A case study in the aviation industry THE EFFECTS OF SMART MANUFACTURING ON JOB AUTONOMY:

(1)

THE EFFECTS OF SMART MANUFACTURING

ON JOB AUTONOMY:

A case study in the aviation industry

Master thesis, MSc Technology and Operations Management University of Groningen, Faculty of Business and Economics

January, 2017

THOMAS JAN VAN DEN BERG Student number: 2022109

Grachtstraat 56 9717 HL Groningen t.j.van.den.berg@student.rug.nl

Word count: 12.582

Supervisor / University of Groningen (RUG) Dr. J.A.C. Bokhorst

Co-assessor / University of Groningen (RUG) N. Ziengs, MSc.

(2)

(3)

3

Abstract

The fourth industrial revolution is on the verge breaking out and it presents itself in the form of smart manufacturing. Academic consensus about what smart manufacturing exactly comprises has however not yet been reached. This paper adds to the existing body of literature of SM by providing a clear definition of SM, based on a systematic literature review. Furthermore, changes in manufacturing practices affect work design and therefore in turn influence job autonomy. A single case study was conducted to formulate expectations about the possible effects of smart manufacturing on job autonomy. The study was conducted at an aviation parts manufacturer in the Netherlands which is currently in the first implementation phase of smart manufacturing. It was found that smart manufacturing, through its three stages of digitization, intelligence and connectivity, will probably hollow out job autonomy. Work-methods and work scheduling autonomy will most likely be eliminated and decision making autonomy will either increase or remain at equal level. Future research is encouraged to verify and research the strength of the suggested relationships.

(4)

4

Preface

Writing a thesis is an intensive and long, but rewarding process. It is also a process that I have not undertaken on my own. Supervisors and fellow students provided constructive criticism and helped me to improve my thesis step by step, to the current paper you are reading now. Furthermore, I received a lot of support from friends and family who were always interested, supported me through encouraging words and just being there, and they let me vent if things did not go as smooth as I anticipated. My colleagues at Fokker were very hospitable. Thank you for your companionship involvement, but most of all, all the laughs we have had together. To all of you: thank you very much!

(7)

7

I. Introduction

Today, we stand on the brink of the fourth industrial revolution. No more than two and a half centuries ago, the first revolution started by using steam as energy source for machinery. Electricity powered the second revolution and IT increased production in the third revolution. At this moment in history, we have reached a new summit of manufacturing as the internet has started connecting not only people, but also machines and factories: the fourth industrial revolution (Kagermann, Wahlster & Helbig, 2013). The production practices associated with this fourth industrial revolution are regarded to as ‘Smart manufacturing’ (Kang, Lee, Choi, Kim, Park, Son, Kim & Noh, 2016).

Smart manufacturing (SM) encompasses among other things increased levels of integration and automation. It is suspected that this will have a large impact on the current way of manufacturing (Kagermann, Wahlster & Helbig, 2013). The imminent changes brought forward by these new opportunities will most likely change the role of employees. These roles can be described through work design characteristics: ‘the attributes of the task, job, and social and

organizational environment’ (Humphrey, Nahrgang & Morgeson, 2007). One of these

characteristics, job autonomy, ‘the degree to which the job provides substantial freedom,

independence, and discretion to the individual in scheduling the work and in determining the procedures to be used in carrying it out’ (Hackman & Oldham, 1976, p. 258), is especially likely

to be influenced. According to Dworschak and Zaiser (2014) the technological advancements could either “increasingly eliminate the need for autonomous human decision making” or function as supporting tools while the employees “would need – and could acquire or develop –

process-specific knowledge in order to make decisions or intervene with malfunctions” (p. 348).

(8)

8

currently obscure. Thus, research is necessary to provide the desired clarity about this relationship and underlying mechanisms.

While SM is the most commonly used term in relation to this new, internet-based, revolution (Kang, et al., 2016), there is no clear consensus about the construct1. For example, other used terms include but are not limited to industry 4.0, industrial internet, smart industry and factory of the future (e.g. Kagermann, Wahlster & Helbig, 2013; Merfeld, 2014). Furthermore, a decent body of academic literature and research on SM has not yet been developed2. More pressing even, SM currently lacks a clear, comprehensive definition. That does not mean that preliminary definitions have not been provided. For example, SM has been defined as ‘A set of advanced

sensing instrumentation, monitoring controls, and process optimization technologies and practices that merge information and communication technologies with the manufacturing environment for the real-time management of energy, productivity, and costs across factories and companies’ (Malik, 2016, p. 10). The current definitions are however not exhaustive in nature.

The absence of a clear, academic definition has however not withheld firms to start reporting results based on their take on SM. Koning and Hartman (2016) report for example performance increases and production waste decreases of up to 55% and 49% respectively at firms which have implemented SM. Without a meaningful way to compare these numbers based on SM fundamentals or definition, these results are currently of little to no value though.

This paper will therefore first of all provide a state-of-the-art overview on the multifaceted construct SM to help provide the needed clarity in this field. Using a systematic literature review relevant articles were selected to create a solid foundation, as this is a prerequisite since “a

researcher cannot perform significant research without first understanding the literature in the field” (Boote & Beile, 2005, p. 3). This resulted in a framework and a definition of SM, which

provides construct clarity and facilitates therefore a meaningful academic discussion in future papers. Secondly, based on earlier comparable developments in manufacturing industries and their effects on job autonomy, an expectation about the effects of SM on job autonomy will be provided. The combination of these two inquiries will form the theoretical background of this paper. Thirdly, the main research question will be answered: How does smart manufacturing

influence job autonomy? To answer this question a single case study was conducted at Fokker

1_{Constructs are defined in this paper in accordance to Bacharach (1989, p. 498) as ‘approximated units (..) which by}

their very nature cannot be observed directly’.

(9)

9

Aerostructures in Hoogeveen (Fokker). Fokker, a manufacturer of aircraft parts, is currently in the first phase of implementing SM in its production processes. It provides therefore an exquisite chance and an ideal environment to research the possible effects of SM implementation on job autonomy. It also provides the opportunity to investigate possible ways to deal with these aforementioned effects.

