The Factory as a Learning Laboratory

Just after 2:00 A.M. on 26 March 1991, the night crew at the Chaparral Steel minimill in Midlothian, Texas, cast the first production run of “near net-shape” steel beams in the United States. General Manager Lou Colatriano grinned with tired satisfaction and headed home for his first break in forty-four hours. From a sketch on a paper napkin to the first red-hot slab of metal that emerged from a patented mold to streak down the new mill line, the elapsed time was twenty-seven months. That included design, mold development, modeling, pilot runs, and construction, and it represented expertise from five companies on three continents. “One of our core competencies,” explained CEO Gordon Forward, “is the rapid realization of new technology into products. We are a learning organization.”

Every manufacturing company in the United States would like to be able to make that statement. Yet a steel mill seems an unlikely place to look for lessons on the quick commercial realization of inventions. In fact, so does any factory, as innovation is generally associated with research laboratories and development organizations. Moreover, we usually assume that the pressure to get product out the door conflicts with learning. But as speed to market becomes an increasingly important criterion of competitive success, we need to rethink our concept of what a factory is. Factories can be learning laboratories.

For decades, U.S. factories were passive service organizations to the rest of the company, churning out products designed without benefit of manufacturing input and burying product and process defects under mountains of inventory. The past decade has seen a transformation of many of these operations into low-inventory organizations dedicated to total quality and to active participation in new product development. What is the next frontier for production? In this article I suggest that it is running operations as learning laboratories.

What Is a Learning Laboratory?

A learning laboratory is an organization dedicated to knowledge creation, collection, and control. Contribution to knowledge is a key criterion for all activities, albeit not the only one. In a learning laboratory, tremendous amounts of knowledge and skill are embedded in physical equipment and processes and embodied in people. More important, however, are the nontechnical aspects, the managerial practices and underlying values that constantly renew and support the knowledge bases.

In this article I put Chaparral Steel under the microscope as an example of a highly successful learning laboratory; its leadership has put tremendous effort into creating a consistent learning system. Many of Chaparral’s practices are found piecemeal (often experimentally) in other U.S. organizations. As my references suggest, scholars studying best practices among Japanese manufacturers have made strikingly parallel observations, and Chaparral’s policies are consistent with prescriptions by organizational learning theorists. Whether you are managing a fabrication shop or claims processing in an insurance office, Chaparral’s management system offers a potentially useful model.

Taking an Organic System View

Chaparral is the tenth largest U.S. steel producer. Its high-quality standards have been rewarded by the market, and it consistently sets records for productivity, compared to both U.S. and Asian competitors (see the sidebar, “The Evidence on Chaparral Steel”). Clearly, whatever it is doing works.

A close look at the company reveals an organic learning system so tightly integrated that Forward says he can tour competitors through the plant, show them almost “everything, and we will be giving away nothing because they can’t take it home with them.” His confidence derives from the fact that the learning laboratory cannot be constructed piecemeal. It is comprehensible only as an organic whole; close scrutiny is required to appreciate its delicacy. To complicate matters, such a corporate ecosystem is in continuous flux, constantly regenerating itself. Even if a competitor identifies important elements of the system, emulation will require time. By then, Chaparral managers trust they will have moved on to the next innovation. The Chaparral system has evolved in response to a turbulent competitive environment, as Forward observes: ‘We have to go like hell all the time. If the price of what we sell goes up too high, … all of a sudden lots of folks will be jumping in. And they can get into business in eighteen months or so …. We constantly chip away the ground we stand on. We have to keep out front all the time.”1

A learning laboratory does not occur spontaneously but is designed, created, and maintained through constant managerial attention to communicating the underlying values, checking the management systems’ smallest details for consistency, and adapting any inharmonious elements. Thus managers designing a learning laboratory need to adopt holistic, systems thinking. They will have to acknowledge the practical utility and bottom-line impact of corporate values. They may have to confront chronic underestimation of the interdependence between incentive and education systems and corporate strategy. Moreover, such systems thinking must permeate the organization’s every level. Everyone in the firm must appreciate the self-reinforcing nature of knowledge-creating activities. Only by comprehending the whole system can one understand why, when a fragment of the learning laboratory is pulled out to be examined (a particular project, a specific learning activity), it comes out vinelike, trailing roots back to deeply held values and widely observed management practices. It is this intense interconnectedness that makes such systems difficult to imitate and fragile — but effective.2

Learning requires creation and control of both external and internal knowledge for both current and future operations. Therefore, four distinguishing activities are critical to a learning laboratory: (1) problem solving (in current operations); (2) internal knowledge integration (across functions and projects); (3) innovation and experimentation (to build for the future); and (4) integration of external information flows (see Figure 1).3 Each activity is the operational expression of an underlying value and is strongly supported by a compatible managerial system of procedures and incentives. Thus each activity, value, and managerial system functions as an internally consistent subsystem. Although I will describe each subsystem separately, the four are mutually aligned and interrelated; that is, values and managerial systems underlying one subsystem also support the other three. An in-depth look at the four subsystems at Chaparral suggests some of the principles that distinguish a learning laboratory.4

Before exploring these subsystems, however, we need a brief explanation of Chaparral’s near net-shape project, an innovative foray directly into the traditional territory of “Big Steel” that illustrates the company’s knowledge creation and control subsystems.

