Finally, the accelerating rate of decline in sales suggested that the whole structure of the business, not just the size of the plants and the amount of hourly headcount, was no longer appropriate. Pratt as a whole desperately needed a rethink.
A Second Change Agent
George David was now carefully observing the crisis at Pratt because it was beginning to affect the whole of UTC. Historically, Pratt had been both UTC’s largest operating unit and by far its most profitable one. The sudden loss of profits at Pratt was now driving down earnings and share prices at the parent UTC, despite good performance from the other businesses.
As David looked around in the fall of 1992, he decided he needed a second “change agent,” someone to replace Pratt’s president, who, as a lifetime P & W employee, understandably reflected the traditional Pratt way of doing business. There was one obvious candidate, forty-three-year-old Karl Krapek, then serving as president of Carrier. David knew that Krapek understood lean thinking and he also knew that Krapek would steamroller any obstacle to get the job done. “Mr. Krapek,” he observes dryly, “is the most relentless executive at following up in the world today.”
We have now heard many accounts of how the “light came on” as managers first grasped lean principles. Krapek’s enlightenment started early but it was a full decade before he was in a position to put lean principles to work on a larger scale. After graduating from the General Motors Institute as an industrial engineer (and after a graduate degree at Purdue in the same subject), he was given increasingly important operating management jobs within GM. In 1979, at age thirty-one, he became one of the youngest assembly-plant managers in GM history, running the five-thousand-worker Pontiac assembly plant in Pontiac, Michigan.
One of the most striking features of the plant, as he noted upon first taking over, was the massive inventory of finished engines ready for installation. Indeed, in the deeprecession beginning in 1979, the Pontiac plant had a three-month supply of engines. This caused endless difficulties and it occurred to Krapek that the plant’s performance could be improved dramatically if engines were only made and shipped to the plant as actually needed.
He devised a plan to clear out the stockpiled engines and then get deliveries from the nearby Flint, Michigan, engine plant every thirty minutes, just as needed. The concept worked brilliantly as it got started and the positive effects on many aspects of plant operations were apparent. Krapek began to think of how to expand on this fundamental lean principle. Then disaster struck. A shipment from the Flint plant failed to arrive and the entire plant had to be shut down, sending the workforce home four hours early. Senior management at GM demanded to know how he had allowed his plant to operate with no buffers! Krapek was severely reprimanded and threatened with being fired.
After an appeal to a higher level, Krapek was allowed to continue in his job, but he suddenly understood what many managers have discovered before and since: It’s impossible to introduce lean, flow concepts piecemeal and in an organization where the senior management doesn’t understand them and where the very structure of the organization doesn’t support them. When George David, then at Otis Elevator, called with a job offer, Krapek was ready to depart for an organization he hoped would be more capable of change.
Perhaps the most fortuitous aspect of moving to Otis was the firm’s location in Hartford. When Krapek first heard about the events at Jake Brake and other Danaher companies in 1987, he took a personal interest. However, because nearly 80 percent of the “manufacturing” conducted by Otis was at the construction site where the elevator was being installed, it was not immediately apparent how to apply lean principles.
In 1990, when Krapek moved from Otis Elevator to become president of Carrier, he inherited a true manufacturing challenge where very nearly 100 percent of costs were incurred inside Carrier’s plants or inside those of its suppliers. He was prepared by his early experience at Pontiac to accept lean thinking, so he consulted with Art Byrne on what to do and retained Iwata, Nakao, and their associates to help. They quickly began to convert the operation from departmentalized batches to cells for single-piece flow and made dramatic progress.
When the phone rang in the fall of 1992, Krapek was ready and able but unenthusiastic. “George David called and said, ‘You have to go to Pratt.’ We were doing great things at Carrier but were only part of the way along in the lean conversion. I said I wanted to stay. In addition, I told him, ‘I came from General Motors and I don’t want to go back to General Motors.’ I meant that I did not want to return to a highly departmentalized, rigid bureaucracy trying to operate as it always had in a totally changed world. But David pointed out, ‘You’re no longer middle management as you were at GM. You will be the president. If you don’t want Pratt to be General Motors, change it into Toyota or something even better!’ I really had no choice, so I went.”
When Krapek got to Pratt at the end of 1992, he knew he had to devise a dramatic plan to reconfigure the whole company and implement it very quickly. A new analysis of market trends showed that new engine sales had practically come to a halt and that shop load was heading for 5.4 million hours by 1994, down 50 percent from the peak in 1991–92, and that it might never rebound much from this point. However, the multilayered, departmentalized structure of the company, with all its associated overheads, had not changed and nothing flowed easily across the functional and departmental walls. What’s more, Pratt was still trying to do too many things itself.
Krapek’s first action was to speed up an evaluation already launched by Coran to determine which physical activities Pratt should be performing. As a result, sheet-metal forming, the fabrication of steel engine discs, and the manufacture of gears and gearboxes were soon contracted to suppliers.
