Embed Size (px)
Citation preview
1
SIMULATING THE COMPLEX CLADISTIC EVOLUTION OF MANUFACTURING: A TOOL FOR SUSTAINABLE RE-ENGINEERING?
‡James Baldwin*: [email protected] Peter M. Allen†: [email protected] Belinda Winder*: [email protected] Keith Ridgway*: [email protected]
*Ibberson Centre, Department of Mechanical Engineering, University of Sheffield.
†Complex Systems Management Centre, Cranfield University.
‡Corresponding author
PAPER FOR THE SPECIAL SESSION AT THE ASEAT/IOFIIR CONFERENCE ON INDUSTRIAL ECOLOGY, 7-9 APRIL 2003, MANCHESTER, UK
Abstract Although there are now many models and tools for the sustainable development of industry, it appears that problems now lie in their incorporation into existing industrial practices. Uncertainty in decision-making and unknown barriers are now deemed to be behind the faltering shift to more sustainable technical, organisational and social advances. To combat similar problems, manufacturing cladistics was first developed in the early 1990s not only as a means of classifying manufacturing organisations but also, and perhaps more importantly, as a tool to guide organisational re-engineering. The aim of this paper is to promote debate as to the applicability of the cladistics approach for sustainable industrial development. Having reviewed the development of manufacturing cladistics as well as some potential shortcomings when applied to social systems, the results of a recent exploratory study are presented. The study was an attempt to integrate manufacturing cladistics with evolutionary complex systems methodology. Finally, the complex cladistic evolution of sustainable manufacturing is considered, particularly, the methodology that would be required to both select characteristics such as ‘design for environment’, ‘waste minimisation’ and ‘energy cascading’, and to gauge their interaction. The aim would be to explore the evolutionary differences between sustainable and non-sustainable organisations, to guide re-engineering, and, through the use of evolutionary complex systems methodology, explore new structures that could offer industry novel solutions for sustainable and competitive change programmes. 1. Sustainable Industrial Development In terms of sustainable development, industry is in a unique position as both the cause of most environmental and social problems either directly or indirectly, and, at the same time, the main mechanism for change, i.e. economic growth. The manufacturing sector is vital, and is the cornerstone for the continued economic health of the UK, but as Our Common Future (WCED, 1987) points out, sustainable development must be ‘development which meets the needs of the present without compromising the ability of future generations to meet their own needs’. Strategies, models and tools that have been developed for, and are specific to the sustainability of business and industry include, among others: cleaner production principles; waste auditing and waste minimisation; life cycle assessment; design for environment; zero emissions; resource suitability and cascading; and, social and environmental management systems and auditing techniques (including Investors in People, Ethical Trading Initiative, ISO 14000, and EU Eco-Management and Audit Schemes). One of the main barriers now to sustainable industrial development is not the lack of strategies, models and tools, but how to implement them, and more importantly how to introduce them into existing practices whilst still maintaining or ideally improving competitiveness. The ESRC, in a recent call of the Sustainable Technologies Programme, recognises many of these general problems and dilemmas. There are three main areas of further research that need pursuing. The first area concerns, for example, questions as to how sustainable systems evolve and how practices emerge and what processes are involved in the transformations from non-sustainable to
2
sustainable manufacturing? Similarly, how do manufacturing systems rise to domination, are path-dependent processes involved, how do certain systems or practices ‘lock-in’, and how can these systems be replaced? How is diversity and flexibility retained – through organisational customisation? How do innovations on technical, organisational and social levels interact in transformations? How can timescales be reduced in these kinds of transformations? A second area involves the decision-making processes of management. Unpredictability, risk and uncertainty are all important factors in decision-making. For example, how certain are the estimated impacts of certain technologies or practices in terms of economics, society and the environment? How can uncertainty and risk be reduced? How much do foresight and precaution play in certain decisions? Why do technologies and practices succeed in one organisation but not in another? A final area of further research must involve questions relating to the interaction of existing technologies and practices with new and sustainable ones. What barriers, institutional, psychological and economic, exist in the adoption of sustainable technologies? Can all costs, benefits and risks be calculated beforehand? Are benefits short or long term and are they evenly and fairly distributed? The aim of this paper is to introduce two areas of research that have the potential to provide solutions to many of these problems or at least create new knowledge to help answer some of the most pressing questions. The first area is manufacturing cladistics, a classification system where best practice can be mapped that allows organisations to locate their position in evolution, the position of competitors and the chance to re-engineer their organisation using the classification as a guide. The second area of research is a combination of manufacturing cladistics and evolutionary complex systems’ modelling. One of the first applications of this approach was to model the evolution of population of species in ecosystems. Using an evolutionary framework designed to model, through simulation, the evolution of manufacturing form, new structural organisations may be explored. In addition, the introduction and impacts of new technologies and practices to the existing structure may also be investigated. This synthesis of approaches may turn out to be an exciting development and a fresh perspective for industrial ecology. 1.2. Manufacturing Cladistics McCarthy, Ridgway and others (e.g. McCarthy & Ridgway, 2000; McCarthy, Leseure, Ridgway & Fieller, 1997) first developed manufacturing cladistics as not only a classification scheme, but as an aid for organisational change initiatives. When classifications are applied to manufacturing, McCarthy et al (2000; p. 78) argue that a classification ‘would facilitate the storage, alignment and development of structural models of manufacturing systems [that] . . would provide researchers and consultants with a generic library of structural solutions for enabling manufacturing systems to maximise their operating effectiveness’. In classification science there are two main biological principles – phenetics and phylogenetics. From these, three main classification disciplines have emerged – the phenetic, evolutionary and cladistic approaches. These schools can be thought of as lying on one dimension – the evolutionary scale. Phenetics is non-evolutionary and lies at one extreme. Cladistics in based purely on evolutionary principles and lies at the other extreme. Evolutionary classifications are a synthesis of the two and lie in the middle of the scale (McCarthy, 1995a). Ridley (1993), after reviewing the three schools to assess their ability to construct natural and objective classifications (rather than artificial and subjective), concluded that only cladistics could fully satisfy these criteria. As it is an evolutionary classification scheme it not only describes the attributes of existing entities but also the ancestral characteristics. The more distant the entities, the further apart their respective positioning in the classification (see figure 1). By using evolution as an external reference point (evolutionary history cannot be changed), classifications will be unique and unambiguous. Cladistics therefore is the most accepted approach in biology and as such is the most appropriate starting point for manufacturing classifications (McCarthy et al, 2000).
