Lean Thinking

by Daniel T Jones

  F IGURE 8.2: J ET E NGINE D ELIVERIES & P ARTS S ALES

  In consequence, employment at Pratt was more stable than orders until 1990, as shown in Figure 8.3 . There were periodic layoffs, but these were likely to be short and it was easy for Pratt employees to think they would always have a job, particularly if they had a few years of seniority.

  When the Eagle First Came to Earth

  Big companies like IBM, General Motors, and Pratt usually receive (but ignore) a number of warnings that the world has changed before the roof finally caves in, and the collapse of both the military and civilian markets in 1991 was not Pratt’s first wake-up call. That came in 1984, when Pratt so infuriated military customers with its failure to fix operational problems with the F100 engine that GE was brought in as a second source of supply and given roughly half the U.S. military’s business for F16s. 26

  At the same time, the launch of Pratt’s PW2037 engine for the Boeing 757 infuriated airline customers. The Pratt engine’s fuel consumption was superior to that of the competing Rolls-Royce RB211-535 and pricing was competitive, but the Pratt engine had a terrible record of mechanical problems, causing flight cancellations when introduced into service. As Fred Hetzer, the project engineer on the PW2037, remembers, “We were like the aging slugger in baseball who can still see the ball clearly but can’t swing the bat fast enough to hit it. We knew about the problems in the PW2037 a year before they surfaced with commercial customers, and we worked day and night to fix them, but the organization was so sluggish and cross-functional communication was so difficult that we just couldn’t get them fixed in time.” As a result, Pratt had a superior engine ready first but wound up with only half the business in the forty-thousand-pound-thrust class.

  F IGURE 8.3: E MPLOYMENT AT P RATT & W HITNEY (000s)

  Finally, Pratt badly misjudged the trend of demand in the jet engine market. Thinking that large, double-aisle aircraft were the primary growth market and reluctant to compete against its currently best-selling JT8D, Pratt failed to develop a replacement engine for the JT8D powering the 727s and 737s. When Boeing decided in the early 1980s to modernize the 737 by lengthening the fuselage to carry more passengers and updating its systems, Pratt did not have an engine with modern, high-bypass technology and lower specific fuel consumption. A consortium formed by GE in the United States and Snecma in France (CFM) did and ran off with most of the business for what became by far the world’s best-selling airplane. When Airbus introduced the A320 to compete against the 737, 100-to-160 passenger, single-aisle jets became by far the largest aircraft market segment. 2 7

  Leaner but Not Lean; Necessary but Not Sufficient

  Suddenly, in the mid-1980s, Pratt faced competition in all its major product categories and its market share began to slip across the board. In addition, total engine deliveries in the industry began to fall due to the shift from four- to two-engine designs. Pratt’s management was not completely asleep and three innovations, which seemed earthshaking at the time, were introduced in response, one in production and two to bridge the chasm between product development and production.

  The major innovation in the physical production system, introduced in 1984, was the “focused” factory with flow lines and business units organized by categories of parts. Pratt’s factory structure emerging from three hot wars (World War II, Korea, and Vietnam) and one cold war was a hodgepodge of isolated shops working on parts with no relation to the part being made in the next shop. In one notable case, the distance traveled by a part within Pratt plants (not counting the distance traveled between plants) was measured and found to total eighteen miles.

  In 1984, Pratt reorganized its facilities so each would take responsibility for a major category of engine parts. The massive North Haven plant would work primarily on turbine blades while the Southington plant would work primarily on rotors and discs and the Middletown plant would take on all final assembly work. Within each plant, activities were further reorganized so that many of the steps in the physical processing of each category of part 28 were grouped and lined up in a logical progression in a “flow line,” insofar as tool designs would permit. Note that this is exactly the concept described in 1936 by Carlton Ward, the assistant general manager of production at Pratt.

  Finally, each part category—for example, high-temperature turbine blades for the JT8D engine—was placed in a “business unit” whose leader knew the cost for his operations. The business unit head was fully in charge of getting the parts made at cost and on time, in accordance with the master schedule (now run off a massive computerized Material Requirements Planning system).

  By the mid-1980s, Pratt’s senior managers were aware that as the jet engine matured it was becoming sensible to apply similar design principles to “standard” design problems confronting each category of part. For example, why not specify the same grade of chromium for each high-temperature turbine blade, rather than fiddling endlessly with minor changes in the alloy mixture, which produced negligible improvements in performance? Yet it was apparent that Pratt’s design engineers working on each category of part were doing the exact opposite. They were doing what comes naturally in engineering cultures far removed from the customer by endlessly reengineering designs in search of novelty and a better solution, no matter how slight the performance gain. In consequence, quite different production methods might be specified for practically identical parts, making their manufacture with the same tools in a common flow cell and business unit impossible.

  As senior managers became convinced that many novel designs were novel in name only while costing the company millions in spiraling development and production costs, a solution emerged in the form of cross-functional teams 29 to evaluate each part and process widely used in Pratt engines—for example, turbine airfoils—and to agree on “norms” for part design, material selection, and processing techniques. If any engineer wished to adopt a new design approach differing from the norm, it was her or his responsibility to convince the relevant team that it was superior. In practice, this system greatly reduced the number of novel schemes proposed and reduced costs.