Next, the two thousand parts in a jet engine were grouped into seven product categories—rotors and shafts, turbine airfoils, combustors and cases, nacelles, forged compressor airfoils, compressor stator assemblies, and general machined parts. The old organization structure, based on plants, was abandoned, to be replaced by a new system of Product Centers, one for each category of parts plus an eighth center for Final Assembly. Each was given a general manager reporting to Coran; at the same time, the centralized purchasing, quality assurance, and detail part design functions in Operations and Engineering were reconfigured, with most employees reassigned to the Product Centers. This meant closing a large fraction of Pratt’s plant space and moving a substantial fraction of total manufacturing activities from one plant to another so that, for example, all of the production work involved in making a rotor could be conducted in nearly continuous flow in one large room in the Middletown, Connecticut, plant.
One great problem facing Krapek was that a massive, immediate reduction in Pratt’s headcount was required and some facilities in Connecticut had to be abandoned. As Krapek noted, “Our weekly output of three large engines and six small engines plus spare parts can literally be fitted into my office. So why do we need ten million square feet of manufacturing and warehousing space?”
In addition, Pratt’s union had to accept the notions of multiskilling, job rotation, multimachine operation, and continuous movement of jobs and work between plants to accommodate a changing value stream. By contrast, as of 1992, almost all hourly workers tended a single machine and simply watched as it conducted its operations, interceding to gauge parts as appropriate. They were constrained in their scope of activities by the division of labor into 1,151 union-sanctioned job classifications—or about one job classification for every ten hourly workers—and jobs were assigned on the basis of seniority through an elaborate “bumping-rights” system which often caused dozens or hundreds of job reassignments when the pattern of working was modified only slightly.
George David and Karl Krapek conducted a series of high-level negotiations in the spring of 1993 with the International Association of Machinists and with the State of Connecticut before finally reaching an agreement that hourly headcount would be reduced permanently (
with total Pratt head-count falling from 51,000 in 1991 to 29,000 by the end of 1994), that flexible working and active participation in job design and the development of standard work would be the new norm, and that the state would help with retraining of displaced employees. In return, Pratt agreed that as long as the ambitious productivity improvement targets were met, no more work would be outsourced to suppliers or moved to Pratt operations in other states.
Remove the Anchor-Draggers
A second great problem facing Krapek and Coran, with the downsizing and labor-management issues resolved, was that Pratt’s existing managers either couldn’t or wouldn’t operate the new Product Centers. Although three of the eight general managers of the new centers announced in August 1993 were from outside Pratt (all with cellular manufacturing experience at General Electric) and knew what to do, many of the Pratt old-timers couldn’t seem to get it.
The problems were of two sorts. At the North Haven turbine airfoil facility, the longtime Pratt managers really threw their hearts into change and attempted a very ambitious move from batches to single-piece flow, but they simply didn’t have the skills to pull it off. The backlog of orders grew alarmingly and customers began to scream.
Traditionally, managers caught in this predicament at Pratt were fired. (The slogan among parts plant managers had always been, “Ship on time and you’ll be fine [even if you’re shipping junk].”) However, Mark Coran was determined to instill a new spirit in which managers who earnestly tried to manage in a new and better way would not be punished for failure. He therefore moved the plant management to other jobs in Pratt and went outside to find Ed Northern, a former GE manager with extensive experience in lean operations, to carry through the lean transformation.
The other problem was that some general managers simply refused to change their methods. In the spring of 1994, Chihiro Nakao had conducted some more theater in the main assembly hall at Middletown when he walked in, quickly looked around, and then told the general manager of assembly that the time needed to assemble an engine would need to be reduced from thirty days to three, the space needed would have to be cut in half, the amount of human effort required would need to be cut by two thirds, and inventories of parts and engines on hand would need to be reduced by more than 90 percent. What was more, assembly of these massive machines would need to be converted from bench assembly to a moving track in continuous flow. And it would be necessary to start immediately.
The general manager and his deputies argued that this was simply impossible to do quickly for such a complex product in such a complex organization as Pratt & Whitney, using highly skilled craftsmen to correct mistakes made farther upstream. They promised to work on a long-term plan, but it was apparent that nothing would happen soon; shortly afterward they were asked to leave Pratt and Bob Weiner, another Pratt outsider, was installed as the new general manager.
Over the three-year period from 1991 to 1994, the number of senior managers in Pratt’s Operations Group was reduced from seventy-two to thirty-six, and only seventeen of the remaining thirty-six were with the company as of 1991. To make the lean transformation happen in this extraordinarily in-grown organization it proved necessary to replace a much higher fraction of the management than in the other organizations we have examined.
Fixing the Two Key Activities
Because Pratt conducts two basic activities in physical production—fabrication of individual parts from castings or forgings and assembly of these parts (along with many more from suppliers) into complete engines—the physical transformation of Pratt that followed comes clearly into view if we look for a minute at what Ed Northern did to transform turbine blade fabrication and what Bob Weiner did to transform final assembly.
