centralized servers gather data from tanks across the facility, a process that would otherwise call for at least manual switching and an on-site check to be centrally controlled. Even hand-cranked valves are gradually being replaced by hydraulic computerized releases. The equipment itself is built on oversized, spacious skids that are easy to reach inside of for maintenance; they make line-of-sight quality checks possible. "We had the luxury of space, and we took advantage of it," said plant manager Mike Covarrubias.

BioNext could supply as much as 10% of the world's mammalian recombinant protein supply when it comes online. It will be needed. Analysts project Enbrel will see at least 17% sales growth during the next three years, rising to $2.5 billion in revenue by 2007.


Making a missile every eight minutes required a total manufacturing overhaul.

The second hijacked airplane had scarcely struck the World Trade Center on Sept. 11, 2001, when the phones began ringing at Raytheon's Tucson missile factory. By December, Raytheon had won a $156 million contract from the U.S. Air Force to make 10,000 new kits for the Paveway laser-guided bomb--a two-foot-long module containing the laser and GPS guidance that tells the bomb where to go and flies it there. Before they were done, Paveway kit orders would leap fourfold.

Unfortunately ramping up production wasn't as simple as asking employees to work harder, adding another shift, or ordering more parts. In August 2001, Tucson was producing about 350 missiles a month--roughly half its maximum capacity. The factory was cramped. Inventory was stacked between the delivery area and the production line, with barely enough room left to squeeze a forklift through it. "We needed four times the amount of material on hand to run the factory," said production operations director Ricky Nelson. "The reality was we weren't going to be able to carve that out."

For a solution Raytheon went back to the lean-production playbook. It distilled its assembly process down to the barest value-added components and rooted out waste--time spent searching for misplaced items and moving materials around the plant. It solicited suggestions from workers and experimented with new solutions such as an assembly line built by employees. It eventually increased inventory turns 400% and reduced the cost per missile by 40%. For its success, it received the 2004 Shingo Prize.

Raytheon, which invented the guided bomb, is the world's largest weapons systems producer, with about 43% of the domestic market. More than half the missiles fired on the opening night of the Iraq war in 2003 were made by Raytheon. But it learned lean production from Hughes, which it had acquired in 1997, as well as from some others. Hughes had spent $20 million reengineering its infrastructure. It had stripped factory floors of their fixtures, moved plumbing, electric, and communication lines to exposed overhead routing fixtures so they could be easily relocated, and put wheels on every piece of machinery to create an open, easily arrangeable plant floor adaptable to changing needs. The ideas were not new in the manufacturing world, but they were new at Raytheon. "It all seems pretty obvious now, but we had many years of assembly heritage built on how America did manufacturing," says president Louise Francesconi. "I'll call it the unmaking of that. It was really revolutionary."

At the Tucson facility, workers install global-positioning and laser-guidance systems into bullet-shaped missile heads. About a dozen people work at four or five mini-stations in the main assembly area. Before 9/11, Paveway operations manager Pat McKinney had already been working to simplify and shrink his subassembly stations. The line had been made more efficient, but it was designed for only 350 missiles per month. Pushing for 1,500 a month forced more thorough analysis and drastic action.

McKinney began counting the minutes in a workday to identify his optimal Takt time--the period needed to produce a single kit--and set a rhythm for the factory. "If I were going to get to 1,500 missiles a month," he recalled, "I needed to build one every eight minutes." Under the old system Paveways were assembled at individual workstations that each had its own tool sets and parts supplies. McKinney split the line into three sections so that each assembler does only a portion of the work before passing it along to the next station. He also moved the stations closer together so that the work in progress could be passed back and forth. Now each operator uses one-third the amount of tools and inventory in less space. "We literally hauled away semi truck--loads of excess tooling and benches," McKinney says.

Even under the new arrangement, four workers were required to pass 45-pound guidance head units down the line by hand. Repeating that task 150 times meant that each worker was lifting more than three tons daily. "It was really physical," says Wayne Minnerd, a beefy man with a gray goatee and a sleeveless T-shirt. "All I could do was drag my butt home at the end of a shift." After work one day, McKinney's staff went to Home Depot and bought about 100 feet of plywood and five dozen rubber caster wheels. They used them to build a missile trolley: a thin strip of board with 60 or so wheels that cradles the Paveway assembly. For less than a couple of hundred dollars Raytheon eliminated lifting and reduced cycle times by 30%, as well as streamlined production and substantially decreased the chances of on-the-job injuries.

But plywood and caster wheels alone weren't enough to get Raytheon to its production goal. So McKinney targeted non-value-added activity like parts flow and transportation. Missile pieces were traveling 2.6 miles from the time they arrived on the delivery dock, went through the assembly process, and then were sent back out to the shipping dock as a completed product. Today they move 700 yards. One employee who wore a pedometer says he now walks one mile a day at work instead of 30 miles.

Suppliers were eager to cooperate. In the past they built a part, bundled it in bubble-wrap, and then taped it into a box for shipment. The process then reversed when the part arrived at the Tucson plant. To simplify the process Raytheon devised reusable containers with precut padded casings. Once the company received a parts shipment, workers could empty it in a matter of minutes and return the casing to the supplier for a refill.

Changes have rippled through the rest of the plant. Parts that had been stocked alongside the assembly line are now stored in the trailers in which they are delivered. When a trailer is emptied, it is replaced with a waiting one and gets hauled off for a refill. Inventory has been reduced from four months' production to ten days. Inventory turns shot up from 3.2 per month in December 2000 to 12 a month this past February. All told, the Tucson plant is capable of producing 1,500 units a month, with less material than it used when it was making 350 per month.

Learning from its experience with Paveway, Raytheon is spreading the lessons of lean production to just about every facility on its 3.1-million-square-foot Tucson complex. The tank-fired TOW missile factory now uses simple lean refinements like modular, removable workstations to accommodate volatile production schedules. When the factory is down for as long as six months, the TOW operation is removed, and the space used for something else. When Raytheon redesigned its Tomahawk factory with a new layout, the entire floor was rearranged in a single eight-hour shift. Across the company rework has fallen 20%, and the number of finished products that pass inspection the first time jumped from 87% in early 2000 to 95% now. "As we continue to grow," Francesconi says, "our improvements will be in finding more and more ways to act small"--and get lean.

Stories from the Industrial Management & Technology section can be found at fortune.com/imt. Executives in manufacturing and research, and others eligible to receive FORTUNE's Industrial Edition, can subscribe by calling 888-394-5472.

FEEDBACK fortunemail_letters@fortunemail.com. Stories from the Industrial Management & Technology section can be found at fortune.com/imt. Executives in manufacturing and research, and others eligible to receive FORTUNE's Industrial Edition, can subscribe by calling 888-394-5472.