The Near Net-Shape Project

One way to push equipment performance and ensure learning is to set goals for each project considerably beyond current production capabilities.5 Chaparral managers set a very ambitious goal for the near net-shape project: to produce large (eighteen-and twenty-four-inch wide) structural steel I-beams for the same perpound cost as the simple round reinforcing bars (“rebars”), the company’s first product. “We knew what it took to make rebar,” explains General Manager Duff Hunt. “The challenge was to produce a more difficult product for the same cost. . . . We decided on the twenty-four-inch because we wanted to explore the most technically challenging product. Because there may be some idiosyncrasies about the process that we need to learn, we also chose a second size … based on the amount of North American consumption.” Reaching this cost objective (half of Big Steel’s) required drastically reducing the energy costs (roughly 25 percent of total) of rolling the steel into the required end shape. The nearer to the final (“net”) shape the molten steel could be cast, the less rolling required.

The only known processes for casting steel to its near net shape were far too capital and labor intensive for Chaparral, which views large investments in either capital or labor as a threat to future flexibility. The steel-making molds and processes developed for the near net-shape project therefore embody knowledge beyond anything available on the market and, in fact, beyond anything the leading vendors of steel-making molds thought possible. The approximately four-foot copper alloy tube through which the molten steel passes to emerge in nearly I-beam shape is set a very ambitious goal for the near net-shape project: to produce large (eighteen- and twenty-four-inch wide) structural steel I-beams for the same per-pound cost as the simple round reinforcing bars (“rebars”), the company’s first product. “We knew what it took to make rebar,” explains General Manager Duff Hunt. “The challenge was to produce a more difficult product for the same cost. … We decided on the twenty-four-inch because we wanted to explore the most technically challenging product. Because there may be some idiosyncrasies about the process that we need to learn, we also chose a second size … based on the amount of North American consumption.” Reaching this cost objective (half of Big Steel’s) required drastically reducing the energy costs (roughly 25 percent of total) of rolling the steel into the required end shape deceptively simple in appearance. However, the mold is patented, which suggests that the combined expertise of Chaparral designers and Italian and German mold fabricators endowed it with cutting-edge knowledge. The mold is contoured so the hot metal won’t bind as it shrinks going through, and the fabrication methods produce an absolutely smooth finish on the sides, to lessen chances of rending the thin skin surrounding the molten steel as it emerges. It is difficult to imagine how Chaparral could have succeeded in this knowledge-intensive project had the company not been run as a learning laboratory.

The Learning laboratory System

Some of the activities, values, and managerial practices described below characterize any high-quality organization. Others are very unusual — at least in the United States. Since they operate as a system, we have to look at the whole operation to understand why emulating just some parts may never produce true learning laboratories.

Subsystem One: Owning the Problem and Solving It

The first critical learning subsystem is the triad of (1) the independent problem solving required for continuous improvement of current processes, (2) egalitarianism as an underlying value, and (3) shared rewards as the reinforcing incentives system (see Figure 2). Empowered individuals, who command respect in the organization and who feel ownership in the system, have the self-confidence, freedom, and motivation to continuously solve problems. The principle involved here turns the old production saw of “if it ain’t broke, don’t fix it” on its head and maintains: “if it ain’t being fixed continuously, it’s broke.”

Activity: Independent Problem Solving.

Learning starts with empowered individuals who can identify and solve problems independently because they have a clear sense of operating objectives. The often-articulated vision at Chaparral is to lead the world in the low-cost, safe production of high-quality steel. This resembles the mission of many other companies that are not learning organizations. For a learning lab, the test of a vision is the extent to which it permeates the whole organization, guiding every micro and macro decision. Unless the vision can be directly translated into operational principles, that is, into guidelines for running the factory lines, it may have little effect on actual shop-floor behavior. Chaparral’s goal of leading the world requires innovation beyond the current cutting edge of production techniques. Maintaining a cost advantage requires constant improvements in productivity. The vision dictates that those improvements cannot come at the expense of quality or employee safety. Therefore the goal for every hour, the criterion for every person’s activity, is crystal clear: make ever more steel — increasingly better than anyone else. “We never stray far from the market,” says Hunt.

In a learning environment, progress has to be everyone’s business — not just that of a few specialists. At Chaparral, who owns a production problem and responsibility for its solution? An incident during the first few weeks of operating the near net-shape caster, when cooling hoses were bursting, provides some insight. “When something like that comes up, and there seems no immediate solution,” explains a senior operator, “you go see what the problem is. You don’t say, ‘That’s not my area,’ or ‘I don’t know that much about it.’ You just show up.” In this case, a group of operators, a welder, some foremen, and a buyer spontaneously gathered to discuss the problem and just as spontaneously scattered to seek solutions. “Everybody telephoned some person they thought might know how to fix the problem — vendors, experts — and within three to four hours we were getting calls back,” says the senior operator. “Service people were showing up, and we worked the problem out. If it had been just one guy, probably a foreman, and everyone walked out, … it would have taken him ten times longer to find a solution.”

Because the performance goal is very clear, supervisors do not need to micromanage the line, and the organization runs lean. For instance, two months after the first run of the near net-shape casting, the pulpit controls operator is carefully checking the timing on the line with a stopwatch. The red-hot beams pass through the rolling mill stand once, then stop, reverse, and go through again. Meanwhile, the flow of steel behind the beam being rolled is diverted. Every second of unnecessary diversion costs money because the diverted steel will have to be reheated to be rolled. Therefore the operator wants to achieve split-second timing. Asked who suggested he perform this function (which is often given to a process engineer elsewhere), he is surprised at the question: “No one.” He considers it obvious that improvement is always a part of his job: “We’re still learning, but we will get it. You just take it upon yourself to roll as much steel as you can whether you are the pulpit operator, the roller, the furnace operator. … We want to do the best we possibly can. Everybody here has the attitude to pick up the pace.”

