Making Design Rules: A Multidomain Perspective - INFORMS ...

Organization Science

informs

Vol. 17, No. 2, March–April 2006, pp. 179–189 issn 1047-7039 eissn 1526-5455 06 1702 0179

®

doi 10.1287/orsc.1060.0180 © 2006 INFORMS

Making Design Rules: A Multidomain Perspective Stefano Brusoni

CESPRI and CRORA, Bocconi University, Via Sarfatti, 25, 20136 Milan, Italy, and Silvio Tronchetti Provera Foundation, Via G. Negri, 10, 20123, Milan, Italy, [email protected]

Andrea Prencipe

Faculty of Economics, University G. d’Annunzio, Viale Pindaro 42, 65127 Pescara, Italy, and SPRU, University of Sussex, Brighton, BN1 9QE United Kingdom, [email protected]

D

esign rules allocate functions to modules, identify operating principles, and set interfaces among modules that determine how organizations evolve. A case study of radical innovation in tire manufacturing illustrates the transition from old to new design rules through the joint adaptation of the manufacturing organization and the product to reflect changes in the underlying engineering knowledge. The case shows how knowledge evolution mediates organizational and technological change and makes any organization design openended and evolving. Key words: modularity; design rules; dynamic multiple networks; tires; tire manufacturing

1.

Introduction

were observed in the aircraft engine (Prencipe 1997), hard disk drive (Chesbrough and Kusunoki 2001), and automotive industries (Takeishi 2002). These studies illustrated that firms consist of different domains, e.g., organizational structures, technological architectures, etc., that may obey different design rules. The evolution of firms’ knowledge bases also plays a fundamental role in mediating the relationship between product and organization design (Brusoni et al. 2001). Through an in-depth analysis of radical innovation in tire manufacturing, this paper illustrates how new knowledge enables technological and organizational evolution. The paper is organized as follows. The next section describes the research setting and methods. Section 3 analyzes why Pirelli developed the robotized manufacturing process. Section 4 analyzes how the new process was developed. Section 5 discusses the findings, and §6 concludes.

How do organizations adapt in response to technological innovation and environmental transformations? To answer this question, organizations are conceptualized as entities made up of many interconnected elements (Lawrence and Lorsch 1967, Perrow 1967, Thompson 1967, Galbraith 1973, Woodward 1965, Miller 1987, Whittington et al. 1999, Grandori 2001). These elements can be combined so that internal and external fits correlate with superior performance (e.g., Drazin and Van de Ven 1985). Recent research has analyzed the processes through which organizations evolve toward fit (e.g., Siggelkow 2002) and suggests that some elements are core to an organization and others are not. Evolution toward fit is a process of strengthening and restructuring connections among core elements. This paper considers how new core elements come into being using the concept of design rules (Baldwin and Clark 2000). Design rules are principles that define how an artifact works, what it does, and how it is manufactured. They allocate functions to components, identify operating principles central to each component, and set interfaces among components. Research on design rules revolves around the concept of modularity. Modular design rules are based on the twinned principles of interface standardization and components decoupling. These principles enable designers to make modifications to one part of a system without communicating with the designers of other parts. Thus, modular products can lead to modular organizations, as product design rules define both the technological and organizational architecture of firms (e.g., Sanchez and Mahoney 1996, Schilling 2000, Sturgeon 2002, Langlois 2003). Empirical studies questioned such findings: Nonmodular organizations that produce modular products

2.

Research Setting

The tire industry is often depicted as an archetypical example of a mature business (Sull et al. 1997, French 1991). Since the early 1990s, however, it has witnessed an explosion of R&D initiatives. In the late 1990s, the trade literature described ongoing technology developments that were leading to a fully robotized and modular tire production process. Pirelli Tires, a leading player, developed a process called the Modular Integrated Robotized System (MIRS). The trade press reported that MIRS would usher in an era of customized tires. It also said that robotized, modular production had the potential to change century-old practices that emphasized the mass production of tires, the standardization 179

180

of tire products, and economies of scale. Business models based on robotized processes, in contrast, emphasize economies of scope, i.e., the production of small batches of differentiated tires; flexibility, i.e., rapid responses to changes in market trends; and customization, i.e., abilities to fulfill specific engineering requirements for different car models. In other words, the diffusion of robotized production would lead to fundamental changes throughout any organization that adopted it. Design rules are usually hidden, but we thought we could make them visible by focusing on their initial development. We approached Pirelli to get data about MIRS. Pirelli appointed a company tutor, Renato Caretta, to introduce us to the manufacturing process. Our tutor was a senior engineer, a member of the managing board, a former senior R&D executive officer, and the person who led and conceived of MIRS. As our objective was to investigate empirically a phenomenon that has received little attention—i.e., the emergence of new design rules—we adopted a strategy of theoretical sampling to generate data. First, we identified three individuals who played key roles in Pirelli’s project. They defined major milestones, e.g., the opening of the first pilot plant; and major organizational transformations, e.g., when and why the team hired new people. We then expanded our interview sample to include additional people who influenced the product and software design. We stopped sampling when new interviews did not add new theoretical categories (Glaser and Strauss 1967), i.e., when we concluded that the events we were discovering could be interpreted within the knowledge, technology, and organization domains. Between April and June 2004, we carried out a first round of interviews, analyzed the data collected, and clarified specific points via phone calls and e-mails. We conducted a second round of interviews between July and September and clarified details through follow-up phone calls and e-mails. We interviewed 27 different individuals, some several times, for approximately 75 interview-hours. Initially, interviews were unstructured and unscheduled, e.g., some occurred while the interviewer and interviewee walked through the plant, and we asked generative questions (Strauss 1987). The intent was to understand the organization and its evolution, and to identify key events and key participants. We also sought to identify other knowledgeable people who could confirm and clarify matters. We used these interviews to distinguish differences between traditional and innovative tire manufacturing in terms of their design rules, e.g., modularity versus nonmodularity; articulation of design heuristics versus tacit knowledge; integrated know-how versus specialized know-how. We singled out core elements that would serve as a basis for constituting design rules, e.g., rigid drum versus flat support deposition; digital control versus visual control; asymmetric deposition versus symmetric deposition.