3
Figure 1. A taxonomic hierarchy presented dendrogrammatically (adapted from McCarthy, 1995b)
McCarthy et al (1997) lists seven basic steps in constructing a manufacturing cladogram:
1. Determine the clade (taxon). In terms of manufacturing this would be the industrial sector – the organisation of interest as well as its common ancestors.
2. Determine the characters. These are variables, features or attributes from which comparisons may be drawn. The character, however, must be of an evolutionary nature.
3. Code the characters. This step in the process deals with character labelling, so that decisions can be made whether they exist in the organisations under study.
4. Ascertain character polarity. Characters may be primitive or derived. Primitive characters are present in ancestors whereas derived characters are not.
5. Construction of the conceptual cladogram. This stage is concerned with constructing a ‘best estimate’ cladogram from historical accounts.
6. Construction of the factual cladogram. Data collection consists of material from contemporary organisations, through interviews, questionnaires, and company records. This is then combined with the conceptual cladogram, producing a full cladogram.
7. Decide taxa nomenclature. This stage deals with the labelling of the manufacturing systems. Labels should convey the essence of the entity, convey the main characters, indicate the evolutionary position, be unambiguous, and ensure universal communication.
McCarthy (1995b) was the first to pursue this approach in depth and was primarily concerned with the process of actually constructing a manufacturing cladogram, the detailed analysis of the appropriate statistical procedures available and the thorough exercise of comparing and contrasting the results obtained from the different clustering techniques. As much of the research was concerned with the development of this process, the main work in the disciplines of organisational systematics, numerical taxonomy and biological classifications was also reviewed. Another important feature of this research was the procedural translation to manufacturing and industry and to the social sciences in general. There are now several good examples of the application of manufacturing cladistics including classifications based on the hand-tool industry (Leseure, 1998; Leseure, 2000), management styles (Goh, 2000) and the automotive industry (Leseure, 1998; McCarthy et al 1997). The cladistics research concerning the automotive industry
4
is discussed in more detail next as it provides the basis for the further collaborative research conducted by Cranfield and Sheffield Universities reported below. 1.3. The Automotive Industry McCarthy et al’s (1997; see also Leseure, 1998) paper provided an illustrative example (see figure 2 and table 1) of the automotive assembly plant cladogram to ‘introduce the reader to the mechanics and benefits of producing a cladistic classification’ (p.272). The data for this research came from a number of sources including the International Motor Vehicle Program (cited in Womack et al, 1990), historical accounts and information from car manufacturers. Figure 2. A cladogram of automotive assembly plants (from McCarthy et al, 1997)
Table 1. Characteristics of automotive assembly plants (from McCarthy, 1997)
1. Standardisation of parts 2. Assembly time standards 3. Assembly line layout 4. Reduction of craft skills 5. Automation (machine paced shop) 6. Pull production system 7. Reduction of lot size 8. Pull procurement 9. Operator based machine maintenance 10. Quality circles 11. Employee innovation prizes 12. Job rotation 13. Large volume production 14. Suppliers selected primarily on price 15. Exchange of workers with suppliers
16. Socialisation training (master/apprentice learning) 17. Proactive training programs 18. Product range reduction 19. Automation 20. Multiple sub-contracting 21. Quality systems (tools, procedures, ISO9000) 22. Quality philosophy (TQM, way of working, culture) 23. Open book policy with suppliers; sharing of cost 24. Flexible multi-functional workforce 25. Set-up time reduction 26. Kaizen change management 27. TQM sourcing; suppliers selected on basis of quality 28. 100% inspection/sampling 29. U-shape layout 30. Preventive maintenance
31. Individual error correction; products are not re-routed to a special fixing station 32. Sequential dependency of workers 33. Line balancing 34. Team policy (motivation, pay and autonomy for team) 35. Toyota verification of assembly line (TVAL) 36. Groups vs teams 37. Job enrichment 38. Manufacturing cells’ 39. Concurrent engineering 40. ABC costing 41. Excess capacity 42. Flexible automation for product versions 43. Agile automation for different products 44. Insourcing
45. Immigrant workforce 46. Dedicated automation 47. Division of labour 48. Employees are system tools and simply operate machines 49. Employees are system developers; if motivated and managed they can solve problems and create value 50. Product focus 51. Parallel processing 52. Dependence on written rules; unwillingness to challenge rules as the economic order quantity 53. Further intensification of labour; employees are considered part of the machine and will be replaced by a machine if possible
5
The Ancient Craft Shops of the 1880s consisted of highly skilled workers who produced unique customised cars and are now extinct (with the exception of enthusiasts). The first major innovation, the standardisation of parts (character state (CS) 1), was introduced by Henry Ford and resulted in the species Standardised Craft Systems. This species did not last long, however, and soon evolved into Modern Craft Systems with the introduction of assembly time standards (CS2) and a division of labour (CS 47). The next evolutionary step to the Neo-Craft Systems, also initiated by Ford, was characterised by pursuing a large volume of production (CS 13) to achieve economies of scale, to treat employees as system tools that simply operate machinery (CS 48), and to focus more on the product (CS 50). As a tighter control over the layout of the assembly line (CS 3) proceeded, as well as explicit master-apprentice training (CS 16), and with a system where one worker is dependent on the previous worker (CS 32), the Skilled, Large Scale Producers emerged. A similar species, the Large Scale Producers, evolved as it became evident that if systems were introduced that reduced the skill requirements of the workers (CS 4) then cars could be manufactured a lot cheaper. Shortly after this, automation with machine-paced shops (CS 5) began to increase in importance and is a dominant characteristic of all other species that followed. In the cladogram, it can be seen that there is a major bifurcation point, with many character states involved, leading to two families – the Mass Producers and the Lean Producers. The classical Mass Producer, now named after Ford, introduced more dedicated automation (CS 46), utilised multiple subcontractors (CS 20), selected suppliers purely on price (CS 14), and introduced middle management with the consequence of an increasing dependence on rules and procedures (CS 52). Two sets of two species then branch from the classical Mass Producer. Down one branch are located the Modern Mass Producer and the Pseudo Lean Producer. Both species have pull production systems (CS 6) and employ preventive maintenance (CS 30), but what