One of the greatest advantages of this attitude, as a maintenance foreman points out, is that “ideas come from just about everybody. The operators working on the equipment have a lot of input because they see the exact problems when they happen.” Moreover, potential improvements are immediately enacted with no wait for management approval or standardization of “best practices.” If it works, it is the de facto standard. If it improves performance, everyone will imitate it. “Whoever can come up with an idea on how to fix it, from the millwrights or myself right on up to the top, … does it right then,” explains a foreman. At Chaparral, there is no formal requirement, as at some Japanese companies, for a certain number of improvement suggestions from each employee. Everyone is involved in some process improvement projects, and, as the foreman explains, “We are all out here to make it run. Probably 90 percent of the problems never even make it to the morning meetings [held among everyone on the shift to discuss problems]. They are fixed in the fidd.”6

The downside to this intrapreneurial attitude is that although managers set the goals, no one has the authority to tell another employee how to accomplish a task. Process engineers and supervisors who know a better procedure often have difficulty convincing operators on the line — much less their peers. “You can’t tell them how to do it,” Administration Vice President Dennis Beach admits somewhat ruefully; “and they don’t do it the way you would.” The engineers concur. They are not called upon enough, in their opinions. However, the benefits from general ownership of all problems are that it is not possible to “pass the buck” and no one expects a steady-state manufacturing process —ever.

Value: Egalitarianism and Respect for the Individual.

A learning environment is premised on egalitarianism, the assumption that all individuals have potential to contribute to the joint enterprise (if they are willing to develop competence). Forward has observed, “We figured that if we could tap the egos of everyone in the company, we could move mountains.”7 Outward symbols of values are important because they convey meaningful messages to all employees. At Chaparral there are no assigned parking places, no different colored hard hats or uniforms reflecting title or position, and the company dining room is a local diner. More unusual, a scant two levels separate the CEO from operators in the rolling mill; a visitor is surprised when an operator stops Forward on a walk through the plant to discuss a new product’s problems. A millwright notes: “If you have-tact, you can tell anybody from Mr. Forward on down exactly what is on your mind. There is no problem in expressing your opinion here.” Repeated requests to interview a particularly knowledgeable project manager are politely rebuffed with a shrug and an explanation from a vice president: “Sorry, he just doesn’t want to. I can’t make him.” However, respect for the individual does not mean equality of responsibility, lack of discipline, or even consensual decision making. Chaparral managers believe that a supervisor should be a leader, trained to make good decisions — including hiring and firing.

Managerial System: Performance Rewards.

Positive thinking and slogans alone cannot create genuine employee investment in innovation and in identifying and solving problems.8 Chaparral has taken unusual steps to ensure that performance and incentive systems back up management’s belief in egalitarianism. In 1986–1987, when employment leveled off, and management confronted potential stagnation, the pay structure was overhauled to reward accumulation of skills as well as performance. Even more radical (especially for a steel mill) was the switch from hourly wages to salary for everyone. There are no time clocks at Chaparral. Forward explains, “When I am ill, I get a day off. Why shouldn’t everyone else?” He is fond of saying that the management system was designed for the 97 percent who are “conscientious people who want to put in a full day’s work.” The 3 percent who abused the system were let go. Beach summarizes the philosophy: “We manage by adultery. We treat everyone like an adult.” Moreover, the pain and pleasure of work are spread around. For instance, in contrast to many factories, operators are not assigned to shifts according to seniority. Everyone rotates into night shift, a practice that fosters knowledge accumulation twenty-four hours a day and guards against a possible disproportionate accumulation of skills and knowledge in the most favored hours. The reward for seniority is thus not greater comfort, but more challenges. How do senior employees feel about that? “During the summer in Texas,” one foreman rationalizes with a grin, “you’d rather work at night.”

Steel-making is hot, dirty, demanding work; the pace of innovation and the constant pressure to produce more, better, and more safely in this organization exacerbate the brutal conditions imposed by the technology. What keeps the workforce from burning out? One factor is the incentive systems that have evolved.9

Bonus schemes are linked to company profits — for everyone. “We think janitors and secretaries are important in our success, too, and they should share in the rewards,” observes Forward. An operator comments, “The more money the company makes, the more money I make. The profit-sharing system creates built-in pride.” Further, 93 percent of the employees are stockholders and together own 3 percent of the stock. In 1988, each employee received one share for every year worked at the company; 62 percent buy additional shares every month through payroll deductions. Although the monetary implications are small, this policy is consistent with the rest of the rewards structure, and some employees find it symbolically important. A furnace controls operator comments, “I feel like this company partly belongs to me. Owning part of the company makes you care. I take better care not to waste anything because I feel like I am paying for it.”

These could be just words, of course. Scholars studying the relationship between such incentive systems and productivity have not reached consensus about the benefits of employee shareholding (partly because shareholding agreements vary widely). However, there is strong support for the notion that profit sharing generally does raise productivity.10 At Chaparral, the dual incentives of employee profit sharing and shareholding seem a natural complement to the rest of the learning laboratory system.

Subsystem Two: Garnering and Integrating Knowledge

The second subsystem revolves around knowledge accrual (see Figure 3). In the learning laboratory, knowledge is highly and visibly valued. Management invests in educating the whole person, not just the technical side, and knowledge flows freely across boundaries. The principle is this: every day, in every project, add to the knowledge resources.

Activity: Integrating Internal Knowledge.

In a learning laboratory, one would expect to see visible embodiment of knowledge creation and control in highly innovative physical systems. Chaparral boasts of such cutting-edge equipment as an automobile shredder that they believe is the fastest and most efficient in the world, a horizontal (instead of vertical) caster, and some of the most advanced digital furnace controls anywhere. Because of the constant push to improve production, Chaparral managers have to design what they need, rather than purchase the best available equipment off-the-shelf Why design in-house? “To keep the knowledge here,” a mill manager explains. Moreover, managers assume that the performance of any purchased equipment can be improved. Some improvements are novel enough to be patented. Rolling mill equipment that its vendor believed was limited to eight-inch slabs is turning out fourteen-inch, and the vendor has tried to buy back the redesign. The two electric arc furnaces, designed originally to melt annual rates of 250,000 and 500,000 tons of scrap metal, respectively, now produce over 600,000 and 1 million tons, respectively.