Brusoni and Prencipe: Making Design Rules: A Multidomain Perspective Organization Science 17(2), pp. 179–189, © 2006 INFORMS

The second round of interviews was problem oriented (Flick 2002, p. 125). We probed whether we had identified key design rules that set apart the two processes and their core elements. For example, we discarded a few elements that we initially thought were critical, e.g., issues connected to production costs. In addition, we validated the distinction between domains by gathering specific stories about difficulties and achievements related to the development of new skills, e.g., the nylon example reported in §4. Relying on these stories, we probed our initial expectations about the relevance of cross-domain connections, how they were activated, and by whom, e.g., the role of software engineers analyzed in §§4 and 5. One researcher carried out all interviews. After each interview, he wrote a short description of the “case,” identified the interviewee and the interview context, and summarized the main points (Stake 1995, Creswell 2003). Interviewees stressed different aspects of the project: e.g., knowledge aspects such as the new skills required to develop the new process; product issues, such as how concerns surrounding product quality using robots rather than labor-intensive processes were managed; and organization issues, such as project team changes during the development process. Our data set includes information from patent, archival, product, and quality assurance databases. We studied archival data and the trade literature to understand how Pirelli articulated its strategies in the mid-1990s prior to the launch of the MIRS development project and afterwards. Internal archival documents told us about the size of the project, the people involved, the resources required, the internal visibility, and top management’s commitment. We studied industry-level patent data to identify the innovative strategies of other firms. We relied on European Patent Office data and the trade press to verify we had interviewed all important project players. Analysis of the chronology of patent applications confirmed claims by Pirelli personnel about preexisting MIRS-related manufacturing process capabilities, a point that became key later in our theoretical interpretation. The company’s internal product data confirmed the increasing segmentation of the tire market. The analysis of the Failure Mode and Effect Analysis (FMEA) database, required to obtain ISO-9000 certification, validated claims about the core technological characteristics of the robotized process. The FMEA database lists causes and solutions for all production process problems, and the frequency of their occurrence. We accessed these data for two generations of the innovative process, and for the most efficient traditional process, in order to assess changes in the conceptual complexity of the processes. The problems and issues that we initially identified were not always the same as those mentioned in the FMEA database, the record of actual engineering and strategic challenges that Pirelli encountered. Also, some


aggregate indicators in the FMEA data did not match up with what our interviewees told us during our first round of interviews. Such discrepancies provoked broad interest, and Pirelli management asked us to be available to its MIRS employees to answer questions and discuss what we had found. For nearly 10 weeks, one of us spent two and a half days per week at Pirelli’s central office collecting and analyzing data on the one hand, but also dealing with Pirelli’s questions concerning how the MIRS development process had actually evolved. This episode exemplifies how the interviewer’s role adapted as our understanding was being “grounded” into the context of study. Initially, he was more like a student, recording the tutor’s description of the industry and MIRS. During problem-oriented interviews, the researcher was like a curious visitor (Agar 1980), i.e., someone who had some knowledge but was still an outsider. As we focused on the analysis of internal documents, the researcher’s role became like that of an initiate (Agar 1980). Finally, the analysis of the FMEA database made the researcher a peripheral member of the group (Adler and Adler 1987), who offered his views and exchanged ideas with others concerning the meaning and possible implications of events in the MIRS development process.

3.

Why Did Pirelli Invest in MIRS?

Although the tire industry has many distinct market niches, it is concentrated: Trade sources in 2000 reported that the top 10 tire manufacturers accounted for 83% of global sales. Over the past three decades, the industry has had shrinking margins due to declining purchases deriving from increased tire longevity, buyer pressure to keep prices down, and increasing raw materials costs. Consequently, leading companies have reduced labor costs, increased throughput, improved process technologies, and rationalized material use. The rapid emergence of new niches—e.g., ultra high-performance tires for sport utility vehicles (SUVs)—and new material availability have also led to improvements in tires. For example, patent statistics show that compared to other technologies, microelectronics applied to tires exhibits the fastest growth rate (Acha and Brusoni 2005). Brusoni and Sgalari (2006) report that all of the main tire manufacturers have introduced some form of robotized production process. The trend toward more automation and flexibility in tire manufacturing dates back to the 1980s. As the automobile industry worked to improve car safety and performance and increased model variety, tire manufacturers strove to reduce average batch size and more efficiently customize tires, e.g., varying tire widths. Pirelli’s internal data show that between 1996 and 2004, the number of tire market segments nearly doubled, while average batch size dropped. Most improvement efforts