(10)

10

II. Theory

Smart manufacturing

The fundamental idea of SM originated in 1991. Weiser (1991) introduced the notion of ‘ubiquitous computing’, integrating computers with the world, including production. This conception is generally seen as the start of what is today described as ‘smart manufacturing’. Research has been relatively slow in the years to follow. However, around 2010, governments started to get involved. Due to national interests (e.g. international competitiveness), three governments promoted research and investments in relation to manufacturing. In Germany3, the US4 and South-Korea5, this has resulted in the advancement of SM and related research (Kang, et al., 2016). More recently China6 also presented a similar strategy and around the globe at least 14 other countries currently deploy similar initiatives (Jirsak, Martinez, Lorenc & Jancik, 2016). Governmental backing is however not the only factor that has been driving the development of SM. A literature review has been conducted7 and resulted in several factors that can be grouped into five main categories. The findings are presented in this subchapter in accordance to these categories: first of all the stimulators of SM will be presented, followed by SM’s core technologies. Thirdly the prerequisites for SM to flourish will be covered, as well as the consequences that will follow directly. Remaining issues that were not covered by the four aforementioned categories will be dealt with before arriving at the last part of this subchapter, the definition of SM.

Stimulators

Besides governmental stimulation there are three other sorts of stimulators: global drivers, increasing customer demands and intensifying global competition. First of all, global drivers are factors that are likely to change the world significantly in the coming decades. Siemieniuch, Sinclair and Henshaw (2015) analyzed several reports and distilled eight of these so called global drivers: population demographics, food security, energy security, resource depletion, emissions and global climate, community security and safety, transportation, globalization of economic and social activity. This is in line with observations of Kagermann, Wahlster and Helbig (2013) and

3_{Industrie 4.0 (as part of the larger ‘High-Tech Strategy 2020 Action Plan’)} 4_{Advanced Manufacturing Partnership (AMP & AMP 2.0)}

5_{Manufacturing Industry Innovation 3.0} 6

Made in China 2025

(11)

11

Lasi, Fettke, Kemper, Feld and Hoffmann (2014). Siemieniuch, Sinclair and Henshaw (2015), based on their analysis, conclude that manufacturing needs to become more sustainable because the demand transcends supply. In order to achieve more sustainable manufacturing they refer to, among other subjects, some of the technologies associated with SM (e.g. Cyber-Physical Systems). Secondly, increasing customer demands in terms of higher volumes, variety of product portfolio, delivery time, individualization on demand and higher quality (Dumitrache, Caramihai & Stanescu, 2013; Heck & Rogers, 2014; Jirsak, Martinez, Lorenc & Jancik, 2016; Khan & Nasser, 2016; Lasi, Fettke, Kemper, Feld & Hoffmann, 2014; Prause & Weigand, 2016; Yao & Lin, 2016) play also a significant role in changing manufacturing practices. For manufacturers to meet these customer demands, they need more efficient ways of production. This is most likely accomplished by integrating IT and cutting edge technology with the production process (Khan & Nasser, 2016; Kolberg, Knobloch & Zühlke, 2016). Thirdly, the intensifying global competition is the last factor stimulating manufacturers to adopt new ways of manufacturing. Issues such as underutilization of means of production, re-shoring, increasing competition on a global level, shorter time-to-market, flexibility in product development, decentralization, pressure to adopt sustainable practices, new emerging technologies, changes in value chain and labor shortage (Brennan et al., 2015; Heck & Rogers, 2014; Jirsak, Martinez, Lorenc & Jancik, 2016; Kagermann, Wahlster & Helbig, 2013; Kagermann, 2015; Lasi, Fettke, Kemper, Feld & Hoffmann, 2014; Li, 2016; Prause & Weigand, 2016; Wan, Cai & Zhou, 2016) force companies to rethink their production methods and strategies.

Thus, a changing world, customers and competitors, induce companies, backed by governments, to invest in and adopt new ways of manufacturing. For these new ways multiple new technologies are (being) developed. An overview is given in the next paragraph.

Core technologies

(12)

12

Krapez, 2016; Wan, Cai & Zhou, 2015; Wang, Wan, Li & Zhang, 2016; Yao, Jin & Zhang, 2015). What lacks however, is a clear consensus about which technologies truly make manufacturing smart. To reach smart manufacturing as it is envisioned by for example Davis and colleagues (2012), “information technology (e.g., data management and modeling) must be

pervasive and integrated throughout the multiple layers of operation and decision-making (e.g., sensors and actuators, operators running the process, supply chain, etc.), and be extended across multiple process units within a factory, as well as across an entire factory and supply chain”

(Bryner, 2012, p. 8).

Two technologies, IoT and CPS, are frequently mentioned and are essential in realizing the aforementioned vision of SM (Bangemann, Riedl, Thron & Diedrich, 2016; Kang et al., 2016; Lee & Lee, 2015; Monostori, et al., 2016). In a comprehensive literature study of three smart manufacturing initiatives in Germany, the US and Korea, Kang and colleagues (2016) found six other recurring technologies, of which 3 are essential, just as IoT and CPS, in the way that they ‘affect each other when applied, thus interoperability is considered more important than any

other technologies’ (Kang, et al., 2016, p. 121). These three technologies are cloud

manufacturing, big data analytics and smart sensors. Altogether, these five technologies make up the core of smart manufacturing and can be defined as follows:

1. Cyber-physical systems. Several definitions circulate (e.g. Kang et al., 2016; Monostori, et al., 2016) but they share the same essence, characterizing CPSs ‘by their

capabilities of accessing physical information intelligent processing of the information with a specific purpose related to the industrial automation domain while applying IT and sharing these information and capabilities within the Internet technologies exploiting environment’ (Bangemann, et al., 2016, p. 953). To put it

bluntly, a CPS is a production machine that has the ability to share and receive all kinds of information through the internet.

2. Cloud manufacturing: ‘A customer-centric manufacturing model that exploits

on-demand access to a shared collection of diversified and distributed manufacturing resources to form temporary and reconfigurable production lines that enhance efficiency, reduce product lifecycle costs, and allow for optimal resource loading in response to variable-demand customer generated tasking’ (Wu, Greer, Rosen &

(13)

13

process flexibility) but is mainly concerned with the end-consumer and views the supply chain from that perspective while SM focuses on the whole supply chain (optimization).

3. Big data analytics. Big data can be described as ‘a data set that is inappropriate to be

used by traditional data process methods due to their wide range, complex structure, and size’ (Kang, et al. 2016, p. 119). In regard to SM analysis of this data will

‘optimize production quality, save energy, and improve equipment service’ (BCG, 2015, p. 5). In recent years there have been major developments in this field, although decision support analytics still need to be developed for a fully functioning SM factory (Lee, Kao, & Yang, 2014).