distinguishes the Pseudo Lean Producer is a reduced lot size (CS 7), a quality system in place (CS 21), a flexible and multi-functional workforce (CS 24), a focus on the reduction of set-up times of the machinery (CS 25), a line balancing procedure (CS 33) where workers from one line help on another line to help time pressures, and an emphasis on teamwork (CS 35). The species of the other branch, the European Mass Producers and the Intensive Mass Producers, exhibit none of the eight characteristics just mentioned, but both share the emphasis of a further intensification (CS 53) of mainly immigrant labour (CS 45). In contrast to the European Mass Producer, the Intensive Mass Producer also exhibits a reduction in the product range (CS 18) and utilises insourcing (CS 44). Unlike all other Mass Producers, the Intensive Mass Producer does not rely on multiple subcontractors (CS –20). There are five species identified on the ‘leaner production’ evolutionary branch. All the following species share all of the characteristics of the Just in Time Systems and include the implementation of the pull production system (CS 6), specialised ‘autonomation’ (CS 19) along with preventive maintenance (CS 30). Also dominant in these species are quality systems (CS 21), quality circles (CS 10), and an overall quality philosophy (CS 22). In line with the quality emphasis, everything is inspected (CS 28) and errors are corrected there and then (CS 31) instead of being sent to a special fixing station, which helps reduce excess capacity (CS 41). The concept of Kaizen change management (CS 26), where step-by step improvements are always sought, is also implemented. There are several differentiating characteristics that define Flexible Manufacturing Systems – characteristics that the next three species also share. These are the implementation of flexible automation, which in consequence reduces both the set-up times of the machinery (CS 42) and the lot size (CS 7) required for operations. There is also an emphasis on teamwork (CS 34; CS 36) and job enrichment (CS 37) to create a flexible multi-functional workforce that can help balance the line (CS 33) when needed. The factory layout changes with an emphasis on a U-
6
shape layout (CS 29) organised in cells (CS 38). ABC costing (CS 40) is also employed. The Toyota Production System has additional characteristics which are pull procurement planning (CS 8) with suppliers selected on the principle of Total Quality Management (CS 27) to complement the pull production system in place. The workers are also emphasised and seen as system developers (CS 49), jobs are rotated (CS 12) and employee innovation prizes (CS 11) are given. This manufacturing system also introduced the ‘Toyota Verification of Assembly Line’ (CS 35), which is a motion study procedure that takes into account repetitiveness and physical strain for job rotation planning. Both the Lean Producer and the Agile Producer, evolved from this species. Both species added concurrent engineering (CS 39), which is a concerted collaboration between design and manufacturing departments. The species also concentrated on improving their ties with the suppliers through exchanging workers (CS 15) and keeping an ‘open book policy’ (CS 23) where costs and profits are shared. Agile Producers, as their name suggests, invested in agile automation for different products (CS 43) and performed processes in parallel (CS 51). As discussed above, manufacturing cladistics was developed not only as a classification scheme but also as a guide for manufacturers. There are three main ways which this approach may aid manufacturers. The first is by way of benchmarking best practice – manufacturers are able to locate their position on the cladogram and the position of their main competitors. The second source of aid is through the use of the cladogram to guide change management and major organisational transformations (although this hasn’t been attempted in practice). The third way this approach may be of aid, as Leseure (2000) has demonstrated, is through developing strategies for particular companies by identifying problem areas through the cladogram. However, there are limitations. This is because the cladogram is essentially a description of the past and as such is only useful to manufacturers who are not competitive. It is of no use for market leaders that are typically world-class manufacturers and may only compare with inferior or unrelated organisations. In addition, the cladogram gives no insight, with the exception of post-hoc analyses, into many of the problems and questions, highlighted in the introduction. In other words, manufacturing cladistics, as it is a representation of the past, gives no indication of what may happen in the future. 1.4. Evolutionary Complex Systems’ Modelling This problem brings us to the main objective of this paper – the potential synthesis of manufacturing cladistics and evolutionary complex systems’ (ECS) modelling. With ECS modelling, structures and the organisation of different practices may be explored through evolutionary ‘runs’. The methodology employed and the preliminary results of the modelling simulations are presented and the potential applications of this approach are discussed. Evolutionary complex systems’ thinking emerged in the 1970s and the work presented here traces its origins back to the insights expressed in Prigogine’s (1973) Nobel Prize winning research. Over the past three decades, research into non-linear processes and open systems have revealed insights into both complexity and evolution. To understand evolution, the objects or phenomena have to be dated and classified which leads to an evolutionary tree that will inevitably indicate certain discontinuities and internal bifurcations as well as invasions corresponding to the introduction of new activities. If these new activities grow, then co-evolutionary processes are involved where the new activity shapes the landscape which in turn shapes the new activity and so on. In terms of manufacturing a new characteristic shapes the organisation that in turn may feed back and influence many, or even all of the other characteristics. Evolutionary models must capture this co-evolution of characteristic and organisation. Allen (1988; 1997) points out that there is a hierarchy of models, where stronger
7
assumptions reduce complex reality to increasing simplicity of understanding. With reality there are no assumptions. Common assumptions in modelling are:
1. That a boundary exists between the system of interest and its environment. 2. That objects are classified resulting in a taxonomy of the components in the system. 3. That the components of the systems are of an average type. In other words components
are homogeneous that are made up of sub-components with diversity normally distributed around the mean. In addition to this the components do not change as a result of their experiences – learning is absent.
4. That the collective or overall behaviour of the system (the variables) results from smoothed, or average processes. This means that the variables are changed by smooth, fixed mechanisms.
5. That the system moves to or is already at a stationary or equilibrium state or has a stable solution. It is assumed that the relationships between variables are fixed and unchanging. Each variable value is derived from the function of the other variable values.