181

focused initially on the point in the manufacturing process where raw materials and components are all assembled to form a crude tire. The introduction of more flexible building machines reduced the average batch size and set-up times. Nevertheless, costs rose rapidly because material inventories serving as intermediate buffers for upstream process steps increased to accommodate the smaller batch sizes downstream. In the early 1990s, Pirelli was in a difficult situation for two reasons. First, the company specialized in the medium and high end of the market, precisely those segments in which car makers required better-performing and more customized tires, and where the trend toward segmentation was strongest. Second, Pirelli was finding it difficult to maintain its reputation as an innovative leader. Its innovative reputation was tarnished when it failed in its bid in 1992 to acquire Continental Tires, one of its main competitors. According to interviewees, some of Pirelli’s key customers saw this failed bid as a signal that Pirelli wanted to turn itself into a commodity tire producer. When this acquisition did not take place, the ownership grip of the Pirelli family loosened, and there were changes in top management, and a major strategic refocus. The first priority of Pirelli’s new top management was to solve the firm’s serious financial issues. They also reorganized Pirelli around core areas, exiting or downsizing unprofitable businesses, and investing to improve the efficiency and innovativeness of core businesses. Some of Pirelli’s nontire product divisions, e.g., fiber optics and real estate, began to overshadow the tire division. Archival data indicated that Pirelli’s market share by tire segment was static or declining. Moreover, in the early 1990s Michelin began to develop a new robotized production process that allowed it to produce small batches of large tires. Michelin, the market leader, was targeting exactly those segments that were Pirelli’s traditional strongholds. These developments characterized the context wherein Pirelli’s top management decided to reinvest to rebuild Pirelli’s tire-making capabilities and innovative reputation. Innovation was emphasized in the rhetoric used to discuss the new strategy, as Pirelli built a new headquarters and research facilities, launched new marketing campaigns, and searched for new distribution channels. In September 1997, the new CEO and owner (Marco Tronchetti Provera) gave Renato Caretta the task of developing a radically new production process to be the flagship of this renewed emphasis on innovation. Renato Caretta’s previous work at Pirelli had led to a number of patents related to process innovations, and his technical reputation within the company was strong. The CEO gave a broad mandate, reportedly saying, “Do whatever you want, but come back with something!” The investment in what would become the MIRS project was a statement of strategic intent from Pirelli’s CEO

Brusoni and Prencipe: Making Design Rules: A Multidomain Perspective

182

Organization Science 17(2), pp. 179–189, © 2006 INFORMS

to its competitors, particularly Michelin, and to Pirelli employees. In fact, the CEO presented MIRS as central to the strategy that would revitalize the entire group. Internal documents also explicitly emphasize that MIRS was to revitalize the firm. However, we found very few archival documents that mention any actual discussion about the decision to invest in MIRS. Interview data revealed that some personnel disagreed with the idea of investing Pirelli’s resources and reputation into what some perceived to be a very risky project. Michelin had already invested in equipment to robotize manufacturing, and so Pirelli was a late entrant. The strong commitment and direction of the CEO and Renato Caretta, however, was also clear. Rather than actively resisting the project, the evidence suggests a wait-and-see attitude by the skeptics, possibly influenced by the strategic turmoil they had witnessed during the unsuccessful attempt to take over Continental.

4.

How Did Pirelli Develop MIRS?

This section analyzes the differences between traditional and robotized tire manufacturing. It identifies their main design rules at the technological and organizational levels. It then examines the process through which such changes were introduced, shifting the focus toward knowledge-level changes. The traditional tire-manufacturing process is nonmodular: Plants are characterized by many nonstandardized organizational and technological interdependencies. Tire manufacturing has traditionally been a discontinuous process where raw materials such as polymers, chemicals and fillers for the rubber compounds, and steel reinforcements are preprocessed, stored in batches, and then cut into discrete components to be assembled into the crude tire. Extruders produce large bands of rubberized material semicontinuously. Plant operators manually shape these materials on flat supports moved about by the operators. Interdependencies are managed by maintaining inventories of raw materials, components, and intermediate products. Changes in product characteristics require extensive human intervention to adapt or change machinery. These manual changes can cause tire imperfections. Traditional tire design and production are highly specialized and disconnected activities. Tire designers, process engineers, and plant operators worked independently: Design and production processes are sequential operations. Communications among designers, engineers, and operators to fine-tune production take place through standardized organizational interfaces, e.g., paperwork exchanges. Tire designers focus on specific tire components, e.g., sidewalls, beads; or specific activities, e.g., the choice of materials, the mould design, etc. After identifying the tire’s performance targets, designers work in parallel on distinct tire or process

components, e.g., beads, treads, and moulds. Design data are not stored centrally, and computer-aided design is used only to a limited extent. Plant operators receive tire specifications from specialist designers and adapt plant machinery. They possess the skills to set up the manufacturing process and implement design specifications. Once the tires are produced, they are tested on cars. Feedback from testing is sent back to the tire designers, who implement changes and start another design cycle. The organization design emphasizes decoupling with few, predetermined interfaces between units and functions. Tacit knowledge underlies a lot of traditional tire manufacturing, and rules of thumb dominate tire design activities. Little articulated knowledge is available to predict how changes in specific tire characteristics may impact tire performance. The MIRS process, in contrast, is modular in nature. The crude tire is built directly onto a rigid drum so that manufacturing collapses into three steps: raw materials processing (done outside MIRS), building and curing, and finishing. Textile plies are knitted in situ around the tire, whereas bead wires—i.e., the string of rubberized metal that holds the tire to the rim, belts, and all reinforcement plies—are deposited onto the drum as preextruded tapes of rubber-coated cords. Each plant consists of a self-contained assembling and curing module. Within each module, pieces of equipment perform specific functions, e.g., each extruder deposits a specific component on the tire drum. Robots position the drums before the extruders, transport them between extruders and eventually to the vulcanizer. The process is scaled up by adding modules consisting of six coordinated robots and one vulcanizer, and so the system can make six tires at a time. After vulcanization, cured tires are removed from the drums for testing and distribution. There are no material inventories maintained between the building and curing processes. Each module is a closed unit that operates continuously. Workers do not access the module unless problems emerge. Modularity is also a property of the tire itself. Each tire component is decomposed into elementary stripes of rubberized materials that designers manipulate and optimize. The extruders operate discontinuously, as each extrudes a small amount of material on the drum. Discontinuous extrusion of small quantities of heterogeneous rubberized materials can lead to tire imperfections, e.g., materials deteriorate rapidly, and mechanical parts are subject to greater stress than in the traditional process. On the other hand, the decomposition of tire components into layers enables engineers to vary the application parameters within each layer and more exactly control tire quality. For example, designers can vary the angle of application or the thickness of certain components in different tire areas, e.g., closer to the centre or the sidewall to improve tire performance. This flexibility, which is characteristic