4. Internet-of-Things (IoT): ‘IoT means networks of electricity, software, sensors,

network connectivity, and embedded ‘things’ or physical objects. It collects or exchanges data’ (Kang, et al. 2016, p. 120). Basically it means that the IoT is the

infrastructure to share information between all sorts of things. In an industrial environment this helps to enable to ‘constantly track performance in real time,

anywhere, anytime’ and ‘enhance situational awareness and avoid information delay and distortion’ (Lee & Lee, 2015, p. 433-434)

5. Smart sensors. The sensors (and underlying and connectivity technologies) can differ, but these are essential for SM since ‘the sensor is the most basic technology for

collecting and controlling data in real time’ (Kang, et al., 2016, p. 120). Thus, the

sensors need to be technological on par with the other machines to be able to not only monitor but also distribute the gathered information. One main sensor technology is Radio-Frequency Identification (RFID) and recent cost droppings have set off multiple efforts in all kinds of industries, including aerospace (Luo, Fang & Huang, 2015). However promising, problems still exist with for example damage due to high temperatures making RFID technology for the time being unfit for production processes which expose products to high temperatures.

(14)

14

is different than for example Computer Integrated Manufacturing (CIM: ‘the manufacturing

approach of using computers to control the entire production process’ (Khan & Nasser, 2016,

p.10)). The goal of CIM was a ‘fully automated production process from procurement and

production to distribution without necessitating human interaction’ (Prause & Weigand, 2016, p.

104) while with SM humans still play a significant role (Zuehlke, 2010). Lastly, these five technologies together should be able to tackle the three challenges of (IT supported) manufacturing as formulated by Alexopoulos, Makris, Xanthakis, Sipsas and Chryssolouris (2015): capturing real-time field data, real-time processing and management of data and role changes of users in relation to information distribution. Therefore, there is no need to include other technologies.

These five technologies all heavily rely on each other but they serve three separate, consequential goals: process information, production knowledge and supply chain connectivity. The first goal, process information, concerns extracting data from the production processes through technologies such as sensors and CPSs (without the IoT component fully activated). This can be described as the digitization phase of SM since all data needs to be available for analysis, therefore electronically. The second goal is to use and combine the gathered data in order to propose adjustments of internal processes of a factory or plant to effectuate the most efficient solutions. Thus, transforming the raw information into knowledge through big data analysis. This can be labeled as the intelligence phase. The third goal is connecting the whole supply chain using IoT and cloud manufacturing making the production, from order to delivery as efficient as possible, called the connectivity phase. SM implementation can thus be divided in these three, sequential phases.

Critical factors

Besides the technological requirements there are two other prerequisites of SM that need addressing: process digitization and cybersecurity. The first factor, process digitization means that before SM can be fully implemented at a manufacturing plant, the production processes need to be fully digitized. Therefore, the first step of SM is digitization of every part in the process (Sommer, 2015). Digitization, “the process of converting continuous, analog information into

discrete, digital and machine-readable form” (De Mauro, Greco & Grimaldi, 2015, p. 98), will

(15)

15

are used intertwiningly, some paperless factories replace their paper documents simply with electronic versions of those documents (Djassemi & Sena, 2006). This does however not mean that a machine can comprehend the information that is on those documents. Therefore, the term process digitization is used in this paper because it includes paperless procedures as well as machine comprehensibility. To avoid misunderstandings, digitization is in this paper not meant as the socio-economic change as a result of for example the increase of digital equipment (Evangelista, Guerrieri & Meliciani, 2014). For further reading about this sort of digitization in relation to smart manufacturing see for example Hirsch-Kreinsen (2016) and Evangelista, Guerrieri and Meliciani (2014). While the importance of process digitization is irrefutable, SME’s lack interest in this development and ignore the chances it provides (Sommer, 2015). One of the possible reasons for this could be that the utilization of digitization ‘requires not only

comprehensive knowledge about the system in consideration, but technology specific knowledge as well’ (Monostori, et al., 2016, p. 632). In other words, the threshold for implementing and

actively using digitization, might be too high for smaller companies.

The second critical factor is cybersecurity. In a general sense ‘security describes the

protection of a system from impermissible external influences’ and ‘applies both to physical influences (..) and to impermissible influencing of an IT system via its communications interfaces’

(DIN/DKE, 2016, p. 23). The IT-dependent nature of SM makes cybersecurity a primary concern regarding implementation of SM systems. The need for increased cybersecurity in relation to smart manufacturing is widely recognized (e.g. Annunziata & Biller, 2015; BCG, 2015; Davis, et al., 2012; DIN/DKE, 2016; Hermann, Pentek & Otto, 2016; Kagermann, Wahlster & Helbig, 2013; Pasqualetti, Dörfler & Bullo, 2013; Zuehlke, 2010). Before complete production lines and even whole supply chains can connect in a relative safe manner via the internet, the risk of cybercrime needs to be minimalized. The more sensitive the manufacturing process becomes (e.g. the production of military equipment), the more attention this issue demands. To demand an entirely secure solution is unrealistic, but ‘it is possible to create security architectures capable of

(16)

16

data protection but also about protecting control over the machines since they also are linked to the internet. Failing to do so can have disastrous consequences8.

Consequences

Given the situation that all (technological) prerequisites are met, what are the potential consequences of SM? First of all, note that the fourth industrial revolution, propelled by SM, is the first industrial revolution that is predicted instead of observed which means that the development is subject to initiatives of involved actors (Hermann, Pentek & Otto, p. 2016). It is expected that it would take about 20 years to fully come to fruition (BCG, 2015). Secondly, there are three distinct types of consequences. The first type is production output, the second type is work design. The third type, societal change, concerns changes in society as a result of the fourth industrial revolution. It is however not within the scope of this research to study this type of consequence since this concerns a completely different field of research. Concerning the first type, production output, it is expected that, when SM is implemented, output changes in multiple ways, such as: customization, flexibility, responsiveness, energy efficiency and environmental effectiveness (BCG, 2015; Davis, et al., 2012; Roland Berger Strategy Consultants, 2014). Furthermore, productivity gains up to 8% are expected (BCG, 2015). However, it will not be cheap. To illustrate, Europe needs to invest €1350 billion over 15 years when it wants to achieve a leading role in global manufacturing (Roland Berger Strategy Consultants, 2014). Financial gains are also expected to be significant though. For example, General Electric expects a total of $250 billion in value creation over the next 15 years as a result of current SM developments (Merfeld, 2014). Moreover, a change towards preventive maintenance is already bearing fruit (Merfeld, 2014) and is expected to lead to industrial assets with no unplanned downtime at all (Annunziatia & Biller, 2015). It is expected that technology will influence and shape other business processes and scopes more and more (Agarwal & Brem, 2015). Lastly, products can and will become increasingly customized due to technological advancements and customer and designer demand, which is a major paradigm shift in contrast to the standardization practices of automation (Chu, et al., 2016; Yao & Lin, 2016).