The physical processes following the molding step are also knowledge intensive. Only one other steel mill in the world does the hot-link rolling employed in the near net-shape casting project, whereby hot cast steel is sent directly into the rolling mills. Using this technique was a big risk because no one knew exactly what the properties of the thinner cast steel would be. The hot link meant less subsequent rolling than usual and hence less “working” of the steel to obtain the desired crystalline structure. The risk paid off when the combination near net-shape casting and hot link turned out to produce steel with exceptionally good metallurgical characteristics. Chaparral processes seem clear outgrowths of skill, yet how does such expertise and knowledge accumulate?

Just as continuous processing has great advantages for manufacturing over most batch processing, so the unimpeded flow of information aids learning more than fragmented, batch-processed information.11 Most organizations are physically structured to emphasize vertical (hierarchical) and horizontal (functional) boundaries. In contrast, Chaparral management emphasizes homogenizing the level of knowledge throughout; few pockets of information are isolated by position, function, or working shift.12

Information flow at Chaparral is obviously aided by its size, deliberately held to under a thousand employees. An individual garnering knowledge on a trip, at a conference, or from an experiment can readily transmit it, as all employees are located in the same place and know each other. The company was also designed to facilitate knowledge flows by encouraging as many accidental meetings as possible. The plant layout accommodates the hands-on style of management favored at Chaparral; even Forward’s office is just steps away from the furnaces and mills. The locker room is located here also, so that at least once a day employees cycle through the one-story headquarters building.13 Consequently, meetings are as likely to be held in the halls as in the conference rooms. Since many decisions go unrecorded and memos are anathema, it is important that people see each other frequently.

Hierarchical boundaries are minimal. This is a do-it-yourself company with no acknowledged staff positions and only a few positions that seem stafflike, such as personnel. There are fifty graduate engineers and technicians, all with line duties. In fact, everyone has line responsibilities, most of them tied directly to steel production, and decision making is pushed down to the lowest possible supervisory level, “where the knowledge is.” Lead operators are selected for their knowledge-transmitting as well as knowledge-creating skills, because much knowledge flows horizontally among peers. Work is structured with the objective of disseminating knowledge. For instance, in commissioning the new mill that receives the near net-shape product (that is, ramping it up to problem-free production), only two teams of operators are being trained. Each team works a twelve-hour shift (with paid overtime). After the initial eight weeks of this grueling schedule, these operators will be dispersed among the rest of the crews to diffuse the knowledge they have created and assimilated about the new process’s idiosyncrasies.

Chaparral has also proven that traditional horizontal boundaries can be redrawn and expectations altered. The quality control department is responsible for reacting to quality problems identified by operators on the line — not by separate inspectors. Production workers do 40 percent of maintenance tasks. A maintenance foreman notes that “at Chaparral, we get involved with the whole process. We are not just tied to one area.”

Although the company has a marketing department, everyone is considered a salesperson. Every employee from CEO to receptionist has a business card to use with customers. Security guards do data entry while on night duty and are trained paramedics as well. Such multifunctional experience is encouraged not only to make the organization more flexible but because management believes it discourages territorial possessiveness over information.

No research and development (R&D) department exists separate from production. Forward maintains that “everybody is in research and development. The plant is our laboratory.” This statement has implications beyond the obvious that there is no research function. Some of the problem solving and experimentation that are done in the midst of production involve research. However, interfacing with the possessors of the latest scientific knowledge is not restricted to an elite group of specialists. Knowledge accumulation in a learning lab cannot be the responsibility of a few special people. Forward describes the large, separate research centers in some companies as “lovely, really nice. But the first time I went into one of them I thought I was entering Forest Lawn [Cemetery]. After you spend some time there, you realize you are in Forest Lawn. Not because there are no good ideas there, but because the good ideas are dying there all the time.”14

At Chaparral, even operators participate in R&D activities. When the world’s leading supplier was constructing the patented mold for the near net-shape project, the three-person team that shuttled back and forth to the site in Germany, learning about the fabrication process and serving as a source of information about the intended production process, included an operator. As described later in Subsystem Four, Chaparral employees at all levels are constantly tapping into the latest, most current knowledge banks around the world.

Value: Shared Knowledge.

Since performance drives everything in this company, and individual incentives are tied to performance, employees seem engaged in a marathon relay race, where winning as a company team takes precedence over individual ownership of ideas, and knowledge is liberally shared. There are acknowledged experts, such as the director of operations who is an “equipment whiz,” and the mold expert who was largely responsible for the near-net shape design, and the organization does not lack for large egos. However, Vice President of Operations Dave Fournie says that “you don’t have to have credit for particular ideas to be thought good at your job. Lots of innovations take more than one good idea. They go through a gestation period, and lots of people figure out how to make sense of it. The point is to focus on the good of the whole. That’s why we don’t have suggestion boxes, where you hide ideas so someone else won’t steal them.”

Chaparral employees are often unable to identify the source of production innovation. Production Manager Paul Wilson explains, “It is hard to say who fathers an idea. It doesn’t make any difference. Everyone shares in the pride of doing, and if the experiment fails, everyone shares in the failure. In other places, a few people do a lot of innovating. Here a lot of people do little bits that add up.” A millwright makes similar observations: “No one is looked down upon. If I am supposed to know more than somebody else but that other person catches [an oversight], well that’s fine — he just bailed me out. I am more than likely going to help him out a different time.”