Allen (1998; 2000; 2001) argues that the following four models relate directly to the number of simplifying assumptions made. If all five assumptions are justified the result is an equilibrium model. Although extremely simple, the model appears to enable perfect long-term prediction and complete understanding and knowledge. The values of every variable and their interactions are precisely known. So, for example, in these types of models the outcome of a particular decision or action is known beforehand whereas in reality, of course, consequences of decisions/actions are only known after. Feedback processes, particularly positive feedback, where growth leads to more growth and decline to more decline, are ignored. In short, equilibrium models are essentially only a descriptive approach applied in post-hoc situations where consequences from decisions/actions have already happened. Despite the known limitations of this approach, equilibrium models are still used today to deal with problems in economics, spatial geography and in environmental science (Allen, 1998; 2000; 2001). When making the first four assumptions but not equilibrium, the result is a mechanical model that corresponds to non-linear dynamics or System Dynamics. However, with assumption 4, that all interactions and events are of an average type, the direction or trajectory of the system is smooth, deterministic and perfectly predictable and is the most probable path. This implies perfect ‘determinism’, and that the ‘actors’ within the system have in fact no way of changing the course of events. This is of course not true in most situations. Models that make only the first three simplifying assumptions are known as self-organising dynamic models. By not making the assumption that all interactions are average, systems are able to spontaneously re-configure spatially/organisationally. What this means is that all system trajectories and thus configurations are possible, not just the most probable, i.e. interactions and events with different probabilities can and do arise in self-organising systems. Good or bad luck are both captured in these descriptions – chance fluctuations at particular points in time, particularly instabilities, are crucial and may lead to positive feedback that in turn may lead to qualitatively different spatial regimes. Systems with non-linear interactions between components are able to change or collectively adapt to impinging externalities or environmental conditions. This is incorporated in the mathematics as ‘noise’ – indeed, the main difference between self-organising systems and non-linear dynamic systems is this inclusion of noise. With self-organising models, instead of having only one possible trajectory and dynamic regime to investigate, all possible trajectories are investigated and these can lead to different possible attractors. This therefore changes the purpose of the model from predictive to more exploratory. According to Allen (1998; 2000; 2001), what distinguishes self-organisational processes from evolutionary processes is the internal variability in the components themselves. Whereas with
8
self-organising systems there are diverse interactions between individual components only, evolutionary complex systems assume also that there is internal diversity of the individual components – the components and sub-components co-evolve. That is, the individual components have the ability to innovate, mutate or learn through their experiences. This corresponds to the individuality of people or species. For example, with reproduction, genetic information is never passed on perfectly. Successive progeny makes sure that pure conditions are never replicated. Behaviour ‘space’ or potential behaviours are continually explored. But different behaviours may be neutral, successful or unsuccessful in the system and may be selected for or against. Through positive feedback, successful behaviours are amplified while unsuccessful ones are suppressed. The aim of this current research is to use the mathematical evolutionary model, with the least assumptions, to model the evolution of the automotive industry. 2. Methodology This section firstly describes the method behind the data collection process and secondly a description of the model. The synthesis was based on the research on the automotive industry. There were several reasons behind this decision. The first reason was because the evolution of the automotive industry is well known and publicised through both academic and trade publications. The factory layout, worker organisation and the practices employed are also well known with many pervasive in all manufacturing. The second reason was that the actual cladogram was attractive and logical and with 53 characteristics and 16 organisational forms, the data collected would prove to be rich. The third reason is that the automotive cladogram has been published several times (e.g. McCarthy et al, 1997; McCarthy et al, 2000; McCarthy & Ridgway, 2000). 2.1. Data Collection As this was an altogether new investigative approach, the methodology was of an exploratory nature rather than an established one. The aim of the research was not to apply the findings but to determine if the approach had a potential viability for application, in short, to see whether the data collected actually worked with the model and if similar findings were found to that in the study of the automotive industry. With hindsight the methodology had several minor flaws that will be highlighted in the discussion. Therefore, it must be emphasised that this investigation was more of an exploration of methodology than of application. Nonetheless, several fundamental observations can still be elicited and the potential for application demonstrated. Data was collected from a questionnaire survey. The objective of the questionnaire was to gauge the opinions of manufacturers of how the characteristics of the automotive industry (see table 1) interacted on one another. The answer options were based on a three-point scale - negative, neutral and positive. That is, the question that was asked was whether both characteristics together had either a negative, neutral or positive effect on productivity. Figure 3 is an example of a small portion of the questionnaire. Figure 3. Example of small portion of the questionnaire matrix
CHARACTERISTICS 1. Standardisation of parts
2. Assembly time standards + 2. Assembly time standards 3. Assembly line layout 0 + 3. Assembly line layout
4. Reduction of craft skills + - 4. Reduction of craft skills 5. Automation (machine paced shops) - 5. Automation (machine paced shops)
6. Pull production system 0 6. Pull production system
9
After, piloting the questionnaire, the final package sent to the companies was personalised as much as possible. The respondents were also offered the final research paper as an incentive. The questionnaire was then sent by post to 1,565 to manufacturing organisations, from a company list obtained through Yell.com, throughout most of the UK (England, Wales and Scotland). The covering letter was addressed to the managing director, manufacturing manager, operations manager, or a person of a similar status pre-determined by telephone. Of the 1,565 questionnaires, 73 were returned equating to just less than an expected 5% return rate (4.66%). 