The second consequence type, work design is also very likely to change (Dworschak & Zaiser, 2014). In this regard two scenarios were developed (Windelband, et al., 2013). The first scenario is called the ‘automation scenario’, in which CPSs make decisions and guide the

(17)

17

employees in their jobs. This scenario ‘has a powerful deskilling impact on the employees’ (Dworschak & Zaiser, 2014, p. 348). The second scenario is called the ‘tool scenario’. In this scenario the employees still make the decisions, albeit supported by information and machine assistance. Which scenario is the most likely depends on the organization, since a company can prefer on or the other based on the market situation or how their production process is organized. A more comprehensive review of possible outcomes is provided by Hirsch-Kreinsen (2016). The general consensus seems to be that highly repetitive tasks will be replaced by automation but for example, BCG (2015) argues that total labor demand will rise due to higher demand for other jobs (e.g. mechanical engineering). A more in-depth look at this subject is presented in the subchapter Expected relation.

Miscellaneous issues

Before defining SM, four issues remain to be dealt with. The first issue is that, to the best of my knowledge, a fully functioning ‘smart factory’ does not yet exist. Multiple initiatives are being undertaken to implement specific parts of SM9, a comprehensive take on the smart factory/supply chain does not yet exist. SM is therefore currently still a hypothetical construct, or at best a collection of practices and ideas which have not been proven to work effectively together. A larger problem may be that the technologies to be used have not yet been developed or have matured enough to be used in production environments (Leitao, Colombo & Karnouskos, 2016). Even when the technology and equipment would be ready for industrial use, the second issues arises: manufacturing equipment usually has long depreciation periods. That means that manufacturers will not change the production machines within relative short timeframes. Therefore, it will take a significant amount of time before SM becomes mainstream and a feasible, empirical researchable subject. The third issue amplifies this slow timetable. Productivity has always lagged behind the technological developments in the other industrial revolutions (Schuh, Potente, Varandani, Hausberg & Fraenken, 2014). It can therefore be a while, before the real fruits of this fourth industrial revolution will be ripe for the picking. Even after implementation of the fully developed new production processes. The fourth issue is that there are major differences in development and research between large enterprises and SME’s (Davis et al., 2012; Radziwon et al., 2014). Large enterprises have economies of scale and larger budgets

9

(18)

18

which makes investing in new, cutting-edge technology to become fully SM-compliant much more achievable. Therefore, the initial number of SM manufacturers is likely to be small but it can also lead to differences in SM-adoption which makes unilateral research in this field harder to conduct. Furthermore, large enterprises can outcompete SME’s rather easily when the outputs start to differ significantly as a result of the implementation of SM. How this competition will take shape is however more a political problem than a scientific one.

Figure 2.1 – Framework of Smart Manufacturing

Smart manufacturing Smart manufacturing Stimulators Governmental initiatives Customer demands Global drivers Global competition Stimulators Governmental initiatives Customer demands Global drivers Global competition Core technologies Cyber-Physical Systems Internet-of-Things Cloud manufacturing Big data analysis Smart sensors

Core technologies Cyber-Physical Systems Internet-of-Things Cloud manufacturing Big data analysis Smart sensors Prerequisites Technological advancement Process digitization (Cyber) security Prerequisites Technological advancement Process digitization (Cyber) security Consequences Production output Work design Societal change Consequences Production output Work design Societal change Definition

Building on all of the aforementioned information, formulating a definition of SM is still no sinecure. However, combining all the collected information, a framework of SM can be build (figure 2.1). This helps in indicating the essential factors of SM. What basically constitutes SM, are the technologies and the intended consequences, in other words the means and goals. The other factors, stimulators (e.g. governmental support) and critical factors (e.g. technological advancement), provide a context for the (future) development of SM but do not describe what SM is. Furthermore, these factors could also be applied to other constructs while the technologies and (intended) consequences are intrinsically linked to SM.

(19)

19

further sculpt a clear image of what comprises SM. As a result, Smart Manufacturing is defined as:

A flexible and transparent way of manufacturing, fundamentally based on IoT, CPS, big data analytics, smart sensors and cloud manufacturing technologies, that integrates a whole supply

chain in order to produce more efficiently, sustainable and customizable to satisfy growing customer demands through the real-time management of information.

Job autonomy

During the second industrial revolution the ‘scientific management’ movement came into existence, mostly known for the ideas of Taylor (1911) and Gilbreth (1911). In short, they focused on production processes and proposed standardized work and strict managerial control. Weber (1947) and Fayol (1949) focused on the organizational consequences of this approach and introduced for example the principle of ‘bureaucracy’ (Weber, 1947). In the 1920’s another, contrary trend arose: the human relations approach. This approach focused on (intrinsically) motivating employees as opposed to the focus on control of the scientific management approach. The Hawthorne studies are still regarded to as exemplary for this movement. Later on more focus came on job design (e.g. Davis & Canter, 1955). One of the more tangible products that resulted of this human relations approach is the still widely recognized Motivating Potential Score (MPS) of Hackman and Oldham (1976).

Definition

Hackman and Oldham (1976) identified five factors which could improve motivation of employees: autonomy, skill variety, task identity, task significance and feedback from the job. These factors are the input variables for the MPS, which enables the calculation of ‘the overall

motivating potential of a job’ (p. 258). Humphrey, Nahrgang and Morgeson (2007) provide in

their more recent meta-analytic summary a broader context. They classify the abovementioned five factors (and others such as job complexity and specialization) as motivational characteristics. Together with social characteristics (e.g. social support) and work context characteristics (e.g. ergonomics) they make up the work design characteristics. These characteristics are the main predictors of how employees experience their job.