  It was also apparent by the late 1980s that the project engineer system of weak coordination was producing poor results, so Pratt augmented this approach with a new system of Integrated Product Development (IPD) being promoted among major defense contractors by the U.S. Air Force. The idea was to form cross-functional IPD teams to resolve major cross-functional conflicts in engine development as they arose. This concept fitted nicely with Total Quality Management, a “program” also embraced under the moniker of “Q-Plus” by Pratt in the late 1980s.

  The results of these three innovations were significant but not sufficient. Time-to-market for the new PW4084, entering airline service in June 1995, shrank from five years under the old project engineer system without IPD to about four years with IPD, and the number of engineering hours declined by a similar fraction. Meanwhile, the new plant layouts dramatically reduced travel of parts within the production system but each step in the so-called flow lines still had a pile of inventory on either side because each machine was producing large batches between setups. One worker was still assigned to each machine, often simply waiting for something to go wrong, and many machines were so massive and dedicated that they could not be incorporated into flow lines. What was worse, the system went steadily backwards after being set up in 1984 (just as it had in the 1930s) because Pratt management was not prepared to continuously realign its massive machines as processing steps and part designs changed. As a consequence, physical lead times for physical production of engines from initial order and raw materials to shipped unit shrank from the traditional twenty-four months to eighteen by the end of the 1980s, but then stagnated even though the actual time needed to physically make an engine using lean methods was only a few months or even weeks.

  In 1991, Pratt was no doubt leaner than in 1983. (When the same part’s journey through Pratt plants was plotted in this latter year, it was found t
o travel only nine miles rather than eighteen.) The shop floor looked much the way it had under Carlton Ward in 1936, when some measure of flow was still in place, and the IPD system had restored some of the engineering coordination possible when Pratt & Whitney conducted its affairs in one large room. These steps were necessary and are important to note here because they provided the critical foundation for what was required next, but Pratt was not yet lean enough to survive once the crisis struck.

  The Creative Crisis of 1991

  When the world as it was understood at Pratt came to an end in 1991, there was an understandable sense of confusion and a plethora of competing ideas about what to do.

  One school of thought—the product engineers’ dream—called for pursuing a technology strategy by pushing ahead rapidly with the next generation of technology. This was the Advanced Ducted Propfan (ADP), utilizing a truly massive fan with reversible blades at the front of the engine. This concept could pull the plane forward with increased fuel economy and stop it on landing by reversing the pitch of its blades and pushing air in the opposite direction. 30

  However, the jet engine had matured to such a degree that the most optimistic estimates about the performance of this engine indicated that it could reduce fuel consumption by 6 to 8 percent at the cost of greatly increased mechanical complexity. It would push passengers along no faster and would probably require more effort for airlines to maintain. In addition, the ADP was some years away from production and depended critically on development of new lightweight composite structures capable of containing the massive fan blades in the event one or more separated in flight. 31 Although the ADP was an attractive option to pursue over the longer term (particularly if energy prices increased and the American government helped pay for its development), 32 it could hardly provide a large enough leap in performance in a short enough time to save Pratt & Whitney.

  Another school of thought—the financial planner’s dream—called for progressively downsizing the company by finding risk-bearing foreign partners for each major component in Pratt engines. These components are the large fan at the front; the compressor behind it, which squeezes air into the combustion chamber; the combustion chamber, where the dense air is mixed with the fuel and ignited; the turbine, which recovers energy from the exhaust stream as it exits the combustion area (sending the energy backwards by means of a shaft through the center of the engine to turn the compressor and fan at the front); the exhaust nozzle; the nacelle, which streamlines the exterior of the engine, contains the thrust reversers, and captures errant blades; and the accessories, such as the fuel and engine control systems.

  Under this approach, Pratt would become the “system integrator” bringing the parts together, but would need to design and make very little itself. Because many foreign firms would see participation in one part of the product as a route of entry for making whole engines, it would be easy to find foreign partners willing to cover the great bulk of development and capital costs. What was more, including foreign firms in new engine programs would help deal with the political problem of selling large orders to foreign military forces and state-owned airlines. The problem for Pratt would be the risk of being supplanted as the system integrator for subsequent engines by one or more of its risk-sharing partners, backed by foreign governments anxious to develop an aerospace industry. Indeed, this approach might easily become an involuntary exit strategy.

  A third school of thought called for rethinking the three major activities within Pratt & Whitney—the development of new products, the selling and order-taking process, and physical production—in light of lean principles, beginning with physical production. The idea was simply to start with the existing company, rapidly make it much lower-cost and much more reactive to the voice of the customer, and then consider what to do next. This was the strategy pursued by Mark Coran for Pratt’s manufacturing operations in the fall of 1991.

  From Big to Not So Big and “Flow” to Flow

  Coran’s first step was to tackle the obvious fact that Pratt had much more space, tools, and people than it would ever need again, even if it did not improve its productivity. He therefore announced in December of 1991 that 2.8 million of Pratt’s 11 million square feet of manufacturing space would be closed.

  He next announced that every product, insofar as possible, would be made in continuous flow with the aid of lean techniques, in order to reduce costs by 35 percent (in constant dollars) over the next four years and to dramatically reduce the lead time for physical production from eighteen months down to four. He imported Bob D’Amore, a lean thinker from UTC headquarters who had learned lean principles as a participant in the turnaround of Harley-Davidson in the mid-1980s, to head the new Continuous Improvement Office. D’Amore reported directly to Coran and was given the task of going through the whole Pratt production system and devising a plan for getting every production activity into some type of continuous-flow cell. This was to be Pratt’s initial kaikaku.

  Next, Coran started work on drastically reducing Pratt’s supplier base so a small number of suppliers in long-term relationships could be helped to improve their performance, and then dispatched process improvement teams to help with this task.

  It was very hard work. Pratt’s hourly workforce and middle managers had typically worked for the company all their lives and were often the children and even the grandchildren of Pratt employees. They had seen the ups and downs of the engine business for decades and many preferred to see the current situation as simply the latest cycle. It would surely pass and things could continue as before.

  In addition, the ideas Bob D’Amore was pushing challenged everything the workforce had always known. For example, D’Amore wanted to regroup machines in tight cells so that one operator could tend two, three, or more machines, while practice at Pratt for generations had been that each operator was assigned his or her own machine. In addition, he criticized Pratt’s bigger-and-more-complex-is-better tool philosophy as being directly contrary to lean thinking. What was more, Pratt couldn’t guarantee anyone—hourly worker or manager—a job after the new system was in place.

  Mark Coran remembers the situation as rather like an invasion in which a small party waded ashore and tried to take control of a vast territory simply on the strength of their new ideas. “It was hard, hard work, and by the spring of 1992, I was doubtful that Bob and I were going to make it. Every manager was talking about making a leap but nothing was actually happening.”

  Fortunately, Coran got critical help from a high place and had some good luck as well. George David had just moved up to become president of United Technologies and had completed his education in lean thinking. This was facilitated in 1991 by Art Byrne, who gave a talk to one of the periodic meetings of the presidents of all the UTC operating companies. 33

  As David remembers, “He asked us a very simple question: Why did we need so many people, so much manufacturing space, so many tools, and so much inventory to get so little done? He argued that we miserably failed to manage our assets, compared with a best-in-class lean company like Danaher or Toyota. I was bowled over with the examples of waste he pointed out in our businesses.

  “So I went over to see what he was doing at Wiremold in the fall of 1991 and it was a revelation. I’ve been an operating executive for years with a good feel for engineering but I’ve never run a plant. After I watched Art Byrne, Yoshiki Iwata, and Chihiro Nakao doing hands-on kaizens on the shop floor at Wiremold I saw the light.” So when Mark Coran told David a short time later about his frustration in pushing lean thinking through Pratt, David immediately suggested that reinforcements should be sent in the form of Iwata and Nakao.

  There was a problem, however. Shingijutsu was on the verge of signing a long-term contract to work for General Electric’s Aircraft Engine Group. When David learned of this, he raced in person to meet Iwata and Nakao at a hotel in Simsbury, Connecticut, and emerged with a multiyear agreement for them to help Pratt instead. As David remembers, “I was thrille
d. We desperately needed their knowledge and we snatched them away from GE at the last minute.”

  Lean Knowledge Is Not Enough

  Nakao’s initial foray into Pratt in May of 1992 was pure theater, like his visit to Jacobs Chuck. In the space of a week, a series of activities at Pratt’s massive Middletown, Connecticut, plant were consolidated and the amount of effort, space, and tooling needed was reduced by 75 percent. Jaws dropped and a wide range of continuing improvement activities were started, which pushed D’Amore’s original thinking much farther and faster. As Mark Coran noted later, “Our lean sensei ’s central contribution was to change permanently our sense of what was possible and in what time frame.”

  However, the new engine market was now starting to “crater” along with spare-parts orders, which had continued to plummet since 1991. Even as D’Amore struggled to unkink the existing value stream, the amount of work to be done was falling by the day, from the peak of 11 million shop hours (on a annualized basis) sustained from June 1991 through July 1992 down to an annualized rate of 8.8 million by December 1992.

  In addition, it was suddenly apparent that Pratt could not sustain the isolated operational gains it was making because there was no support structure for the new, compact cells. Bob D’Amore’s Continuous Improvement Office had neither the resources nor the authority to follow up on the myriad of loose ends at the conclusion of every improvement exercise. Nor was it able to provide day-to-day coaching to line managers on how to maintain the progress already made and how to improve on it. Even more unsettling, it was becoming apparent that many managers were actively resisting the new system. As a result, the spectacular gains achieved in the one-week improvement blitzes were quickly being lost as managers and workers went back to old ways.