Managerial System: Apprenticeships and Education.

An organization that values knowledge must provide mechanisms for continuous learning. Chaparral management has sent some employees to school to obtain advanced degrees, but it also invests heavily in an unusual formal apprenticeship program for everyone in the plant, which it developed with the Bureau of Apprenticeship and Training in the U.S. Department of Labor. (Most apprenticeships are run by unions.) As Forward notes, “Expertise must be in the hands of the people that make the product.”

The roughly three-and-a-half-year program allows apprentices to progress to the level of senior operator/ craftsman by successfully completing 7,280 hours of on-the-job training and designated formal schooling. The foremen of individual crews schedule the on-the-job training and evaluate the candidate’s systematic progression through various tasks in the factory. For example, 2,200 hours in steel-pouring operations is one qualification in the ladle metallurgy apprenticeship. In addition, all apprentices complete study programs in safety, operating processes across the entire company, mathematics as related to operations, metallurgy, and basic mechanical maintenance “in an effort to give all operators a solid basic understanding of the equipment used and the operating processes.” Scheduling this study, which requires four hours of unpaid work a week, is left up to the individuals, who have five options: home study with manuals and videos; in-plant study in instructor-monitored study rooms; formal classes; personalized tutoring by instructors; and study in work areas as operations permit. (Operators may also get credit for prior experience by proving knowledge on the job.) Courses include basic engineering knowledge such as conduction in liquids and gases, as well as very specific skills such as ladder logic programming for understanding programmable controllers and lubricant storage and handling. Perhaps more surprising are such sessions as “working with other people” under “troubleshooting skills.”

The most unusual aspect is the instructors. Selected foremen rotate in from the factory floor to teach. “It creates a lot of credibility for the education program on the factory floor,” explains one instructor. “What’s more,” he adds wryly, “I have to live with what I teach. So I’d better do a good job.”

Chaparral also invests in outside courses of all types, tending to train its own people rather than hiring those with a particular skill. A maintenance foreman attended a special engineering course because “no one knew anything about vibration analysis so I was asked to go study it.” External education includes nontechnical courses as well. In few factories would a furnace controls operator attend a Dale Carnegie-type course to help him enhance the interpersonal skills critical to smooth teamwork. Line managers are expected to be technical experts, constantly up-to-date with cutting-edge technology — whether that be some aspect of metallurgy or human resources. Therefore every manager attends conferences and cultivates a professional network of information sources. The manager of the scrap shredding operations, for instance, has just attended a conference on recycling — in the future, Chaparral’s expertise in recycling may be an important corporate capability.15

Subsystem Three: Challenging the Status Quo

The third subsystem in a learning laboratory involves constantly pushing knowledge frontiers (see Figure 4). The company must select employees for their desire to challenge their own and others’ thinking. They must see risk as positive, because it comes with the experimentation critical to innovation. The company must select suppliers for their superior capabilities — and for their willingness to be pushed beyond the bounds of their current knowledge. The principle involved is this: always reach beyond your grasp.

Activity: Continuous Experimentation.

Learning requires constant pushing beyond the known, and Chaparral employees are skilled experimenters. A visitor was surprised to find that extensive overhead slides explaining the formal Taguchi experimental designs that were guiding the development of the horizontal caster had been prepared for the board of directors. This extremely technical presentation was to help directors understand the methodical knowledge-creation process, enable them to identify critical decision points, and thereby better equip them to evaluate the risk. Another example of knowledge-creating experimentation more often found in research laboratories than factories is the one-sixth scale model of the near net-shape caster, which uses water to approximate the flow of steel. Standing alongside the rolling mill, close to a water source, this model allowed testing of several types of baffles and lengths of mold. Resulting observations guided much of the eventual mold design decisions.

Innumerable large and small experiments are more of the “cut-and-try” variety. “We aren’t always as systematic about our learning as we could be,” Forward admits. Many creative simulations are conducted right on the production line. The advantage is that the more closely the experimental environment approximates the final production environment, the more immediately relevant is the information generated. The disadvantage is the obvious potential to disrupt production — which is why most factories prefer to isolate experimentation. In one of many projects leading to the near net-shape casting, a prototype of metal splashboards was first constructed out of plywood. By continuously soaking the wood in water, the crews were able to keep it from being consumed by molten steel just long enough to prove the concept. “We were the local hardware store’s favorite plywood customers for a while,” one employee recalled. Similarly, during the design of the near net-shape caster, several prototype molds of almost pure copper were used to determine if casting a dog-bone-shaped slab was possible. The soft molds held together just long enough to show this new casting process’s feasibility.16

The operating rule is this: if you have an idea, try it. Line managers authorize tens of thousands of dollars for experiments without higher authority. Hunt explains, “We use products to do research. We can close the feedback loop between researchers and users by using new methods and new materials within our own facility.” Wilson agrees: “In other companies, the word is — don’t rock the boat. Here we rock the hell out of the boat. We don’t know the factory’s limits. We want it to change, to evolve.”

Not every suggestion is instantly accepted. When a maintenance operator at a conference spied new digital furnace pulpit controls, he then had to convince his supervisor to invest. For almost two years, the supervisor remained uninterested in visiting the Mexican vendor and a site using the controls. Yet the operator persisted. Finally, his supervisor understood the potential of these state-of-the-art controls and agreed to their installation and customization to Chaparral operators’ specifications.