In this research, as previously mentioned, the focus will be on the effect of SM on one specific factor: job autonomy. Job autonomy is commonly defined as ‘The degree to which the

(20)

20

the work and in determining the procedures to be used in carrying it out’ (Hackman & Oldham,

1976, p. 258). However, more recently scholars have suggested that this unidimensional construct can be split in three different underlying constructs: work scheduling autonomy, work methods autonomy (Jackson, Wall, Martin & Davids, 1993) and decision-making autonomy (Karasek, Brisson, Kawakami, Houtman, Bongers & Amick, 1998). Work scheduling autonomy can be defined as ‘the freedom to control the scheduling and timing of work’, work methods autonomy as ‘the freedom to control which methods and procedures are utilized’ and decision-making autonomy as ‘the freedom to make decisions at work’ (Humphrey, Nahrgang & Morgeson, 2007, p. 1336). These different aspects of job autonomy could sort different relations with related constructs. Another possibility is a strength difference in those relations between the three job autonomy aspects. In either case, splitting job autonomy provides more insight in the mechanisms concerned.

Job autonomy focus justification

(21)

21

more insight in the work of the employees. This could limit employee initiative because this is a form of (managerial) control (Rantakari, 2012). On the other hand, delegating decision authority about a course of action to employees, one of possible outcomes discussed earlier, increases employee initiative since it reduces ‘the likelihood of managerial interference that would make

the employee effort redundant’ (Rantakari, 2012, p. 172). Furthermore, improved feedback

procedures could encourage people to think out of the box to improve the production process since job autonomy is positively associated with creative work involvement (Volmer, Spurk & Niessen, 2012). This holds especially in new projects where process efficiency is still relatively low. Thus, employee initiative is important and can be encouraged through delegating authority and improving feedback procedures, which enhances decision-making autonomy. The second reason for the focus on job autonomy is feasibility. The available time is limited for this specific research. Therefore, it makes sense to focus on a well-developed construct such as job autonomy, which “is perhaps the most widely studied job characteristic” (Bayo-Moriones, Bello-Pintado & Merino-Díaz-de-Cerio, 2010, p. 65). Furthermore, this study fits within a larger PhD project on SM and can therefore be used as input for other studies in this project.

Relations with other constructs

Earlier research has found many relations between job autonomy and other constructs. Not all constructs are however relevant for this research (e.g. benevolent leadership, Wang & Cheng, 2010) or not strongly correlated enough to be interesting (e.g. overload, Humphrey, Nahrgang & Morgeson, 2007). By selecting the high impact (ρ > 0,30) and significant (p > 0,05) relations related to the retention/availability of the workforce in the model of Humphrey, Nahrgang & Morgeson (2007), three relevant constructs can be distinguished: burnout, organizational commitment and job satisfaction. The research concerning the relationship between job autonomy and burnout has however resulted in some contradicting results. Kim and Stoner (2008) for example did not find significant results for this relationship. However, they could associate burnout positively and directly to turnover intention. Remarkably, meta-analysis shows that turnover intention has no meaningful association with job autonomy (Humphrey, Nahrgang & Morgeson, 2007). While job autonomy may have no direct relationship with turnover intentions, it does however, as stated before, positively correlate with organizational commitment. In turn, organizational commitment, ‘a psychological state that binds the individual

(22)

22

turnover (Mathieu & Zajac, 1990). This means that decreasing job autonomy can still have the effect of increased turnover intentions in an organization through its effect on organizational commitment. Furthermore, a meta-analysis shows that job satisfaction, just as organizational commitment and burnout, negatively correlates with turnover intentions (Tett & Meyer, 1993). The same logic can therefore be applied: lower job autonomy can also lead to increased turnover via lower job satisfaction. This is especially relevant since decision-making autonomy correlates the strongest with job satisfaction (r = 0,50) of the underlying job autonomy constructs (Humphrey, Nahrgang & Morgeson, 2007). One of the most effective ways to decrease turnover intention for organizations is to focus on supervisor support. Both on individual and aggregate level, supervisor support is (very) positively associated with job satisfaction (Griffin, Patterson & West, 2001), which thus in turn decreases turnover intention. A more direct, but also very influential relation is that of managerial leadership and the retention of key people (Bontis & Fitz-enz, 2002). This would suggest that the possible negative effects of decreasing job autonomy can be mitigated through support and leadership. In summary, a decrease in job autonomy is very likely to have negative impact on the workforce. It can be definitive (employee turnover) or temporal (burnouts of employees). In both cases it results in problems for a company, especially a company which relies heavily on craftsmanship in its production. (Specialized) craftsmanship is hard to replace and therefore a focus on employee retention seems a logical priority for such an organization.

Job autonomy can not only be viewed as an antecedent construct, but also as a consequent construct. The biggest influencer of job autonomy, relatively easy changed by an organization, is job enrichment. Job enrichment is positively associated with job autonomy on both individual as well as aggregate level (Griffin, Patterson & West, 2001). Job enrichment practices are for example job rotation, multiskilling of employees and maintenance responsibilities. As stated, an organization can relatively easily change these policies and therefore influence job autonomy. Changing or entering job enrichment policies can therefore also be very well used to counter other, negative antecedents of job autonomy.

Expected relation

(23)

23

formulate a well-founded expectation, the effects of earlier developments on job autonomy in manufacturing industries are examined here. The focus is on two developments that changed work designs: lean production/automation and advanced manufacturing techniques (AMTs). Lean production is widely used and considered state of the art when it comes to production systems (Kolberg, Knobloch & Zühlke, 2016; Kolberg & Zühlke, 2015). However, Lean Production ‘has reached its limits to meet future market demands’ (Kolberg, Knobloch & Zühlke, 2016, p. 1). Therefore, since the 1990s efforts have been made to integrate automation practices with lean production, creating ‘lean automation’ (Kolberg, Knobloch & Zühlke, 2016). Recently it was stated that lean automation can be seamlessly integrated with SM (Sanders, Elangeswaran & Wulfsberg, 2016). Therefore, I assume that meaningful insights can be extracted from the implementation of lean for the effects of SM. AMTs, an extensive set of technologies to increase the efficiency and flexibility in a manufacturing environment (Wu, 2012), have been on the rise since three decades due to increasing computer power. It is not within in the scope of this study to elaborate broadly on this subject, but an overview of current AMT’s and examples from industries where they are used is provided by Khan and Nasser (2016). For this study it suffices to note that AMTs currently focus on small-batch, customized production goods and data-driven processes (Khan & Nasser, 2016). This is achieved by making use of multiple technologies and process digitization. Both practices are related to SM and therefore it seems reasonable to incorporate the effects of AMTs in this research.