The Billion-Dollar Room
Ed Northern manages a single, vast room, in North Haven, Connecticut. It measures 1,000 feet by 1,000 feet and can be easily surveyed from the front door. In this room in 1991, 1,350 Pratt employees used 600 sophisticated machines to manufacture $1 billion worth of turbine blades and guide vanes for jet engines. 34 Because jet engines themselves are usually sold below cost—indeed, in some recent cases, practically given away—and because the frequently replaced guide vanes and turbine blades (often called the “razor blades” of the jet engine business) are sold at multiples of their actual production cost, what happens in Ed Northern’s room largely determines whether Pratt & Whitney can make a living.
The problem in 1993 was that North Haven’s costs were so high that Pratt was not garnering enough profits on its “razor blades” to sustain its “razor” (jet engine) business. Even worse, in the effort to switch over to lean methods, North Haven was failing to meet its shipping schedules. Back orders were soaring and Pratt’s cash flow was severely affected. When Ed Northern first walked into the room in August 1993, he faced a life-or-death task.
Ed Northern’s light, like many others we have met, came on in the early 1980s, in his case at the GE Aircraft Engine Group where he first tried single-piece flow. He had some early successes but eventually left for Inter Turbine, a small firm making a living by repairing damaged turbine blades for airline maintenance shops. Inter Turbine, however, lacked the technology base or financial resources to move beyond a narrow market niche, so when Mark Coran called in the summer of 1993, promising Ed complete freedom to institute lean methods at North Haven, he readily accepted.
The room Ed Northern first saw was laid out in the “flow” lines introduced in 1984, except that changing part designs and processing needs had run head-on into the massive, immovable processing machines so that whatever flow had been achieved in 1984 had become a series of dams and stagnant pools by 1993. In addition, he found truly appalling quality. In many processes, first-time-through acceptable quality was less than 10 percent. Parts were going through the system over and over and it was impossible to meet the production schedule.
Northern immediately took a series of steps which we hope are becoming familiar. He assessed his headcount and determined that he would never need more than 60 percent of his 1,350 workers. At the same time, he surveyed the line management and found that a substantial fraction would never be able to work in the environment he planned. A onetime headcount reduction and rapid management changes quickly produced a personnel level he knew he could defend and a management team he could lead.
The next step was to construct a value stream map for the entire turbine blade and guide vane business, to reconfigure the business units so they precisely channeled the flow of value for each product family, and to reconfigure every machine so it could be moved easily at any time by the work-force. 35 Then it was time to move the machines into cells laid out in the same sequence as processing steps so that single-piece flow occurred in as many cases as possible.
The results were immediate and startling. Over the next two years, overdue parts fell from $80 million to zero, inventory was cut in half, the manufacturing cost of many parts was cut in half as well, and labor productivity nearly doubled. In short, just what we would expect. But then it was time to confront the monuments problem .
The Monument of Monuments
Lean thinkers call a “monument” any machine which is too big to be moved and whose scale requires operating in a batch mode. (They would apply the same term to a hub airport, a centralized computer system, or a centralized engineering department—to anything that requires batches to operate and can’t be moved as the value stream changes.) Because continuous improvement and changing processing requirements require the continuous movement of machines, monuments are evil, another form of muda.
The monument in question in North Haven was the massive, $80 million complex of twelve Hauni-Blohm blade grinding centers, custom-made in Germany and installed in 1988 as Pratt attempted a high-tech leap over its competitors. The idea had been very simple: Totally automate the grinding of the blade roots for turbine blades using the world’s fastest and most sophisticated equipment.
Prior to the late 1980s, North Haven h
ad placed each blade in a series of nine grinding machines for a total processing time of eighty-four minutes. The objective was to grind smooth the base of each turbine blade so it would snap snugly into the disc holding it in the engine. This approach was labor-intensive, due to direct labor needed to watch machines, conduct frequent gauging, and position parts in machines. In addition, indirect effort was needed to move parts from machines to storage areas and then to the next machine, now located some ways away in the degraded “flow” system.
The new system used twelve massive grinding centers with twelve axes of motion. Each center could perform all of the grinding steps formerly accomplished by nine machines and could grind a blade in only three minutes. What was more, the centers were fed and unloaded robotically and the parts were carried to and from storage by an automated guided vehicle (AGV). No direct or indirect hourly labor was required.
Still, there were problems. The forces applied to the blade by the grinders were so severe that the blade was destroyed if held by standard positioning fixtures which concentrated the tremendous forces at a few points on the blade. Therefore, it was necessary to encapsulate the blade, excepting the area to be ground, in a low-temperature alloy to spread the forces evenly over the whole blade. Encapsulation, conducted by a machine with a large vat of liquid metal, expensive molds, and long changeover times, was a batch process, so it was necessary to take the encapsulated parts to a storage area until they were needed by the Blohm machines. This task was handled by AGVs and an automated storage and retrieval system. (ASRS, as it was called, was identical in concept to the system Toyota tried in its Chicago warehouse, as described in Chapter 4 .)
Lean Thinking Page 21