During the development of the near net-shape beams, Chaparral frequently disconcerted development partners by its innovation speed. One innovation followed so hard on the heels of the previous one that they almost overlapped. When the German mold developers finished a prototype on schedule and called to ask how they should ship it, they were astonished to be told, “Don’t ship it at all; cut it up and make it look like this.” The designers had already learned enough from the prototyping process to improve the design. A new mold would be required. What surprised partners was that Chaparral did not appear to begrudge the “wasted” $40,000 per mold, since the knowledge engendered by each prototype enabled the next step in innovation.

Value: Positive Risk.

In a research laboratory, risk is accepted as the norm, since the cutting edge is always fraught with uncertainty. In contrast, risk is usually anathema in a production environment. Managers of a learning factory must tolerate, even welcome, a certain amount of risk as a concomitant of knowledge acquisition. Chaparral managers avoid riskless projects because a “sure thing” holds no promise of competitive advantage — no opportunity to outlearn competitors. Says Forward, “We look at risk differently from other people. We always ask what is the risk of doing nothing. We don’t bet the company, but if we’re not taking some calculated risks, if we stop growing, we may die.”

This positive attitude toward risk permeates the company. If everyone experiments, learns, and innovates, then neither success nor failure can be heavily personalized. If individuals are singled out for praise, then they have an incentive to protect ideas as intellectual property rather than seek embellishment from friendly critics and codevelopers. If individuals are singled out for blame, then the risk of failure may overwhelm the impulse to innovate. What happens when you try something and it doesn’t work? What is the penalty? “Everybody makes mistakes,” a Chaparral foreman responds. “You don’t have to cover up a mistake here. You just fix it and keep on going.” The philosophy appears to apply even when the failed experiment is very expensive. In 1986, when Fournie was medium section mill superintendent, he championed the installation of a $1.5 million arc saw for cutting finished beams. Not only did the magnetic fields attract any small unattached pieces of metal for yards around, including pens and watches, but the engineers were never able to refine the equipment to the point of effective operation. Since promoted to vice president of operations, Fournie is somewhat amused to find that visitors “can’t believe you can make a mistake like that and not get crucified.” He tries to take the same attitude toward those who work for him: “You don’t start them out on $1.5 million projects, but you have to give them freedom to make mistakes. The reward for having ideas is getting to carry them out. You give them bigger and bigger chunks and evaluate.” Operators on the line have a consistent view of the atmosphere for innovation: “You have an idea — good, bad, indifferent — just spit it out and we’ll talk about it. Someone may laugh [at it] but you’ll laugh back next week.”

A potential hazard of this positive attitude toward risk is that no one wants to admit a mission is impossible. “Once we say we will do something,” Fournie explains, “we hang on and try like nobody’s ever tried before. Tenacity makes lots of projects work here that don’t work other places — but when it just can’t be done, it’s hard to call the project off.” One of his key criteria for determining if people are ready for promotion is whether they know when to ask for help, when to admit that they are in over their heads: “It’s a tough call … the hardest decision to learn to make.” He speaks from experience. He not only set up the arc saw project; he killed it.

Managerial System: Hiring Practices and Career Paths.

The most important managerial system in a learning laboratory is selecting and retaining the right employees. Because employees must be innovators, constantly challenging the status quo, they are selected as much for their potential, their attitude toward learning, and their enthusiasm as for a specific background. Although top managers and a few specialized “gurus” at Chaparral have extensive steel experience, when Chaparral was first set up, management decided not to look for workers with industry experience. Beach explains: “We were looking for bright, enthusiastic, articulate people, and we preferred people who had not been exposed to other companies’ bad habits.” The company therefore hired (and continues to hire) from the immediate geographic area, seeking ranchers and farmers with mechanical ability but no steel experience. Chaparral looks for “a twinkle in the eye, a zest for life,” for “basically conscientious people who can put in a strong day’s work and enjoy what they’re doing.” This love of work is critical in an organization that deliberately runs somewhat understaffed to avoid laying people off during market downturns. A learning laboratory is not a good match for someone who works solely for a paycheck or who equates promotion with increasingly easy work.

Chaparral’s original applicants went through six weeks of intensive training with daily evaluations and faced stiff competition. Top performers were given their choice of jobs and an immediate 20 percent pay raise. Highly selective hiring procedures continue to reflect concern that new employees fit into the Chaparral culture, that at least one supervisor be personally committed to their training and progress, and that the team have a stake in their success. Although personnel does some preliminary screening, current applicants undergo one or more days of demanding interviews with at least five employees, including two foremen, before they join Chaparral. Only one out of ten applicants selected for interviews can expect to be hired, and the final decision belongs to the foreman with direct responsibility. These very cautious, resource-intensive selection practices may account in part for the extremely dedicated workforce, which boasts an absentee rate about one-fourth that represented by the National Association of Manufacturers.

Since people are the key resource in a knowledge-intensive organization, keeping them is critical.17 Beach explains: “From the very beginning, we designed this organization with Maslow’s hierarchy of needs in mind [i.e., that once people’s basic needs for food, shelter, and belonging are satisfied, they will aspire to a fourth level of need, self-esteem, and finally, the fifth, self-actualization.] I know it’s not stylish, but we really believe in that hierarchy, so we constantly look at what will help people become self-actualized, at their ego needs. … People like a challenge and a well-defined goal out there.”18

A strong aid in motivating continuous innovation is a clear path for advancement, not just in salary but in position. Asked what he looks for in a job, a millwright responds: “As long as I am moving forward, I can think for myself: and I feel that I am contributing, I will be happy.” Chaparral managers believe that skilled, innovative people will leave an organization if they see no possibility for personal growth. During the 1986–1987 reorganization to address possible stagnation, some maintenance and shipping operations were reorganized into autonomous, self-directed teams with rotating leadership. Management regards this change as an organizational experiment that may provide a model for other parts of the company.

The company invests in cross training at a number of levels. Management has learned the necessity of training in advance of the need because of problems experienced early in the company’s history, when operators who were automatically moved up into lead and then supervisory positions as vacancies opened proved underprepared. In response, Chaparral management established such idiosyncratic practices as “viceing.” In other companies, when a foreman or supervisor is absent, usually the foreman from a prior shift stays on in that position. When a similar absence occurs at Chaparral, the prior foreman works the extra hours, at his usual pay level, but in a subordinate role. The most senior operator or craftsman is then temporarily promoted to “vice foreman” to cover the supervisory position. Thus the company retains operating experience in the form of the prior shift foreman and simultaneously trains the senior operator for future work as a foreman. A supervisor explains, “You get a little pay increase and a whole lot more responsibility I got a chance to see whether I really wanted to be a foreman or not, and it gives you a little respect for the boss’s job.”

Managers are similarly prepared for the future. Recently the production managers for the three mills were given the title of general manager and asked to learn each other’s jobs and to cover for each other. This crosstraining is intended to prepare them for general management of an entire operation, when Chaparral starts up another site.19

Subsystem Four: Creating a Virtual Research Organization through Networking

A learning laboratory obviously needs access to the latest knowledge, embodied in the best minds and best equipment available. However, not all companies can afford an internal research organization. Moreover, no company can cover all the technological advances, worldwide, that may affect its future. Therefore the principle behind the fourth and final subsystem is this: create a virtual research organization through extensive networking and alliances — for learning and for economic reasons (see Figure 5).

Activity: Integrating External Knowledge.

At Chaparral, employees constantly scan the world for technical expertise that others have already invested in. Managers never hesitate to invent when necessary but only after assuring themselves through extensive searches that no available system will suit their needs. While building the horizontal caster, they made repeated trips to the few other world sites that had somewhat similar equipment. Chaparral also constantly benchmarks its capabilities, not just against immediate competitors but also against best-of-class companies, even those from totally different industries. Three on-site laboratories support production through chemical and physical product analysis, but the company has created a virtual research organization through extensive networking and alliances. Information obtained externally is rapidly incorporated through development projects, flowing through the created network almost as readily as it does inside the walls of the learning laboratory, because in both cases people working directly in production transmit the knowledge.

The network was heavily used in the near net-shape project. At its outset, a team of managers and foremen visited Japan and determined that one firm had the most advanced “profile” casting process in the world. Yet it was inflexible, would cost “more than the entire company,” and was more labor intensive than desired. Chaparral therefore enlisted the help of German and Italian suppliers. At first extremely skeptical of this small company from the middle of Texas and hence unwilling to try to make the radical mold design, the German supplier began to believe that seemingly impossible goals could be achieved only when it saw how far Chaparral had progressed with the help of the Italians. Chaparral employees visited the German vendor every few weeks, and ultimately Chaparral’s expertise combined with the German’s cutting-edge knowledge to produce a patented mold that neither company could have made alone.

Concerned that the relatively thin cast steel emerging from this mold might have undesirable metallurgical qualities, Chaparral reached again into its network for testing. Since it had not yet built its own rolling mill, and none existed anywhere that could receive the novel shape, managers identified a production laboratory in Mexico with very flexible equipment. Chaparral’s superintendent of steel-melt technology went to Mexico to direct the simulation of the future mill design, using Mexican equipment and workers. To their delight, Chaparral managers discovered that the samples demonstrated superior characteristics.

The development and testing laboratories in the German and Mexican firms served as virtual extensions of the corporation, for they possessed special equipment and skills that complemented Chaparral’s design capabilities. However, knowing such sources of expertise exist would be useless if the factory were not able to tap them, to jointly create more knowledge, and then to absorb that knowledge into the production system.

Chaparral’s collegial knowledge network seems more characteristic of a research laboratory than of a factory. Possibly because of Forward’s early career as a research metallurgist, the company aggressively pursues the latest knowledge. Chaparral sought the coveted Japanese certification for steel, not because managers thought they would ever sell much steel in Japan, but because they believed the Japanese would go through the company’s process carefully — and Chaparral would learn much from the exercise. Forward takes a distinctly nontraditional view of environmental scanning for a factory: “By the time you hear about a technology in a paper at a conference, it is too late.” This philosophy explains why Chaparral invests in unorthodox knowledge-gathering mechanisms, for instance, by cosponsoring a research conference with the Colorado School of Mining about a new alloy under investigation. Forward himself treks back regularly to his alma mater, MIT, to consult with university experts.20

Very occasionally, this method of R&D has failed the company. The most notable example is the previously mentioned electric arc saw. No one in the plant or in its network of experts, including the vendor, was able to solve the technical difficulties. The mill manager left the saw in place for almost a year to allow time for everyone to attempt solutions and to accept failure, while he sought an alternative technology: It is not clear that an in-house R&D facility would have helped in this case because the required invention lay in the unfamiliar realm of electromagnetic fields. Managers at Chaparral believe that through their virtual R&D organization, they can actually tap a larger variety of knowledge bases than they would be able to support internally, given their size. Still, they will have to continue investing in internal expertise in order to integrate the externally obtained knowledge.

Value: Openness to Knowledge from Outside.

Knowledge garnered through such networks can flourish only in an environment that rejects the “not invented here” mentality. At Chaparral, “not reinvented here” is the operative slogan. There is no value in recreating something — only in building on the best existing knowledge. People in a learning laboratory value the capability to absorb and use knowledge as much as to create it. They understand that all invention is a process of synthesis. As its practices suggest, a key Chaparral value is global outreach — openness to innovation, whatever its origin. Knowledge is valued not so much for the pedigree of its source but for its usefulness.

Managerial System: Resources for Alliances and Networks.

To support information gathering and reinforce global outreach, the company invests heavily in employee travel (and regards the expenses as just that — investments), often sending a team, including foremen and technical staff as well as vice presidents and operators, to investigate a new technology or to benchmark against competitors. Newly acquired knowledge need not filter down through the ranks, because the people who absorbed it are the ones who will apply it. In 1990, seventy-eight people from production, several of them operators, visited a customer site at least once. They also visit other minimills. Asked why he visited a sister plant, an operator states: “To see if I could pick up any new ideas to use here.” And did he? “Yes, several.” And he points out some small operational changes. Forward elaborates: “We send the people who can best tell us what’s going on — whoever they are.” Chaparral managers also invest in long-term relationships with suppliers to extend the network. In 1985, while seeking Japanese suppliers, they identified a high-quality company that was not doing much development. As a general manager explains, although it was clear that initially the supplier “would learn more than we” from the alliance, Chaparral spent the money to send employees over to “hand hold” so as to develop a capability that would be useful in the future.

This constant dispersal of mixed employee teams to customers, competitors, and suppliers throughout the world serves a dual purpose. The visits are regarded as learning sabbaticals that keep life “exciting” for employees.

They are a source of information for the company as a whole. “We want them to … come back with new ideas about how to make improvements or new ways to understand the problem.”21

Conclusion

The factory in fact any backroom operation, is not usually regarded as an arena for experimentation and learning. Chaparral Steel challenges this concept of operations. Factories can function as learning laboratories (see Table 1). The most important characteristic of such organizations is that they are totally integrated systems. They are difficult to imitate because every employee, from CEO to line operator, is technically capable and interested in learning. Moreover, the whole organization is designed around the creation and control of knowledge. The four subsystems described above are not only internally linked but tremendously dependent upon each other. Continuous education depends upon the careful selection of willing learners. Sending workers throughout the world to garner ideas is cost effective only if they are empowered to apply what they’ve learned. The organization is unlikely to be open to outside knowledge if it does not place a strong value on sharing knowledge or does not give rewards for bettering the whole company’s performance. Thus continuous learning depends upon the sense of ownership derived from the incentive systems, upon the pride of accomplishment derived from special educational systems, upon values embedded in policies and managerial practices, as well as upon specific technical skills. The line operator appears to take the same perspective on the conduct of daily activities as the CEO. Chaparral is tremendously consistent.

Paradoxically, the system’s interdependence is also a potential weakness, as competencies often are.22 A learning laboratory may have trouble recreating itself Any organization is likely to be somewhat limited by its “congenital knowledge” and by the stamp placed upon it by its founders. 23 A significant challenge for Chaparral is how to grow. Forward has noted that “to stand still is to fall behind,” and the company’s credo to provide growth for its skilled people requires some forward momentum. Therefore, it must either grow larger where it is or clone itself in a new location. How will it transplant the deep worker knowledge, the motivation, the commitment, and the informal systems of knowledge sharing?

For other companies interested in creating factories as learning laboratories, the questions are: Can it be done in a plant within a large corporation, where many of the managerial systems have already been set corporatewide and therefore are not at the plant manager’s discretion? Can an existing plant be transformed when plant managers may not have the luxury of selecting people as freely as Chaparral did? Can a company less geographically isolated hope to reap returns on investing in its employees’ intellectual advancement, or will they be lured away by other companies?

Similar questions were raised a decade ago when U.S. manufacturers first began to understand how Japanese companies were competing on the basis of quality. Initially overwhelmed by the difference in activities, values, and managerial systems implied by the total quality approach, many U.S. managers were pessimistic about their ability to change their operations to the extent needed and reluctant to invest in employee education. Yet today, many U.S.-based factories achieve quality levels never even aspired to in 1980.24

For many of these improvements, the Japanese were our teachers, having learned from Deming and Juran.25 According to some researchers, Japanese managers may also be ahead of their U.S. counterparts in creating learning laboratories, since “Japanese companies are usually adept at organizational learning.”26 As the references at the end of this article suggest, many of Chaparral’s managerial practices and values also characterize best practices in Japan, such as investing extensively in formal and informal education; searching worldwide for the best technology and methods and absorbing that knowledge into home operations; and valuing employee empowerment, problem solving, and risk taking.27 Apparently, such learning systems are not uniquely Japanese.

Of course, creating a learning laboratory in a “green-field” site is easier than in an existing plant. As not all U.S. factories can start up from scratch, creating a learning laboratory implies big changes. But do we really have any choice? No financial formulas, corporate reshuffling of departments, or exhortations by corporate management to integrate across functions will foster the creativity and productivity needed for international competition.

Experts on change management suggest that three critical elements are required for altering current practices: (1) dissatisfaction with the status quo; (2) a clear model of what the changed organization will look like; and (3) a process for reaching that model, that vision of the future.28 Examples such as Chaparral can aid all three. One way of stimulating dissatisfaction with current practices and hence motivation to change is to observe other companies where an alternative management style appears to be yielding superior results. By benchmarking against such companies, managers can derive principles to incorporate into their own particular visions. Chaparral offers a model of a factory as a learning laboratory, and if the specifics are not transferable, the principles underlying the Chaparral vision are.

The precise process for implementing these principles will differ markedly from company to company. It is possible to interrupt a factory’s current systems by introducing new equipment, new learning skills and activities, new knowledge-creating management systems, or new values. But interrupting a current system is only the first step. As the Chaparral example demonstrates, learning skills, management procedures, and values are interrelated. Values unsupported by management systems are vapid; management systems that run counter to values are likely to be sabotaged; learning activities unsupported by values and management practices will be short-lived. If a learning capability is to be developed, the whole system must eventually be addressed.