Longitudinal research suggests that lean practices have a negative impact on job autonomy (Parker, 2003). This can mainly be contributed to the standardization of workflow which limits all the three job autonomy constructs (Parker, 2003; De Treville & Antonakis, 2006). However, when employees are actively involved with developing operational procedures and believe that enforced constraints are beneficial is for the process, autonomy can increase (De Treville, Antonakis & Edelson, 2005). This holds especially for decision making autonomy.

(24)

24

result in standardization of operational processes due to interdependencies of processes, machines and employees. These interdependencies in turn limit job autonomy (Turnbull, 1988). This is similar to the aforementioned automation scenario (Dworschak & Zaiser, 2014).

Thus, lean practices seem to decrease all forms of job autonomy but can enhance decision making autonomy by involving employees in development of procedures. AMTs improve decision making by empowering employees to solve problems on their own. Furthermore, it seems that AMTs also limit the work methods and scheduling autonomy. An overview of all the relations in this chapter can be found in figure 2.2.

The distinction between the different phases of SM and the sub constructs of job autonomy, enables a more precise materialization of the general research question ‘How does smart manufacturing influence job autonomy?’ Including these more specified constructs and phases, the research question becomes: How do the three phases of SM influence the work-methods-, work scheduling- and decision making autonomy of employees?

Figure 2.2 - Relational framework

-Job autonomy

Job autonomy

Work scheduling autonomy

Work methods autonomy

(25)

25

III. Methodology

In the effort to research the influence of SM on job autonomy, both the initial ‘How?’ and subsequent ‘Why’ questions were asked. It therefore seemed more than obvious to conduct a case study. For a case study is not only very well suited to answer the ‘How?’ aspect of a research question (Yin, 2014), it can also provide answers to questions such as ‘Why?’ ‘with a relatively

full understanding of the nature and complexity of the complete phenomenon’ (Voss, Frohlich &

Tsikriktis, 2002, p. 164). Moreover, a case study provides a better insight in the field of the subject and can help understand relations (Eisenhardt, 1989). No other study design could offer the same combination and therefore a case study was the chosen design.

Case description

Fokker Technologies consists of four business units: Fokker Aerostructures, Fokker Landing Gear, Fokker Elmo and Fokker Services. The group achieved in 2013 a turnover of €762 million with 4688 employees and is since 2015 owned by GKN Aerospace. Plants and sites can be found all around the globe (e.g. Mexico, Romania and USA). For this research I used the plant of Fokker Aerostructures in Hoogeveen (Fokker), the Netherlands. At this moment around 1000 employees work at this plant.

In the composite factory employs 165 employees of whom are 143 male (86,7%) and 22 female (13,3%). On average, an employee is 47 years old and has an average tenure of 18 years. This shows an organizational commitment of the employees towards Fokker as well as a commitment of Fokker towards its employees. Human capital is, due to among other things the needed craftsmanship for production, very important for Fokker. It therefore pays for example its employees above competitive salaries and provides several fringe benefits such as evening school. In casual conversations employees also indicate to be proud to be working for Fokker. To keep the workforce up in shape Fokker employs multiple initiatives such as free fruit at work and a voluntary, yearly health check. Employee absence due to sickness is on average around 5%.

(26)

26

the nature of contracts and long term commitments in this industry10. This study focuses on the Apache11 program, specifically the lamination department. The Apache program has been running for years and as a result has little to no surprises and is therefore easily manageable. That made it an ideal candidate for Fokker to implement smart manufacturing in this production process first.

Data collection

In order to formulate meaningful conclusions based on the findings of this study, triangulation, ‘the use and combination of different methods to study the same phenomenon’ (Voss, Tsikriktsis & Frolich, 2002, p. 206), was employed. Even more specific, to improve the reliability of this data triangulation was used (Patton, 2002): the use of multiple data sources. Three of the six main data sources, documentation, interviews and participant-observations (Yin, 2014), effectuated this reliability improvement. The other three sources, archival records, direct observations and physical artifacts, were either not available (records and artifacts) or when considered too intrusive to result in meaningful data (direct observations). Therefore, I used documentation, interviews and participant-observations as sources. First of all, for the interview data I interviewed employees (both on operational and tactical level) to be able to report (expected) changes. Secondly, internal documents (e.g. process descriptions) were reviewed to objectively view the (proposed) changes. Thirdly, I observed and interacted with employees to validate the first two data sources with real world examples. Also, this sort of data collection helped to get a better understanding of the organization and helped in some cases with the interpretation of data.

The interviews comprised two different groups: operators (production employees) and Fokker 4.012 stakeholders. The operators were interviewed to establish a baseline concerning experienced job autonomy before SM implementation. The goal of these interviews was to uncover the perceptions of the employees about their original job autonomy and underlying reasons. The interviews had a maximum duration of 10 minutes each and answers were noted. The format of the interviews was semi-structured. This provides possibilities for flexibility while also ensuring reproducibility of the same interview with different participants in different

10_{As most notable example the Joint Strike Fighter (JSF a.k.a. F-35) production and maintenance contracts} 11_{Current production concerns the AH-64E Guardian (a.k.a. AH-64D Block III). For further information:}

https://en.wikipedia.org/wiki/Boeing_AH-64_Apache

(27)

27

timeframes13. To effectuate solid semi-structured interviews, a protocol was developed based on the methodologies described by Emans (2007), see Appendix IV. The used constructs in the protocol were based on the existing questionnaire, including validated scales, the Work Design Questionnaire (WDQ) (Morgeson & Humphrey, 2006). An overview of the operationalization and Dutch translations of the job autonomy constructs are provided in Appendix III. Before using the protocol it was validated by testing it on an employee from another department. The remarks concerned phrasing and choice of words and the protocol were adapted accordingly. The interviews with the Fokker 4.0 stakeholders were conducted in similar fashion. These stakeholders were selected based on their jobs, background and responsibilities in the Fokker 4.0 project. After collaboration about suitability of the stakeholders for this study with the head project manager of Fokker 4.0, a shortlist of selected persons resulted in the selection of these three specific stakeholders. The questions in the interview protocol were founded on the theoretical body of SM as result of the systematic literature review (see Appendix VI for the used interview protocol). The questions however focused on Fokker 4.0 specific issues. The interviews were semi-structured, lasted 30 minutes each and were recorded. Afterwards the interviews were transcribed14 and analyzed. Quotations of these stakeholders in this paper were translated from Dutch to English.