2.2. The Evolutionary Complex Systems’ Model The evolutionary complex systems’ model has been developed into a computer program called Turbo Basic® and run in the Microsoft Dos® operating system. It is based on the equations given in Allen (1984; 1985) and Allen et al (1986). There are numerous variables that can be manipulated and calibrated. There are four variables directly related to the running of the model that need to be explained. The first is the running time of the evolutionary model. This may be adjusted to accommodate reaching the final solution. Solutions are typically found between 10,000- and 50,000-time units. The second variable is the number of characteristics launched in the model. This can be controlled so that specific or experimental organisational forms may be introduced and explored. The third variable is the starting value of the characteristic. Characteristics may have a value between 0-25 units. The value may be thought of as the success or importance to the organisation. In addition, if a new characteristic is launched, the starting value also indicates the commitment of the organisation to this characteristic. An analogy can be made with an ecosystem – if an ecosystem (organisation) were defined by the constituent species (characteristics) then the value of the species would relate to their population total and hence its success and importance to the ecosystem. Most of the runs reported in the Results and Discussion section (below), launch the characteristics with a starting value of 5. Due to the interactions between characteristics during the evolutionary run, the characteristic value changes reflecting the overall importance of the characteristic to the organisation. The final variable to be highlighted is the rate at which random or specific characteristics may be introduced. Character introduction reflects innovation. This function may be turned off all together reflecting an organisation that does not innovate. To get sensible results the innovation rate should be adjusted so that a degree of organisational stability is achieved before introducing the next characteristic. Typically, character introductions are between every 300 – 3000+ time units. An organisation that launches characteristics every 300-time units, for example, would be very innovative. This function may also be programmed to launch characteristics at random time intervals. There are several other variables that may also be manipulated (e.g. the distance or locality of interaction, the rate or speed of interactions between the characteristics and the seeding of different evolutionary runs) but are more concerned with mathematical side of the model and as such are beyond the scope of this paper. The aim of the model for this paper is twofold. The first is to test the stability of 14 of the 16 organisational forms that make up the automotive industry. There are several objectives of this aim. The first is to demonstrate how the model works and the potential usefulness of this approach. The second objective is to see if the model results reflect the results of the actual cladogram. The third is to determine if there are any flaws in the methodology or design of the data collection procedure. The final objective is to relate the results of this experiment to the questions surrounding sustainability highlighted in the introduction. The second aim of this paper is to launch a particular organisational form at its most stable solution, in this case the Modern
10
Mass Producer, and then introduce characteristics either singularly or in combinations. There are two main objectives of this aim. The first is again to relate the results to the sustainability issues, particularly concerning decision-making, uncertainty and specific barriers to the introduction of new practices. The second objective is to demonstrate the exploratory capacity of the model, particularly when evolving completely new organisational structure. 3. Results Following the aims of this particular study, the first objective was to simulate characteristic combinations as detailed by McCarthy et al (1997; see also Leseure, 1998). The Ancient craft system (no characteristics) and the Standardised Craft System (only one characteristic) were not simulated, as there would be no character interaction. Therefore each of the remaining 14 organisational forms was tested for the stability of the characteristic combinations (see table 2). As can be seen from table 2, thirteen of the fourteen organisational forms had unstable characteristics (refer to table 1 for the description of each characteristic). These are very interesting results and may be due to several reasons. Table 2. Table of results: Stability of characteristics of the 16 organisational forms
Stable? Unstable Characteristics Time 1. Modern Craft Systems Yes NONE 10000 2. Neo-Craft Systems No 47, 48 10000 3. Skilled, Large Scale Producer No 47, 48 15000 4. Large Scale Producer No 4, 47, 48 10000 5. Mass Producers (Fordism) No 4, 14, 47, 48 10000 6. Pseudo-Lean Producers No 4, 7, 14, 20, 32, 47, 48, 52 15000 7. Modern Mass Producers No 4, 14, 47, 48 12000 8. European Mass Producers No 4, 14, 45, 47, 48, 52 15000 9. Intensive Mass Producers No 4, 14, 45, 47, 48, 52 15000 10. Just in Time Systems No 4, 28, 48 15000 11. Flexible Manufacturing Systems No 4, 7, 28, 31, 32, 41, 48 15000 12. Toyota Production Systems No 4, 7, 28, 31, 32, 41, 47, 48 15000 13. Lean Producers No 4, 7, 23, 28, 31, 32, 47, 48 15000 14. Agile Producers No 4, 7, 23, 28, 31, 32, 40, 41, 48 25000 15. All Characteristics Launched No 4, 7, 14, 20, 23, 28, 31, 32, 35, 40,
41, 45, 46, 47, 48, 52 35000
The first and most obvious reason could be the fault of the questionnaire design. One possibility may be due to misinterpretation of the statements and questions on the questionnaire. For example, ‘reduction of craft skills’ is a lack of a characteristic and may have been misinterpreted. It would have been better perhaps if it was stated as just ‘craft skills’. In the original automotive cladogram it can be seen that this characteristic is essential for most of the organisational forms particularly in both the main mass production and lean production evolutionary branches. In addition, there may also be a question of weighting the characteristics, as some are more important than others. For example, character state 1, standardisation of parts, is far more important than character state 11, employee innovation prizes, or character state 12, job rotation. There are also questions such as timing of introduction, i.e., some characteristics would be introduced earlier than others and some may also be precursors, for example, character state 5, automation, may be thought of as a precursor to flexible (CS 42) and agile automation (CS 43). Another possibility may be due to certain characteristics having ‘bad press’ in the media, which may influence opinions on certain interactions. A good possible example of this is the character state 47, division of labour. This, of course, is essential for all manufacturing organisations (and in most other social systems) with the exception of the first two craft systems where one
11
craftsman typically produced the end product. Similarly, there is also a question of psychological barriers, in terms of perceptions and biased opinions, to certain characteristics. For example, character state 14, suppliers selected primarily on price, is not as popular or as manufacturing-friendly as character state 23, open book policy with suppliers and the sharing of cost. However, it may be a necessity for many manufacturers and possibly for a lot of the manufacturers who responded to the questionnaire. In other words, respondents may not like the policy/practice and indicate that it is negative, but may be nonetheless essential for their operations. These are all very important questions and problems, and need to be considered in future research. Nonetheless, the evolutionary model was still viable and some very interesting results were found, particularly when investigating the introduction of new characteristics. An example of the actual process of simulation for the Modern Mass Producer (MMP) can be seen in figures 4 to 7. Each bar represents a characteristic. There are 53 bars ordered increasingly from left to right. For example, the first bar on the left, which is yellow, represents character state 1, standardisation of parts; the second bar, which is pink, represents character state 2, assembly time standards and so on. All characteristics of the MMP were launched with a starting value of 5 (see figure 4). Figure 5 shows at T=1400 (time units), a relatively short period of time, character states 4, reduction of craft skills, 14, suppliers selected on price, 47, division of labour, and 48, employees are treated as system tools, failed to make an impact from the beginning due to the accumulative interactions of all the other characteristics. The demise of these characteristics is depicted in figures 6 and 7. Figure 5 shows that the influential characteristics at the beginning of the evolutionary run are 1, standardisation of parts, 13, large volume production, 29, U-shape layout, 31, line balancing and 46, dedicated automation. Figures 6 and 7 shows that with the demise of the unstable characteristics, character states 16, socialisation training, 20, multiple subcontracting, 50, product focus, and 52, a dependence on written rules, all increase in their value to the organisation. Figure 4. Modern Mass Producer (MMP) at T=50