of the MIRS process, makes it particularly suitable for producing high-performing tires. MIRS requires an integrated organization of the design and production processes, and tire designers gain overall control of the manufacturing process. The dedicated software that controls the robots allows engineers to simultaneously design the tires and set the process parameters that will determine how the robotized production process is implemented. MIRS shifts skills and responsibilities away from plant operators to tire designers. Codification and integration characterize the MIRS underlying knowledge base, and MIRS plant control depends on an information technology infrastructure. Whereas in the traditional process plant operators set up the machinery, implement design changes, and control the process, in the MIRS process tire designers directly control the robots that produce the tires. The development of the IT infrastructure emerged as a major source of problems, and also opportunities, for the MIRS development team. Making the MIRS System How was MIRS developed? Renato Caretta received his extremely broad mandate directly and publicly from the group’s CEO and owner: He and his group were expected to revolutionize tire manufacturing, from rubber production to vulcanization and even tire distribution. The project started in September of 1997. He handpicked three engineers to lead the design group, based on his knowledge and trust of their technical capabilities. These mechanical engineers had extensive knowledge of manufacturing process and had worked with Caretta. Caretta encouraged open discussions and required intensive testing of ideas proposed. It soon became clear to the group that they needed to narrow their focus so that they could draw on the team’s core capabilities, which centered on their knowledge of tire manufacturing machinery. For example, raw materials preparation for robotized production was different from the traditional way of preprocessing raw materials. The challenge stems from the fact that tire components must be decomposed into very thin strips of materials, as opposed to the wider bands extruded in the traditional process. Raw materials consist of rubber and a plethora of additives. If the additives are unevenly distributed, some rubber layers in the crude tire may not contain all of them. This creates problems during vulcanization and in use. The group debated the raw materials issues at length and concluded that it was too complicated for them to solve, as it required chemical engineering skills that they did not have. Therefore, a separate and dedicated R&D unit was set up to work on the preprocessing of raw materials for the MIRS process. By the end of the first year, the MIRS group had focused on tire building and the vulcanization process.

183

Tire building was the obvious bottleneck in the manufacturing process. Flexible building machines for traditional production could accommodate decreasing batch sizes, but only at the price of higher inventory costs. One possibility was to build tires on a solid drum, an idea that dates from the 1920s. The advantage of a solid drum is that it eliminates several steps necessary to shape the flat support onto which tire components are extruded. These steps are a major source of product imperfections, and so eliminating them had immediate benefits. Caretta’s past work had led to patents on rigid drum technology, and so it seemed natural to focus on his unexploited knowledge. The bottlenecks to the development of solid drum technology were related to how the drums move and how composite material of adequate quality with the necessary uniformity can be extruded onto the drum. The group solved the latter problem. Robotized technologies solved the problem of moving the drums around. Having focused on tire building, the group sought to solve the important technical question: “Can rubber and other rubberized materials be extruded onto a rigid drum that moves from extruder to extruder?” Despite holding patents describing how to do this, nobody at Pirelli had actually implemented the process. Our tutor stressed that although he was the group leader, everyone in the group worked on this problem. Many anecdotes suggest the group was characterized by tight interaction, strong opinions, rewards for solutions that worked, and direct criticism when claimed solutions turned out not to work as expected. The main task was to prove that a crude tire could be built on a rigid drum: They built crude tires by driving the extruders by hand, and by carrying the tires to the vulcanizers to see what happened. The feasibility of building a tire in this way was quickly established. The group then focused on the overall logistics of the process, e.g., where to position the robots and the extruders. Solving this and related problems led to the first systematic use of computer-aided design. Two key problems faced the design team. First, they had to decompose the overall process into subtasks so that robots could sequentially perform them. Second, they had to decide how to position the robots. Positioning was constrained by the limited size of the factory floor, and so they had to work out a sequence of robot movements that allowed them to avoid colliding with one another, and also with a column positioned in the middle of the floor. It turned out that by adapting off-the-shelf equipment and using their expertise, they solved most of the hardware implementation problems. For example, they found that robotized arms to move crude tires could be bought from a specialized supplier and could be installed with little adaptation. Making MIRS: Developing the Organizational and Knowledge Architectures Having established the key plant-operating principles, the group had to make them work. Robotized production