The internal production documents provided insight in how production processes and steps were designed in the current situation. These documents included formal reports, quality regulations and design documents. Furthermore, I used planning, design and change documents in combination with user cases in order to determine what the intended outcome of the change was. First of all, I searched on my own for these documents. Secondly, I contacted several people who I thought could be of use for me in the search for relevant documents. It also happened that people approached me when they heard what I was doing. These people showed me different documents and one business architect helped me make sense of the design documents.

The participant-observations were primarily focused on the production process. I have participated at all the production phases during ‘aanmaak’ (making) in which I had the occupation as an intern. During these participations I could ask a lot of questions about the process, work methods, etcetera without being seen as an outsider or disturber, since I also

13

Another student within the same PhD project might replicate the interviews after the completion of the PoC

(28)

28

worked at Fokker. Furthermore, I have been stationed for my regular job as change agent at multiple departments in the back-office to get to know different parts of the organization and ways to see the production process (e.g. production-engineers take another approach than IT-guys). Interesting observations or remarks were manually noted in a logbook.

Data analysis

For the analysis of the data I chose a pattern-matching technique. It “compares an

empirically based pattern (..) with a predicted one” (Yin, 2014, p. 143.) and is therefore the best

suitable technique for this study to assure objective, reliable and valid findings. To analyze the interviews with the operators the collected grades were combined in one table (Appendix V). To interpret the results, the grades were compared with the notes of the interviews and a line of reasoning was formulated. This was presented to the production manager of the Apache lamination department who validated the findings and reasoning. For work-methods autonomy he provided additional insights in the measured differences. For the analysis of the interviews with the Fokker 4.0 stakeholders a codebook was developed using the method of DeCuir-Gunby, Marshall and McCulloch (2011), which can be found in Appendix VII. After developing the codebook it was than applied at the respective interviews to provide empirical and reliable data for this study.

(29)

29

IV. Results

This chapter provides the current perceived level of job autonomy by production employees (‘operators’) and a process-as-is description of the lamination department. These two matters serve as the baseline of job autonomy. Furthermore, expectations of employees entrusted with the implementation of Fokker 4.0 and a process-to-be description of the lamination department will guide the expectations of how SM will influence job autonomy. The reported figures in this chapter can be found in Appendix V.

Operators’ job autonomy baseline Descriptive factors

The Apache lamination department employs 10 operators and 8 of them were willing and able to participate anonymously in this study. All of them are males and their average age is 46. The operators have been working at average for 16 years at Fokker, of which 8 years in the lamination department. These figures are very similar to the earlier presented figures of the whole composite factory which makes this a representative sample for the factory.

Work-methods autonomy

The interviewed operators reported very limited concerning work methods. The reason for the largely absence of autonomy in this regard lies in the importance of tested and validated procedures in the aviation industry. Every procedure, before going live in production, is vigorously tested and documented resulting in a FAI (First Article Inspection). This FAI document proves that a product can be built according to production requirements. It is therefore, from that point on, the foundation for descriptions of how a specific product needs to be manufactured. This includes among other things production instructions, tools to use and auxiliary materials. An investigation to a product failure starts with a check if the all the production steps were carried out as prescribed. Therefore, it is very important for an operator to follow the production instructions to the letter. Not doing this can have severe consequences for the product (e.g. product failure15), the company (e.g. reputational damage) but also for the

15_{One operator told an anecdote about a test flight in which the floor panels of the plane in question let go and}

(30)

30

employee (e.g. dismissal). In other words, the nature of the industry renders work-methods autonomy virtually non-existent.

There are two exceptions regarding the absence of work-methods autonomy: use of workspace and craftsmanship. An operator may for example have a tidy or sloppy workstation and have his own preference of arranging tools. Within limits though. At the end of the day operators need to clean up their workstation and store away tools and materials on predesignated places. Workstations are job related, not operator related. Since operators generally do a different job each day and therefore work at different stations, a clean desk policy was implemented. Craftsmanship concerns itself with all the not specified steps in production. To illustrate, the last step before a product can go to the autoclave is to bag it and make it vacuum sealed. This requires a lot of skill and experience and is very hard to specify in a document. Therefore, it is left to the operators how to execute this step, as long as the product is finished to specifications.

The ambiguity between strict regulation and craftsmanship is not directly reflected in the average grade of 4,0, but there is a clear division of operators into two groups. One group scores work methods autonomy very low (1s and 2s) in contrast to the other group that gives high scores (7s and 8s).

Decision making autonomy

(31)

31

Work scheduling autonomy

The highest perceived autonomy is that of work scheduling. Production schedules are prepared per week by the planning department and are strongly influenced by the autoclave cycles16. Operators have therefore no influence over production schedules. They check every morning when they arrive the job(s) they need to finish that day. Depending on the duration of a certain job, they will be manufacturing one or multiple jobs. These jobs are scheduled so that the whole (set of) job(s) can be finished in one day, and it is expected of the operators to finish the jobs before going home. While deadlines are strict, the operators indicated that the sequence of jobs on a day is completely free, as long as it is all finished at the end of the work day. Furthermore, operators can check their schedules for the whole week at the Thursday before the scheduled week. This provides another layer of flexibility since operators are allowed to mutually switch jobs. This can be for reasons of for example physical injuries or to keep a certain skill up-to-date. Thus, production schedules are strict and an operator has absolutely no influence on this schedule. However, on a day-to-day basis operators are provided with relative high levels of autonomy: free to choice of job sequence and mutually switching jobs. Therefore, operators grade work scheduling autonomy as the highest kind of autonomy with an average grade of 7,4. Corrected for an extreme outlier it even rises to an 8,3.

Conclusion

Overall, the job autonomy of the operators at Fokker is perceived as moderate. The best way to describe it is that a very well formulated and strict frame is in place, but within that frame there is relatively high perceived job autonomy. To control for possible effects of job autonomy on job satisfaction, the interviews with each operator ended with the question how satisfied they were with their job. The answer was very satisfied, represented in an average grade 8,8. The sense of comradery among employees was given as explanation for this grade, just as organizational/supervisor support. The low level of job autonomy seemed of minor to no interest to the operators. Furthermore, they all very well understood that rules and regulations prohibited high levels of job autonomy. Thus, the negative effects of low job autonomy levels seem not to exist in this case, or are at least very well mitigated by different factors such as relations with colleagues and perceived support.