Figure 6. At T=5000
Figure 5. At T=1400
Figure 7. At T=12000
4 14 48 47
4 14 48 47 4 14 48 47
12
The second objective of the investigation was to identify potential obstacles when introducing new technologies or practices to existing manufacturing organisation. The Modern Mass Producer (MMP) was the experimental case study (see figures 4-7). The model was run with all the organisation’s defining characteristics until a stable solution was found. In this case, it was without character states 4, 14, 47 and 48 (figure 7). The value for each characteristic was noted and re-launched at these values. Several experimental characteristics and bundles of characteristics were then introduced. The first area investigated was practices relating to quality, that is character states 10, quality circles, 21, quality systems, and 22, quality philosophy. This area is the most related, out of all other characteristics, to issues concerning sustainability, for example, attempts at minimising waste in time/resources, and prolonging the lifetime of products. Character state 10, quality circles, was introduced singularly with a starting value of 5. The character state soon increased in importance to the organisation and trebled its value. There was a small effect on character state 13, large volume production, which decreased in value from 18 to 15, which is interesting. There were no other apparent effects on the rest of the organisation. However, the introduction of character state 21, quality systems (e.g. procedures, tools, ISO9000), again with a starting value of 5, had an effect on character state 20, multiple subcontracting, which became unstable and failed. A very interesting finding was if quality systems was introduced with little commitment (e.g. a starting value of 2), the characteristic took a much longer time period to establish itself than if introduced with a stronger commitment (e.g. with a value of 5). This is one of the potential problem areas in the introduction of sustainable technologies. A similar effect on multiple subcontracting occurred after the introduction of character state 22, quality philosophy. Obviously, through the views of the respondents, practices associated with quality standards react negatively with the practice of subcontracting out to multiple agencies. Figure 8. Quality policies & MMP T=50
Figure 10. At T=6000
Figure 9. At T=2000
Figure 11. At T=11000
10 21
22
20 13
10 21 22
20 13
10 21 22
20 13
10 21 22
20 13
13
These three characteristics were then introduced as a package that would equate to a new organisational form and another evolutionary ‘branch’ close to the MMP. When this new structure was launched (see figures 8 to 11) character state 13, large volume production, suffers slightly dropping down from an approximate value of 18 to a value of 13. In addition to this, multiple subcontracting failed completely. The combined effect of all three quality-related characteristics, however, had an overall stronger value at the stable solution. Would the effect on large volume production been anticipated beforehand with the introduction of quality policies? This gives some indication of the unpredictability, risk and uncertainty in decision-making. The simulation also demonstrates its potential for identifying problem areas and exploring possible solutions. Having identified that multiple subcontracting reacted negatively with the new practices then other policies relating to the workforce would need to be investigated. Due the last results, the procedure was then repeated for characteristics relating to the workforce policies of the MMP organisation (see figures 12-17). This included introducing characteristics 11, employee innovation prizes, 12, job rotation, 17, proactive training programmes, 24, flexible multifunctional workforce, 33, line balancing, 34, team policy, 36, team ethic, 37, job enrichment, 38, manufacturing cells, and 49, employees as system developers. These characteristics were introduced as a bundle with surprising consequences (see figures 12-17). First of all, there was a initial shock to the system, were most characteristics lost some value and became temporarily unstable (see figures 12 & 13). It took the system a relatively long time to recover (see figure 14 at T=12000), after which most characteristics stabilised including the new introductions. This perhaps reflects real situations when a package of new practices, in this case 10 new policies) challenges the rest of the organisational structure, having effects on both productivity and the internal workings and philosophy of the organisation. Figure 12. MMP & Workforce Policies, T=50
Figure 13. At T=850
Three characteristics, 20, multiple subcontracting, 32, sequential dependency of workers, 52, dependence on written rules, did not recover from the instability and disappeared from the system (see figures 14 & 15). The disappearance of all three of these characteristics, multiple subcontracting, the sequential dependency of workers and dependence on written rules, is perhaps logical as they counteract many of the new practices introduced. Another interesting finding is that when the model seemed to be reaching a stable solution (see figure 15), character state 52, dependence on written rules, suddenly and rapidly decreased in value and soon disappeared (see figures 16 & 17). This finding indicates that the interactions of technologies and practices can occur over long time periods and that some characteristics have certain thresholds that when crossed, can have critical implications for the rest of the organisation. In this instance, it appears
New character states
Unstable old characteristics that go on to recover 20 32
52
14
that the new workforce policies needed time to embed themselves in the organisation before they could influence other parts of the organisation. This demonstrates the potential of the model when looking at problems with longer time horizons. Figure 14. At T=12000
Figure 15. At T=20000
Figure 16. At T=23000
Figure 17. At T=28000
Another area of exploration concerned policies regarding the supply side of the organisation. The characteristics concerned were 8, pull procurement planning, 23, an open book policy with suppliers and a sharing of costs and profits, and character state 27, TQM sourcing where suppliers are selected on the basis of quality. When introduced singularly pull procurement planning had no effect on the rest of the organisation although itself increased in value to the organisation. When both character states 23 and 27 were introduced singularly, multiple subcontracting again failed, whilst the latter characteristics both increased in value. When all three of the characteristics were introduced together, the whole organisation was affected, with small decreases in most of the values of the characteristics. Nonetheless, the system recovered apart from multiple subcontracting which again failed. Character states 6, pull production system, and 30, preventive maintenance also had significant reductions in value (by 4 units) although were still influential in the overall organisation. These findings are a little puzzling, as the new practices/policies are not directly related to the character states that failed. However, as the model takes into account the indirect interactions of all characteristics all at once, subtle influences, as these findings indicate, are often very important in an organisational setting. This is one of the main advantages of this model - the exploration of all potential consequences that seem in some
20
32
52
32
52
52
52
15
instances logical and in others illogical, totally unrelated and surprising. This suggests that not all benefits and risks can be calculated beforehand whilst also highlighting the limited capacity of foresight in certain situations.
Figure 18. MMP & Supplier Policies, T=50
Figure 20. At T=11000
Figure 19. At T=2700
Figure 21. At T=18000
One final area explored with the MMP organisation was the introduction of all these characteristic packages, policies relating to quality, the suppliers and the workforce (15 new character states), that would create an altogether new organisational form (see figures 22-26). Figure 22. MMP with 15 new character states, T=50
Figure 23. At T=1300
As can be seen from figures 22 and 23, the introduction of the 15 new character states produced quite a shock to the system. From previous results, this was again perhaps expected, but again indicates problems when introducing new characteristic bundles. Character states that were
6
23 27
16 13
8
30 20
6 23 27
16 13
8
30 20
6
23 27 16 13
8
30
20
6
23 27 16 13
8
30
20
6
32 16 13 30 20
46 50
6
32 16 13 30 20
46 50
16
particularly affected but recovered some stability, were 6, pull production system, 13, large volume production, 16, socialisation training, 30, preventive maintenance, 46, dedicated automation, and 50, product focus. The strongest characteristics after the initial shock (see figure 23) were character states 1, standardisation of parts, 46, dedicated automation, and 52, dependence on written rules. Figures 24, 25 and 26 depict the demise of the unstable characteristics. Figure 24. T=8000
Figure 26. T=43000
Figure 25. T=15000
Figure 24 shows the first two characteristics to disappear, character states 20, multiple subcontracting, and 23, open book policy with suppliers and sharing of cost. The disappearance of multiple subcontracting is perhaps expected in the light of previous results, but the disappearance of character state 23 is surprising as this characteristic was successful when introduced both on its own and in combination with other supplier policies. There is no obvious reason why this character state failed this time, other than the subtle indirect interactions with other practices and policies discussed above. Figure 24 also shows that character state 32, sequential dependency of workers, in decline, and that character state 52, dependence on written rules, still had a healthy value. Figure 25 shows that character state 52, dependence on written rules, again fails after a relatively long time period. When comparing figures 22 and 26 it can be seen that the only original characteristics that were hardly affected were 1, 2, 3, and 5. One final observation (see figure 26) is that out of the surviving 24 character states, only 7 could be said to have healthy values in terms of the importance to the organisation (character states 1, 24, 30, 38, 46, 49, and 50). 4. Discussion There are several important findings and observations from these results in terms of the questions and further research areas posed in the introduction of this paper. The aim of this paper was to demonstrate the usefulness of this approach to problems such as those encountered when
23 32 20
52
23 32 20
52
1
38
30
24
46 50
49
17
introducing new technologies, practices and policies to an existing organisation. The dataset produced from this study is very rich (this paper has reported on only one aspect of only one of fourteen organisations studied). One important point is that the results presented in this paper are the result of one evolutionary run (reality is of course another). The model can produce an infinite number of evolutionary runs, producing different solutions and highlighting different areas of concern. Indeed, due to this, the model must not be mistaken for a predictive tool but rather an exploratory tool in which possibilities are investigated. The first issue in terms of further research, however, is that this investigation was the first of its kind and there was an expectation that there would be a few problems in design and methodology, for example the importance and possible weighting of some of the characteristics, a clearer and more understandable description of character states in the questionnaire, a chronology of character introduction in reality, and a verification period were results are validated by experts both in industry and academia. The main aim was to address some problems that may be encountered in the introduction of new and sustainable technologies, practices and policies. Of course, sustainability characteristics were not investigated per se, nonetheless, many similar problems had arisen. The main problem studied was the consequences of introducing new practices and as such the decisions behind these introductions. The first issue identified was the management’s view (and possibly the workforce’s view) of some of the characteristics. This psychological barrier (in terms or perceptions and biases), which may manifest in organisational culture, has an important bearing on the success of any new characteristic. The psychological barrier in this case, caused several of the characteristics to fail in all types of organisations (see table 2). Some of these characteristics were seen in the original automotive cladogram as essential prerequisites to most of the organisational forms. Similarly, as was highlighted with character state 21, quality systems, the initial commitment of the organisation is very important in the success of any new characteristic. Commitment is an issue not only for quality policies (e.g., Goh, 2000) but also for sustainable technologies and practices (see for example Berkel, Willems & Lafleur, 1997). The model demonstrated how new practices may emerge and how they interact with other characteristics. These interactions, successes and failures are quite often logical in terms of one characteristic replacing a similar characteristic, but are also sometimes illogical and quite surprising indicative of a high degree of unpredictability. This unpredictability highlights the limited capacity of foresight, and in some aspects, precaution. It is these scenarios that the indirect and somewhat subtle interactions influence the fates of unrelated character states. It was demonstrated that innovations behind organisational transformations, in different spheres in the organisation, for example, quality, supplier and workforce policies, can have unexpected and disastrous consequences on either production or the overall internal consistency or harmony of the organisation. The model, through simulating these processes, stable solutions, and potential consequences, both in the short- and long-term, can be an aid to management in decision-making, in terms of reducing uncertainty. Fully exploring the consequences could also have an overall impact in reducing, for example, the timescales involved in major organisational transformation. In one of the last simulations (above), character state 23, an open book policy with suppliers and a sharing of cost, unexpectedly failed when previously it succeeded. In cases like this, the model can give the modeller more of an insight into possible reasons. Different variables may be manipulated, such as the starting value (commitment), or different character combinations explored, both of which may to lead valuable answers for management. This may also be of use in understanding on a more general level why some technologies work in one organisation but not in another.
18
5. Tool for sustainability? As can be seen from the discussion, this new research using the unique evolutionary framework may provide a useful decision-making aid for manufacturers in their bid to become more sustainable. Through the identification of sustainability characteristics, such as life cycle analysis, waste minimisation and energy efficiency schemes (for other examples see table 3), and the synthesis with more traditional manufacturing characteristics, such as different production systems, factory layouts, and levels of automation (McCarthy et al, 1997), it would be possible to explore the problems of technology transfer, the potential consequences in all areas of the organisation, and new structural formations. Table 3. Potential sustainability characteristics (from van Berkel et al, 1997)
Inventory Tools Improvement Tools Prioritisation Tools Management Tools 1. Life cycle inventory 2. Abridged LCI 3. MET Matrix 4. Eco-Balance 5. Material & energy balance 6. Process flow chart
7. Ecological principles 8. Product improvement
approaches 9. Product improvement matrix 10. Pollution prevention
techniques 11. Pollution prevention strategy 12. Option inventory 13. Blueprint
14. Benchmarks 15. Total cost calculation 16. Life cycle cost calculation 17. Life cycle evaluation 18. Eco portfolio analysis 19. Product summary matrices 20. Eco opportunity 21. Option evaluation
22. Design for environment 23. Cleaner production
indicators 24. Process audit 25. Cleaner production guide 26. ISO14000 27. EU Eco-Management
systems
Future research would have to identify, analyse and weight characteristics that define non-sustainable through to pseudo-sustainable (end-of-pipe mentality) to sustainable (reduce at source) organisational forms. The next stage would be to construct the cladogram describing the evolution of the sustainable manufacturing following the guidelines highlighted in the introduction (see figure 27 for a hypothetical cladogram). At this stage, the cladogram would offer manufacturers a benchmark of past, current and best practice. Organisations are able to identify the position on the cladogram of both themselves and their competitors and use it as a guide in organisational re-engineering for sustainability. Figure 27. A hypothetical cladogram of the evolution of sustainable manufacturing (from McCarthy, 1997)
19
To develop the evolutionary model of sustainable manufacturing the research would have to establish, through sources such as questionnaires and interviews with industrial and academic experts, participating company records and archives, how the sustainability characteristics interact with traditional manufacturing characteristics such as those, for example, that make-up lean production. Through computer simulation, it would then be possible to explore the interactions between sustainability characteristics and traditional manufacturing characteristics and examine opportunities and barriers to new practices, techniques, and technologies that could increase the sustainability of the manufacturing system. In addition, it would also be possible to explore how market conditions, internal organisational structures and practices, government policies and calculations of risk might influence the adoption new, more sustainable technology. The consequence of this would be the identification of the possible problem-areas of particular decisions and the short and long terms effects. The main outcome would be to provide an evolutionary framework and to develop appropriate tools designed to consider possible options, revealing the advantages and associated costs more explicitly which will help reduce the risk and uncertainty in decision-making when implementing sustainable technologies and systems. The research could also establish the kind of innovations and new practices that may be considered over the next decade, and relate these to their possible effect on sustainability of the company, and on society. From this, and from a knowledge of the internal structures of participating companies, guidance can be given that will influence evolution in the direction of increased sustainability. Research in this new field and information and tools that it would generate would arguably be valuable for manufacturing organisations. It will facilitate a reflection of the possible innovations, new ideas, disrupting technologies, threats and opportunities that they face, and in addition will introduce into the discussion the issues surrounding sustainability. This may identify threats to their longer-term survival, and in addition lead to choices of innovations and changes that have greater efficiency and cheaper running costs in the future. The tools could arguably provide practical assistance in seeing options and in assessing their benefits and costs. Clearly, society could also benefit from an accelerated consideration of the factors affecting the sustainability manufacturing systems. This goes beyond the obvious case of the maintenance of manufacturing employment itself, but also a decrease in the burden that manufacturing places on the global and local ecosystems and environment in performing its tasks. The ideas emerging from complex systems thinking have caused considerable excitement, but not as yet a great many practical results. The work on industrial evolution and the transformation of organisations over time offers a real opportunity for making use of these ideas to promote the transition to a manufacturing sector with more sustainable and less damaging impacts on its environment. The theoretical ideas link to some of the most exciting ideas in current discussions about post modernism, the nature of knowledge, and in economic, political and organisational science. References
Allen, P. M. (1984) Self-organisation and evolution in urban systems. In: Crosby, R. (ed) Cities and Regions as Non-Linear Decision Systems. AAAS Selected Symposia, 77, Westview Press, Boulder Colorado, pp. 29-62.
Allen, P. M. (1985) Towards a new synthesis in the modelling of evolving complex systems. Environment and Planning B, 12, 65-84.
Allen, P. M. et al (1986) Towards a general dynamic model of the spatial evolution of urban and regional systems. In: Hutchinson (ed). Advances in Urban Systems Modelling. Plenum Press.
Allen, P. M. (1988) Evolution: Why the whole is greater than the sum of its parts. In: Wolff, Soeder and Drepper. (eds) Ecodynamics. Springer-Verlag, Berlin, pp.2-31.
20
Allen, P. M. (1997) Cities and regions as evolutionary complex systems. Geographical Systems, 4, 103-130.
Allen, P. M. (1998) Evolving complexity in social science. In: Altmann, G, & Koch, W, A. Systems: New Paradigms for the Human Sciences. Walter de Gruyter, Berlin and New York.
Allen, P. M. (2000) Modelling complex economic evolution. In: Schweitzer and Silverberg (eds). Evolution and Selbstorganisation in der Okonmie.
Allen P.M. (2001) Knowledge, Ignorance and the Evolution of Complex Systems. In: Foster, J, & S. Metcalfe, S. (eds) Frontiers of Evolutionary Economics: Competition, Self-Organisation and Innovation Policy. Edward Elgar, Cheltenham, UK.
Department of Environment, Transport and the Regions (1999a) A Better Quality of Life: A Strategy for Sustainable Development for the United Kingdom.
Department of Environment, Transport and the Regions (1999b) Quality of Life Counts: Indicators for a Strategy for Sustainable Development for the United Kingdom - A Baseline Assessment.
Goh, P. (2000) The Implementation of Total Quality Management in Small to Medium Sized Enterprises. Unpublished Ph.D. Thesis, University of Sheffield.
Leseure, M. (1998) Manufacturing Cladistics: Using Phylogenetic Classifications to Understand and Manage the Complexification of Manufacturing Systems. Unpublished Ph.D. Thesis, University of Sheffield.
Leseure, M. (2000) Manufacturing strategies in the hand tool industry. International Journal of Operations and Production Management, 20(12), 1475-1487.
Maddison, W. P, & Maddison, D. R. (1992) MacClade Version 3: Analysis of Phylogeny and Character Evolution. Sinauer Associates Inc., Massachusetts, USA.
McCarthy, I. P. (1995a) Manufacturing classification: Lessons from organizational systematics and biological taxonomy. Integrated Manufacturing Systems, 6(6), 37-48.
McCarthy, I. P. (1995b) The Development of a Manufacturing Classification using Concepts from Organisational Systematics and Biological Taxonomy. Unpublished Ph.D. Thesis, University of Sheffield.
McCarthy, I. P, Leseure, M., Ridgway, K. & Fieller, N. (1997) Building a manufacturing cladogram. International Journal of Technology Management, 13(3), 269-286.
McCarthy, I. P, & Ridgway, K. (2000) Cladistics: A taxonomy for manufacturing organizations. Integrated Manufacturing Systems, 11(1), 16-29.
McCarthy, I. P, Leseure, M., Ridgway, K. & Fieller, N. (2000) Organisational diversity, evolution and cladistic classifications. The International Journal of Management Science, 28, 77-95.
Prigogine, I. (1973) Irreversibility as a symmetry-breaking process. Nature, 246, 67-71. Ridley, M. (1993) Evolution. Blackwell Scientific Publications. Ross, H. (1974) Biological Systematics. Addison-Wesley, Reading, MA. Sheffield First Partnership (1999a) Review of Assisted Areas of Great Britain. Sheffield: A
Positive Response - The Case for Development Area Status for Sheffield and South Yorkshire. Sheffield First Partnership, Sheffield.
Sheffield First Partnership (1999b) Developing Sheffield’s Economy for the 21st Century. Sheffield First Partnership, Sheffield.
Suarez, F, Cusumano, M, & Fine, C. (1995) An empirical study of flexibility in manufacturing. Sloan Management Review, Fall, pp. 25-32.
United Nations Commission Environment on and Development (1992) Agenda 21. UN, New York, 1992.
van Berkel, R, Willems, E, & Lafleur M. (1997) Development of an industrial ecology toolbox for the introduction of industrial ecology in enterprises-I. Journal of Cleaner Production,5(1-2),11-25.
Womack, J, P, Jones, D, T, & Roos, D. (1990) The Machine that Changed the World. MacMillan Publishing, New York.
World Commission on Environment and Development (1987) Our Common Future. Oxford University Press, Oxford, England.