184

requires a sophisticated IT infrastructure to pilot robots and control the manufacturing process. The traditional tire-manufacturing process did not rely on such technology, and so Pirelli had limited software-engineering capabilities. Our tutor attempted to establish a joint venture with a local university, but then decided it would be cheaper and faster to develop software development capabilities in-house. He hired two software engineers, and their original involvement was to develop software to pilot the robots. In MIRS, the timing of robot activities, e.g., how long robots are positioned in front of which extruder and robot positioning, or how to rotate the drum while the extruder emits a tire component, affect tire performance characteristics. Traditionally, the definition of tire performance characteristics is a designer responsibility, and machinery control is delegated to process engineers and plant operators. Now, tire designer expertise was required to define the robot’s operating parameters. To this point, however, MIRS had been a manufacturingoriented effort with no tire designer on the project team. The software engineers writing the software for synchronizing robot movements made it clear that tire designers were needed on the development project. In addition, then, two young tire design engineers were brought into the project, further broadening team composition. Once included in the MIRS team, the software and tire designers confronted a major task. Product design parameters were key inputs for the software programs that piloted the robots. Tire designers traditionally rely on tacit rules of thumb to evaluate relationships between design parameters and tire performance. In addition, traditional manufacturing allowed key operations to be performed manually by plant operators who evaluated whether, for example, materials should be stretched a bit more or less during the building phase. These operator skills were plant specific and machine specific. The robots’ tasks had to be clearly articulated and programmed before production began. Skills and expertise that in traditional tire-making processes were embodied in tacit problem-solving heuristics had to be articulated and codified for the new software-directed processes. The tire designers had to develop new knowledge in the form of explicit cause-effect relationships between tire design parameters, tire-building requirements, and tire performance. The role played by the software engineers in relation to the tire designers forced knowledge developments that were central to the transition from old to new design rules. The case of nylon 0 illustrates the opportunities that the new process made possible, and the new knowledge that was developed to exploit these opportunities. In a tire, layers of nylon (called nylon 0 ) hold together the various layers of rubberized material. The tension and angle with which the nylon layers are applied determine tire performance. In the traditional manufacturing


process, the application parameters of nylon cannot be varied and so angle and tension are the same throughout each batch of tires. The problem is that tires are asymmetrical and at high speed, and/or with high weight loads, these asymmetries create handling problems and reduce tire life. What can be done about this is limited in traditional manufacturing, for changing application parameters means stopping the line and manually adjusting the machines. Further, knowledge of the appropriate tension and application angle of nylon 0 to obtain particular tire performance depends on plant operators’ rules of thumb and experience with the manufacturing machines. In the new robotized process, however, tire designers can easily vary the angle and the tension of nylon, introducing the appropriate parameters through the digital control system. Optimal nylon deposits can also be designed to differ in different parts of tires to optimize high-speed performance. The possibility of systematically varying the application parameters in different parts of the tire, e.g., making the same tire component thicker or thinner according to the position of the specific layer with respect to the rim, made major improvements in tire performance possible. To do so, tire designers had to learn how such changes affected performance and then articulate this understanding to make it usable by the software programmers. As such adjustments were not possible using traditional technology, the tire designers had no knowledge about the actual impact of many parameters on tire performance. A machineset parameter for angle and tension became a product design variable that designers working with MIRS had to understand. For each type of tire developed on MIRS, and each tire component, engineers explored ranges of product and process parameters, produced the tire, and tested them on cars to evaluate their performance. The nylon 0 example captures the type of expertise that developed. Specifically, designers learned how to quantify the impact of parameter variations on tire performance. They also developed a more integrated understanding of the whole process as they learned about the chemical properties of the materials and the process machinery, e.g., extruder behavior. Variations once implemented by plant operators fell under the control of the designers. They also learned how to articulate cause-effect relationships to program the robotized plant through its IT interfaces. Over time and in contrast to the specialized and largely tacit knowledge on which traditional manufacturing depended, MIRS led to the development of integrated and articulated knowledge about the tire-manufacturing process. In fact, robotized production required an integrated (versus specialized) and articulated (versus tacit) knowledge base. The project group, including software engineers and tire designers, focused on determining and accumulating software design parameters from the end of 1999 to 2003, and this process continues. This


185


need for learning and articulation was unanticipated at the beginning of the project. All interviewees confirmed that this was a key opportunity uncovered by the project. Making MIRS: Learning Processes Gain Strategic Focus Whereas MIRS was validated technologically, the role it should play within Pirelli’s business strategy remained unclear. In early 2000, the development team had no tire product segments in mind that were suitable for the new process. Top management asked the project team to identify and deliver commercially viable products. Whereas it is technologically feasible to produce any tire with MIRS technology, the economics of MIRS makes it commercially viable only at the highest ends of the market. The market niches Lamborghini and Ferrari focused on are highly profitable, but they are also tiny, and traditional manufacturing processes served them adequately. Fortunately, two new and highly profitable niches emerged in the early 2000s: Run-flat tires that were linked to the launch of the new Mini, and ultra-high performance (UHP) tires for SUVs, the fastest-growing market niche in Europe. The rigid sidewall of the former, and the large section width of the latter, represented major challenges for the traditional tire-manufacturing process, but not for the MIRS robotized process. The development team quickly seized these opportunities offered them by top management to produce for these two market niches. They knew the technical characteristics of the MIRS process, and they knew that it could produce tires of the requested characteristics. Run flats and UHP were pivotal in turning MIRS into an economically viable endeavor, as they allowed Pirelli to use MIRS to commercialize new products on the basis of its newly developed, integrated body of knowledge. Further development proceeded rapidly: As the MIRS technology was associated with viable commercial products, additional skills developed, relying on the emerging IT infrastructure that allowed the MIRS system to be set up in other plants. MIRS plants were built in Germany, the United Kingdom, and the United States. The Milan centre now manages the databases and IT infrastructure on which all MIRS plants are built. However, MIRS has not replaced traditional manufacturing at Pirelli. The two processes coexist, but they specialize in different product niches. New connections have been established between the two processes. Based on the knowledge developed for MIRS, for example, process innovations—e.g., new ways to fold rubber and tire components in a more uniform and controllable way—have been introduced into Pirelli’s traditional manufacturing plants.

5.

Discussion: Modularization, Demodularization, and Pliotropy

Figures 1 and 2 summarize the key features of the traditional manufacturing and MIRS processes, respectively,

Figure 1

Traditional Tire Design and Manufacturing System

KNOWLEDGE DOMAIN Design rule: specialized knowledge

Rigid drum

Component design

ORGANIZATIONAL DOMAIN

Tread design

Design rule: modular and sequential design process

Choice of raw materials Choice of equipment

Chemicals

TECHNOLOGY DOMAIN Design rule: nonmodular production process

Textiles

Steel wires

Overlapping processing steps

Curing

Building

Note. Continuous arrows lines represent strong connections among domains. Dotted arrows represent weaker connections. Horizontal continuous lines represent evidence of decoupling among phases or activities. Horizontal dotted lines represent evidence of tighter coupling among phases or activities. Source. Adapted from Padgett and Powell (2003) based on authors’ interviews. Figure 2

The New Tire Design and Manufacturing System

KNOWLEDGE DOMAIN Integrated know-how

Design rule: integrated know-how

ORGANIZATIONAL DOMAIN

Tire X Tire Y Tire Z Integrated Integrated Integrated design and design and design and process process process setting setting setting

Design rule: integrated design process

TECHNOLOGY DOMAIN Design rule: modular production and modular product

Chemicals Module

Module

Textiles Module

Module

Steel wires Module

Module

Note. Continuous arrows lines represent strong connections among domains. Dotted arrows represent weaker connections. Horizontal continuous lines represent evidence of decoupling among phases or activities. Horizontal dotted lines represent evidence of tighter coupling among phases or activities. Source. Adapted from Padgett and Powell (2003) based on authors’ interviews.

186

in terms of technological, organizational, and knowledge domains. The figures capture the key empirical study result: The adoption of modular design principles for the tires and the plant did not lead to a modular organization. Rather, modularization at the product and plant levels led to a process of demodularization and integration at the organization level. Changes in organization were driven by the changes in engineering know-how, which were in turn triggered by the adoption of new core elements, i.e., the rigid drum. More specifically, it was integration within the knowledge domain that enabled the effective modularization of the technological domain. Our study describes how the evolutionary dynamics of three domains complement one another, but reached different end states: The nonmodular manufacturing process became modular, while the modular organization became highly integrated. In other words, it is not products that design organizations. Knowledge does. Our results are consistent with and extend research on the relationship between product and organization design. Drawing on an empirical study on automakers’ management of suppliers’ involvement in product development in Japan, Takeishi (2002) showed that whereas the actual tasks of design and manufacturing are outsourced, automakers retain relevant knowledge in order to obtain better component design quality. His results illustrated how patterns of knowledge partitioning differed from patterns of task partitioning. In describing systemic innovation in the hard disk drive industry, Chesbrough and Kusunoki (2001) also illustrated different organizing principles for product and knowledge domains. Our analysis is consistent with recent research on organizational evolution. Using the terminology of Siggelkow (2002), the emergence of the rigid drum element constitutes an example of patching, i.e., the creation of a new core element. This element became core as it generated new connections and feedback loops with other elements within the knowledge domain and also with other domains. Siggelkow (2002) labeled this process thin-to-thick. In parallel to the thin-to-thick process, Pirelli underwent a patch-by-patch process (Siggelkow 2002) as new core elements were added over time as necessary for the development of MIRS, e.g., digital control of robots actions; integrated database management systems. These two processes led to a complete cross-domain rewiring leading to a radical innovation. The analysis of this rewiring of connections among domains (each organized according to its own design rules, summarized in Figures 1 and 2) highlights the importance of within- and across-domain connections in order to introduce radical innovations. Such connections are the mechanisms through which design rules develop that also become hidden in organizational routines and IT systems. Our case shows that task requirements set in motion cross-domain connections: Software engineers played a catalytic role in generating new connections among product and process engineers and across


organizational units. They helped develop new skills and capabilities by combining existing ones, by pushing engineers to articulate their tacit heuristics, and by creating feedback loops between areas that were previously independent. This is consistent with Padgett and Powell’s (2003) work on the complexity of the connections that developed between people in biotechnology and banking in Renaissance Florence. Padgett (2001) also emphasized that enabling connections for radical change differ from the connections that characterize incremental evolution. He labeled pliotropy the presence of nonmodular connections across domains that are necessary to generate nonincremental evolution. In our case, pliotropy emerged in response to the initial effort to develop a manufacturing process around the existing core of rigid drum patents. Pirelli Tires is keenly aware of the importance of connections between manufacturing processes and product know-how. Many interviewees stressed that what allowed Pirelli to overtake Michelin was their clear focus on market high-end products, as Michelin used robotized processes on standard tires and so did not make connections concerning how its new technological achievements could differentiate its business strategy. Figure 3 illustrates the transition from old to new design rules and highlights the richness and variety of connections instrumental to the emergence of new design rules. During the transition phase, design rules are visible (Figure 3). Once established and accepted, they become hidden (Figures 1 and 2). During the 1980s and early 1990s, Pirelli’s research effort generated patents that remained unexploited. When it was created, this new element (represented by a circle in Figure 3, top left) was not central to the organization, as it was linked only to the tire assembly phase (the dotted arrow) and had no feedback loops to other elements in other domains. When a financial crisis, a change in ownership, and the appointment of a new CEO as well as the emergence of new market niches and fierce competition had to be dealt with, everything changed and this unexploited element became core to Pirelli’s new strategy. It was only after 1997 that the rigid drum element developed within the organizational domain (solid arrow from top to middle domain in Stage 2). Renato Caretta led this process and became a central figure in all domains: an experienced process engineer (bottom domain), the inventor of all of the rigid drum-related patents (top), and the leader of the MIRS group (middle). Once the design group validated the technology, the group was enlarged to include software and product engineers. The latter were instrumental to developing the integrated know-how (the dotted circle in the knowledge domain in Stage 3), as they established and strengthened new connections within the organizational and the knowledge domains (between plant operation and tire design), and multiplying across-domains connections (solid arrows).


187


Figure 3

Pliotropy in the Evolution of Tire Manufacturing Technologies

Stage 1 (1997)

Stage 2 (1997 –1999)

Stage 3 (1999 – 2001)

Stage 4 (2001– … ) Integrated know-how

Integrated organization

Modular production Note. Continuous arrows lines represent strong connections among levels or phases. Dotted arrows represent weaker connections. Horizontal continuous lines represent evidence of decoupling among phases or activities. Horizontal dotted lines represent evidence of tighter coupling among phases or activities.

Eventually, a new configuration emerged characterized by new design rules (Stage 4). The new engineering know-how became embodied into a new and integrated organization, within which MIRS design rules are hidden in routines, organizational practices, and IT systems. The analysis of the MIRS developmental path in terms of emergence of new core elements and reinforcement of these connections within and across domains constitutes a framework that may be used to describe the developmental paths of the dynamics of change in other organizations. The framework helps depict the evolutionary dynamics of firms as they build on stable configurations interspersed with prolonged periods of turmoil that ultimately usher in new design rules. Within each configuration, connections between domains are stable and predictable. Changes in configurations are associated with an explosion of new connections both within and across domains. The MIRS case provides an example of the nonmodular nature of connections that allowed a transition from one configuration to another. The case also suggests that there may be no one best way to design a radical change process. Several transition paths from one configuration to another are possible because different domains may be informed by different design rules, and yet still act as a coordinated system. The MIRS case illustrates again that coordination does not necessarily lead to similarity in the organizing principles of different domains. Focusing on the three leading firms in the aircraft engine industry, for example, Brusoni et al. (2001) analyzed the dynamics of changes in product, technology, and organization that accompanied the introduction of a radical shift in one technology component. They concluded that all three firms completed the transition, yet they relied on different strategies to do so, and they ended up relying on different configurations. The three engine manufacturers followed dissimilar developmental paths, evolving toward configurations of product, organization, and technology domains that obeyed different design rules.

6.

Conclusions

By understanding new design rules as a process involving the emergence of new connections within and across domains, this paper provides a bridge between modularity research (Sanchez and Mahoney 1996, Baldwin and Clark 2000, Schilling 2000) and work that conceptualizes organizations as networks of interconnected elements (Perrow 1967, Lawrence and Lorsch 1967, Miller 1987, Levinthal 1997, Rivkin 2000). This introduction of new design rules requires coordinated changes across domains that include not only products and organization, but also evolving bodies of knowledge. Changes in knowledge, organization, production process, and product are linked in a continuous seesaw pattern characterized by both designed and emerging features, ridden with unexpected and unanticipated events, and stop and go, ad hoc organizational and technological adjustments. The case highlights how changes occur both within and across domains. We think the latter represents the most critical part of the story, and these were the changes that most intrigued people at Pirelli. If changes are not introduced in a coordinated and timely manner among all domains, radical innovations will not occur. Crossdomain connections were enabled by hiring people with specialized skills, i.e., software engineers, an unanticipated development that then activated links and feedback loops to other specialized engineers. Such links eventually generated new sorts of awareness and knowledge, enabled new and stable organizational connections, and finally brought about major, commercially viable changes at the process and product levels. This evolving fit among different domains does not imply similarity. Whereas a domain is organized modularly, others may not be. By focusing and analyzing the microlevel processes whereby an established organization introduced new core elements, how they did it, and when they did it, this paper extends research on how established organizations may adopt technologies or strategies that are

188

different from their traditional ways of doing business. Dunbar et al. (1996) discussed how organizations change their frames, i.e., those values and principles that inform their selection criteria and define their decision-making premises. In terms of the framework proposed in this paper, such issues can be analyzed in terms of missing connections among different domains. As new connections are introduced, they allow new frames to emerge. The analysis also links to research that has analyzed the brokering and integrating roles played by focal individuals or organizations within networks. Focal individuals associated with successful product and process innovations act like champions (Allen 1977, Roberts 1987, Rothwell 1992). Technology brokers recognize, store, blend, and transform technologies (Hargadon and Sutton 1997) and in doing so, they enact cross-domain rewirings (Padgett and Powell 2003). Our case study illustrates the unexpected but explainable nature of the development process that leads to new design rules. Elements of design interact continuously and lead to emergent needs and opportunities. The dynamics of a new technology can intersect with existing organization structures at strange angles, and require adjustments on all sides. Design rules play a fundamental role in explaining the evolution of products, while the rules of design remain largely people embodied. The evolution of design rules depends on the skills and capabilities of designers.


References

Acha, V., S. Brusoni. 2005. Complexity is in the eye of the beholder: Lessons from the tire industry. J. Finch, M. Orillard, eds. Complexity and the Economy. Edward Elgar, Cheltenham, UK. Adler, P. A., P. Adler. 1987. Membership Roles in Field Research. Sage Publications, Newbury Park, CA. Agar, M. H. 1980. The Professional Stranger—An Informal Introduction to Ethnography. Academic Press, New York. Allen, T. J. 1977. Managing the Flow of Technology: Technology Transfer and Dissemination of Technological Information Within the R&D Organization. MIT Press, Cambridge, MA. Baldwin, C. Y., K. Clark. 2000. Design Rules, Vol. 1. The Power of Modularity. MIT Press, Cambridge, MA. Brusoni, S., G. Sgalari. 2006. New combinations in old industries: The introduction of radical innovations in tire manufacturing. J. Evolutionary Econom. 16(1–2) 25–43. Brusoni, S., A. Prencipe, K. Pavitt. 2001. Knowledge specialization, organizational coupling, and the boundaries of the firm: Why do firms know more than they make? Admin. Sci. Quart. 46(4) 597–621. Chesbrough, H., K. Kusunoki. 2001. The modularity trap: Innovation, technology phase-shifts, and the resulting limits of virtual organizations. I. Nonaka, D. Teece, eds. Managing Industrial Knowledge. Sage Press, London, UK, 202–230. Creswell, J. W. 2003. Research Design. Sage, Thousand Oaks, CA. Drazin, E., A. H. Van de Ven. 1985. Alternative forms of fit in contingency theory. Admin. Sci. Quart. 30(4) 514–539. Dunbar, R., R. Garud, S. Rahuram. 1996. A frame for deframing in strategic analysis. J. Management Inquiry 5(1) 23–34. Flick, U. 2002. An Introduction to Qualitative Research. Sage, London, UK. French, M. 1991. The U.S. Tire Industry: A History. Twayne Publishers, Boston, MA.

Acknowledgments

The authors wish to thank Anna Canato, Eugenia Cacciatori, Anna Grandori, Michael Hobday, and John Padgett for their comments on earlier drafts of the paper. Special thanks go to the editors of this Organization Science special issue, Roger Dunbar and William Starbuck, and two anonymous referees for their advice and encouragement to refine the authors’ thinking. The authors would also like to thank participants in the workshop on organization design held at New York University in June 2004, as well as Ammon Salter and Sidney Winter and other participants in the DRUID Conference held in Copenhagen in June 2005. The time and efforts of the management and engineers of Pirelli Tires are gratefully acknowledged. Among them, the authors owe a particular debt of gratitude to Renato Caretta and Giorgia Sgalari. None of the opinions reported in the paper can be attributed to Pirelli Tires, the Silvio Tronchetti Provera Foundation, or any of their employees. They reflect the authors’ own interpretation of the case study data. Partial funding from the European Commission (Contract CIT2-CT-2004-506022; KEINS—Knowledge-based Entrepreneurship) is gratefully acknowledged. The usual disclaimers apply.

Galbraith, J. A. 1973. Designing Complex Organizations. AddisonWesley, Boston, MA. Glaser, B. G., A. L. Strauss. 1967. The Discovery of Grounded Theory: Strategies for Qualitative Research. Aldine, New York. Grandori, A. 2001. Neither hierarchy nor identity: Knowledge governance mechanisms and the theory of the firm. J. Management Governance 5 381–399. Hargadon, A., R. Sutton. 1997. Technology brokering and innovation in a product development firm. Admin. Sci. Quart. 42 716–749. Langlois, R. N. 2003. The vanishing hand: The changing dynamics of industrial capitalism. Indust. Corporate Change 12(2) 351–385. Lawrence, P., J. Lorsch. 1967. Organization and Environment. Harvard University Press, Cambridge, MA. Levinthal, D. 1997. Adaptation on rugged landscapes. Management Sci. 43 934–950. Miller, D. 1987. The genesis of configuration. Acad. Management Rev. 12(4) 686–701. Padgett, J. F. 2001. Organizational genesis, identity and control: The transformation of banking in Renaissance Florence. J. E. Rauch, A. Casella, eds. Networks and Markets. Russell Sage, New York, 211–257.


189

Padgett, J., W. Powell. 2003. Economic transformation and trajectories: A dynamic multiple network approach. Accessed July 13, 2005. http://home.uchicago.edu/∼jpadgett/papers/sfi/ intro.chap.pdf/.

Siggelkow, N. 2002. Evolution toward fit. Admin. Sci. Quart. 47 125–159.

Perrow, C. 1967. A framework for the comparative analysis of organizations. Amer. Sociol. Rev. 79 686–704.

Stake, R. E. 1995. The Art of Case Study Research. Sage, Thousands Oak, CA.

Prencipe, A. 1997. Technological capabilities and product evolutionary dynamics: A case study from the aero engine industry. Res. Policy 25 1261–1276.

Strauss, A. L. 1987. Qualitative Analysis for Social Scientist. Cambridge University Press, Cambridge, UK.

Rivkin, J. W. 2000. Imitation of complex strategies. Management Sci. 46(6) 824–844. Roberts, E. 1987. Generating Technological Innovation. Oxford University Press, New York. Rothwell, R. 1992. Successful industrial innovation: Critical factors for the 1990s. R&D Management 22(3) 221–239. Sanchez, R., J. T. Mahoney. 1996. Modularity, flexibility, and knowledge management in product and organization design. Strategic Management J. 17(Winter Special Issue) 63–76. Schilling, M. A. 2000. Towards a general modular systems theory and its application to inter-firm product modularity. Acad. Management Rev. 25(3) 312–324. Siggelkow, N. 2001. Change in the presence of fit: The rise, the fall,

and the renaissance of Liz Claiborne. Acad. Management J. 44(4) 838–857.

Sturgeon, T. 2002. Modular production networks: A new model of industrial organization. Indust. Corporate Change 11(3) 451–496. Sull, D. N., R. Tedlow, D. N. Rosenbloom. 1997. Managerial commitments and technological change in the US tire industry. Indust. Corporate Change 6(2) 461–501. Takeishi, A. 2002. Knowledge partitioning in the inter-firm division of labor: The case of automotive product development. Organ. Sci. 13(3) 321–338. Thompson, J. D. 1967. Organizations in Action: Social Science Bases of Administrative Theory. McGraw-Hill, New York. Whittington, R., A. Pettigrew, S. Peck, E. Fenton, M. Conyon. 1999. Change and complementaries in the new competitive landscape: A European panel study, 1992–1996. Organ. Sci. 10(5) 583–600. Woodward, J. 1965. Industrial Organization: Theory and Practice. Oxford University Press, London, UK.

Making Design Rules: A Multidomain Perspective - INFORMS ...

Making Design Rules: A Multidomain Perspective - INFORMS ...

Suggest Documents