16

(32)

32

Changes in production process Process as-is

Before an operator in the lamination department can start with his job, the raw material needs to be prepared at the ‘Prepreg’ department. After the necessary steps such as cutting the materials according to specifications the materials are shipped to the lamination department on a cart. The first step of an operator is then to pick-up the materials assorted for the job he has to do. He brings it to his workstation and scans the barcode on the accompanying routing document (‘BV’). After scanning, several systems provide the operator with a set of instructions and list of tools to use for this job on two computer screens. The operator then goes on to the actual producing (laminating) the product following the instructions meticulously. Afterwards, he bags the finished product and creates a vacuum in the bag. It is then ready to be picked up by the operators from the autoclave department. The BV is put on top of the product as a signal to the autoclave operators that it can be picked up, which they do at the end of the work day.

All production administration (e.g. measurement notations and stamping control checks) is written up on paper, on the paper BV17. The BV will physically follow its related product through the whole factory. A missing BV can therefore cause production delays, since it contains all the needed production and routing details. Only when the product leaves the factory to the customer the BV is scanned and saved digitally. An overview of the described process is provided in figure 4.1.

Process to-be – near future

The actual production remains the same as compared to the current situation. However, there are differences in three main areas: the IT system, authorization and administration. The new IT system combines all of the existing systems that currently operate separately from each other. That means that the production routing, scheduling, stock levels, etcetera will be dealt with using one system. This will enable a new level of transparency for all people involved (Expert 2) and also provide a better foundation to analyze processes (Expert 3). Both the other areas of difference, authorization and administration, are a direct consequence of this new, all-embracing IT system. First of all, in the current production process there is no formal authorization check. In the new situation every operator will have his assigned authorizations actively checked by the

17

(33)

33

system. Secondly, the administration of the production progress will be completely digital and the paper BV will be history: ‘The paper world will disappear for the operator; he will do his

[administrative] work on screens’ (Expert 1). All measurements will be directly entered into the

system by the operator. Also, his personal stamp will be replaced by a personal badge with which he can ‘digitally’ stamp the control points. Just as is done currently, but then digital and with a different tool. The largest difference for the operator in this specific process is that currently operators can freely choose in which sequence they will do their jobs for the day. In the new situation a sequence will be provided, or as Expert 2 put it: ‘The operator currently determines

(..) the sequence in which he performs his jobs, those simple instructions we can take over’.

Operators are allowed to deviate from this preferred sequence, but only if they offer a reasonable explanation. An overview of the new process is provided in figure 4.2. Lastly, operators will gain a more direct way to provide feedback on production steps through an integrated feedback loop in the user interface they use to execute their jobs (Expert 1).

Process to-be – distant future

In the more distant future it is expected that the level of automation has increased slightly but that the operators are still essential for the production process (Experts 1, 2 & 3). To remain competitive in a relative expensive country as the Netherlands the experts expect that a lot of non-value adding activities can be eliminated (e.g. logistical robots to move molds (Expert 1) and the digital instead of physical distribution of schedules by planners (Expert 2)). Furthermore, better coordination in the supply chain will probably be achieved by sharing information which enables shorter lead times and lower costs (Expert 2) and would eliminate the need for multiple test rounds in different parts of the supply chain (Expert 3). Within ten years it could be possible that production machines schedule their own maintenance and reschedule their production schedules to compensate for an unexpected failing machine (Expert 3). In short, the consensus is that operators will remain at the heart of the production process in which they will be more and more supported by machines and IT leaving the most complex production tasks for the operators to execute.

Conclusion

(34)

34

production will be more supported by machines and a fully integrated IT system than nowadays. The integration of the supply chain will presumably enable the whole production process from raw material to end product to be more efficient and transparent.

Legend figures 4.1 and 4.2

Object Description

Prepreg The previous department in the production process which prepares the raw

material for production.

Operator 1 The primary operator concerned with a specific lamination job.

Production systems A combination of two IT systems used at Fokker. The first one contains

instruction sets and is called TPD. The second system contains the order progress, called DPD.

Pitch team A team consisting out of four functions: team leader, process engineer, planner and a quality checker. They are together responsible for supervising the production team (operators) and the process.

Operator 2 Another operator which is like operator 1 doing his own tasks, but is called in for a short amount of time (ca. 5 minutes) to check for small mistakes and if all stamps have been put.

(35)

35

(36)

36

(37)

37

V. Discussion

Answering the research question

This paper started off with the question how smart manufacturing influences job autonomy. It quickly became apparent that formulating a satisfactory answer was harder than initially thought. Both constructs are multi-faceted, which calls for a more specific research question to be able to provide a more meaningful and substantial answer. The reformulated research question, as mentioned earlier, therefore became: How do the three phases of SM influence the work-methods-, work scheduling- and decision making autonomy of employees?

Based on this case study in the aviation industry, the first phase of SM, digitization, is unlikely to influence work methods autonomy since digitization does not interfere with the actual execution of the job itself. Therefore, logically it is safe to assume that this sort of autonomy is unaffected and will remain on the same level. Decision making autonomy though, will presumably be influenced and increase as a result. The intensified inclusion of employee feedback on the production process accounts for this expected effect. Operators will be more able to actively report potential changes and improvements and that calls for personal initiative and judgement: to see and asses a situation, come up with a suitable solution and report it. This is precisely the mechanism that enhances decision making autonomy and is in line with the reasoning of De Treville, Antonakis and Edelson (2005). The effects may however be mitigated by the already relatively high perceived decision making autonomy in this specific case. In contrast to this increase in autonomy, work scheduling autonomy will most likely be restricted. Declining employees the possibility to schedule their own job sequence and imposing strict schedules will accomplish this specific decline in autonomy. Employees had already no say in larger planning decisions, but losing scheduling autonomy at this personal level will supposedly result in severe changes in perceived levels of autonomy. Also, the lost flexibility concerning mutually switching jobs will lower work scheduling autonomy. Both developments support the findings of Parker (2003) that workflow standardization limits job autonomy. An important note in this regard is that, according to Expert 3, flexibility can only be limited to a certain degree. Without a little slack in operational processes the work flow would stagnate.

A case study in the aviation industry THE EFFECTS OF SMART MANUFACTURING ON JOB AUTONOMY:

THE EFFECTS OF SMART MANUFACTURING

ON JOB AUTONOMY: