2
Abstract
Cyber-Physical Systems (CPSs) include an interconnectedness of collaborative elements that can act autonomous and within various levels of the production process. CPSs can fundamentally change manufacturing practices as we know it, including the role and responsibilities of operators. This study aims to increase the understanding of the impact of Cyber-Physical Systems on knowledge characteristics and autonomy of operators. A qualitative in-depth case study was conducted at a pharmaceutical company that introduced a CPS 1.5 years ago. It was found that the CPS led to more real-time information generation, leading to an increased visibility of the production process. The CPS decreases the perception of job autonomy while the perception of knowledge characteristics of operators is in general positively affected by the CPS. Future research is needed to verify the findings for more matured CPSs.
Keywords: Cyber-physical systems, autonomy, knowledge characteristics, case study
3
Preface
This thesis is the final part of my master study Technology and Operations Management. The last six months was an intense journey, but rewarding in the end. I would like to express my gratefulness to all who provided support during this research. In particular my supervisors at the university for their effort, interest and dedication to provide constructive feedback and support. The supervisor meetings and discussions were very useful for the process of the research and the structuring of my thoughts.
5
1. Introduction
Industrial manufacturing has faced multiple radical changes in last the 250 years. Starting with the mechanisation and steam power in the late 18th century, to mass production on the assembly line in the late 19th century, to the automation and computerization of production in the late 20th century. Currently, the industry is facing a fourth major revolution called industry 4.0. “Industry 4.0 aims for optimization of value chains by implementing autonomously controlled and dynamic production” (Kolberg & Zuhlke, 2015, p. 1871). An important theme within industry 4.0 is the digitization of industrial manufacturing (Hirsch-Kreisen, 2016), by using elements such as are real-time production information and networked systems (Herman, Pentek & Otto, 2015). The developments are expected to have a significant consequence on the role of operators due to an increasingly open network, virtual work platforms, and extensive interaction between human and machine (Kagermann et al., 2013). The contribution of this study is to investigate the impact of these developments on operators by a qualitative case study.
6
system monitoring, diminishing the need of intermediate-skilled workers (Autor, 2013), which is referred to as deskilling (Hirsch-Kreisen, 2016). In the deskilling scenario, CPSs will deteriorate the need for autonomous worker decision-making in low-skilled jobs due to the determination of work by the technology (Dworschak & Zaiser, 2014). Yet, there is limited empirical evidence that supports these ideas since most studies are theoretical and generic (e.g. see Gehrke et al., 2015). More research is needed to gain insight on the direction of the impact of these technological developments on workers, which is useful for the consideration of future job design.
The aim of this study is to provide empirical insight on the impact of CPSs on the worker. We will do so by a qualitative case study at a pharmaceutical company that introduced a Cyber-Physical system two years ago. The associated information system has resulted in a change of tasks that operators have to perform. We want to assess how CPSs affect operator autonomy and the knowledge characteristics by answering the research question: How do CPSs affect the autonomy and knowledge characteristics for operators?
The next chapter will describe the theoretical background of this thesis. The third chapter will discuss the methodology. The fourth chapter will describe the case. The results of the study are shown in the fifth chapter, followed by the discussion in chapter six. The research question will be answered in the concluding chapter, including the limitations and directions for future research.
7
2. Theoretical Background
This section will elaborate further on the CPS concept and its impact on worker autonomy and the knowledge characteristics. In addition, the difference between CPSs and previous developments will be explained including reasons why the CPS concept is relevant for the industry. The main concepts in this section include Cyber-Physical systems and the relation towards job autonomy and knowledge characteristics. 2.1 Cyber-Physical Systems Background
8
9 predicting future scenarios that lead to the ultimate goal of an adaptable system enabling self-optimization and autonomous response (see Figure 1). An example of the autonomous response is a change in production sequence to avoid delivery delays or because a machine malfunction is expected. The second characteristic of CPSs, connectedness, relates to the ability to establish and to connect with other elements of the system. Connectivity replaces the isolated information technology by connected components and is a key requirement for CPS implementation (Schuh et al., 2017, Qin et al., 2016). It is required in CPSs to push data into production before production, and to confirm production results in real-time during and after production (Schuh et al., 2017). Connectivity enables remote control of machine tools to assess for instance machine parameters. Elements such as materials, sensors, machines, products, customers and supply chains are more and more able to connect to each other, exchanging information with each other and control each other autonomously (Qin et al., 2016).
10
11
- Wearable sensors that monitor the condition of the worker including heart rate, breathing, toxic gas exposure. The data can help the managers in decision-making to visualize and anticipate for unsafe working condition (Badarinath & Prabhu, 2017). - Camera-based sensor measurement in a production line that support visualisation and
decision support of workers, to foster production and logistic efficiency (Lewin, Weber & Fay, 2017). It supports real-time system monitoring and control. Simulations based on collected data can help to eliminate weaknesses and errors in the production process.
In addition, CPSs are one of the main elements that led to the development of the industry 4.0 concept, introduced in 2011. Industry 4.0 aims to meet individual customer requirements, increase production flexibility and responsiveness by enabling last-minute production changes and enable end to end supply chain integration (Kagermann et al., 2013). In the context of Industry 4.0, CPS include “smart machines, storage systems and production facilities capable of autonomously exchanging information, triggering actions and controlling each other independently” (Kagermann et al., 2013 p. 5). Considering the scope of this research industry 4.0 will not be explained further (for more information about Industry 4.0 we refer to Kagermann et al., 2013). Nonetheless, it is important to know the implications that industry 4.0 may have since it will most likely increase the adoption and application rate of CPSs. 2.2 Job Design
12
includes five tasks characteristics; autonomy, task identity, skill variety, task significance, and feedback from the job.
The number of studies that considered job design is tremendous and the job characteristics model is still used to assess the design of jobs. There is a good reason since job design is important for the performance of individuals and groups. Morgeson and Humphrey (2006) argue that many studies including the work of Hackman and Oldham lack detail by neglecting many work characteristics. They created the work design questionnaire (WDQ) to provide a more comprehensive measure to assess job design. The WDQ divides job categories into task, knowledge, social and physical characteristics. Task characteristics are primarily concerned with how work occurs and how specific tasks are performed, whereas knowledge characteristics deal with the knowledge demands of work (Humphrey, Nahrgang & Morgeson, 2007). Considering that intelligent data management is a key component within CPSs, we argue that CPSs affect the knowledge demand of workers. Therefore, we focus on the knowledge characteristics as described by Morgeson and Humphrey (2006). In addition, a high level of CPSs intelligence influences the autonomy of workers, as explained earlier. Therefore, the second main construct includes the level of autonomy, which is part of the task characteristics in the WDQ. The justification for autonomy and knowledge characteristics are described next.
Autonomy
13 affected. Additionally, the increased information availability and information about capacity of machines can support the adaptiveness of the production planning in CPSs (Wang et al., 2015). Consequently, the work scheduling autonomy may be affected. This makes autonomy an important aspect to consider in the context of developing CPSs. Knowledge characteristics “Knowledge characteristics reflect the kinds of knowledge, skill, and ability demands that are placed on an individual as a function of what is done on the job” (Morgeson and Humprey, 2006, p. 1323). The knowledge characteristics include information processing, job complexity, problem-solving, skill variety and specialisation. Knowledge characteristics are generally concerned with the enrichment of jobs (Morgeson and Humprey, 2006). “Information processing is the extent to which a job necessitates an incumbent to focus on and manage information” (Humphrey, Nahrgang & Morgeson, 2007, p. 1335). Jobs differ in the degree of information processing and monitoring. The development of CPSs increase the availability of information generation, requiring more information to be monitored and processed (Hirsch-Kreinsen, 2016). Generated data need to be transformed into information and information to understanding (Zuboff, 1988), requiring abstract thinking and analytical reasoning. In addition, CPSs may include machine to machine communication, requiring more in-depth system knowledge of workers and the ability to handle abstract information (Dworschak & Zaiser, 2014). “Job complexity is the extent to which a job is multifaceted and difficult to perform” (Humphrey, Nahrgang & Morgeson, 2007, p. 1335). The information generation of CPSs is expected to result in an increasing complexity of work, leading to unknown requirements for all activities (Hirsch-Kreinsen, 2016). Complex tasks require higher levels of skills which tend to increase the mental demands of individuals and generally have a positive influence on the job satisfaction. Yet, too much complexity may overwhelm individuals, resulting in perceptions of work overload and reducing efficiency (Humphrey, Nahrgang & Morgeson, 2007).
14
problems, in particular during the introduction of CPSs (Hirsch-Kreinsen, 2016). Yet, CPSs may also have a negative effect on the ability of the operator to adequately solve problems. Automation within CPSs for instance can strongly affect the role and requirements of the operator since operators are “expected to intervene into the production process in case of malfunctions only, but cannot develop the respective competences during the troublefree process” (Dworschak and Zaiser, 2014, p. 348).
“Specialization is the extent to which a job involves the performance of tasks requiring specific knowledge and skill” (Humphrey, Nahrgang & Morgeson, 2007, p. 1335). It represents specific knowledge and skill within a specific area. Specialization is concerned with efficiency, routinization and a mechanistic approach of work and is often associated with negative or no effects on job satisfaction (Morgeson & Campion, 2002). Unlike the other knowledge characteristics which increase job motivation and satisfaction of work (Morgeson & Humphrey, 2006).
Skill variety is defined by Hackman and Oldham (1976) as “The degree to which a job requires a variety of different activities in carrying out the work, which involves the use of a number of different skills and talents of the person” (p. 257). Skill variety refers to a broad set of activities and tasks that need to be performed, whereas specialisation focusses on skills for a particular area. Skills requirements changed in the course of the history. Scientific management and job simplification for instance led to a substitution of skilled jobs and increased the specialization of jobs, whereas computerization in the 20th century led to the reduction of moderately skilled jobs (Frey & Osborne, 2013). Yet, future skill requirements are uncertain. In the context of CPS, new complex tasks may be introduced that require additional skills or diminish existing skills. This is also referred to as upskilling and deskilling. The upskilling and deskilling concepts and the relation towards the knowledge characteristics and worker autonomy will be described briefly in the next subsection. Considering the different views and perspectives, studies are required to determine which skills are needed for the CPS implementation (Dworschak and Zaiser, 2014).
2.3 Deskilling and upskilling
15 simplified and routine tasks that are easy to learn. The extensive division of work increases the control of the manager over the process, destructs the craftsmanship of workers and diminishes the autonomous control of workers over the process (Braverman, 1974). Upskilling is the other perspective and is about technological developments that lead to jobs with more complex tasks that require higher levels of skills (Zuboff, 1988, Kagermann et al., 2013). Zuboff argues that information technology leads to an upgrade of qualifications, such as the ability of operators to translate data into information, requiring for instance analytical reasoning abilities, which Zuboff refers to as “intellective skill.” Intellective skill is about the shift from physical cues to sense-making based on abstract cues such as inferential reasoning and system thinking (p. 95). Kagermann et al. expect that technical developments such as CPSs lead to more worker autonomy and self-development opportunities for workers. In addition, CPSs will lead to more data and information generation and accessibility using all kinds of devices and assisting systems. This may affect the quantity of tasks, as well as the introduction of new tasks such as data and information processing (Gehrke et al., 2015). In addition, tasks that are hard to computerise may require more creativity and social intelligence of workers (Frey & Osborne, 2013).
A third perspective is the combination of the upskilling and deskilling perspective, where the general development leads to the polarization of skills between workers (Galli, 1991). The polarization is the result of the reduction of medium-skilled work. Tasks transfer to employees with high-skills such as for example production managers, production engineers and software engineers or can even be carried out by the CPS system itself (Dworschak & Zaiser, 2014). For instance, by automating activities on an intermediate level such as demanding production work, system monitoring or routine based administrative tasks (Hirsch-Kreinzen, 2016). Skill polarization leads to a strong division of labor. Tasks that involve simple activities provide the low-skilled workers little freedom to act, such as standardized monitoring and controlling. Highly qualified workers emerge and take on various tasks and deal with disturbances in the process and production management.
16
showed an increasing polarization of skills between the studied period of 1980 and 2005 which is believed to be a result of computerization (Autor & Dorn, 2012). Considering that CPS developments have a firm digital element, skill polarization may increase with the development of CPSs.
17
3. Methodology
Yin (2014) indicates that case studies are relevant when how questions are used when the investigator has little control over events and when contemporary events are being studied in a real-life context. Considering that CPSs are a rather unexplored area in an empirical sense with general CPS definitions, the description of the phenomenon can increase the understanding of what is going on. The case study methodology provides an excellent in-depth analysis to describe a particular phenomenon. In addition, based on current CPS theories the prediction is made that job autonomy and knowledge characteristics will be affected by CPS maturity, which will be tested in the case study. Thus, this study has a deductive nature. This chapter will discuss in more detail the methodology of the study including the data collection methods and the data analysis.
3.1 Data collection
According to Yin (2014), data in a case study can be collected based on six sources: documents, interviews, direct observation, participant-observations, archival records and physical artefacts (Yin, 2014). The first four data collections methods were used in this study for the understanding the CPS impact on operators in the pharmaceutical industry. This data triangulation increases the construct validity of the study. First, internal documents were studied to reveal the process of the company, the functioning of the CPS and to gain insight on potential implications on workers in the case company. Second, interviews were conducted with operators to measure the implications. Third, direct observations were performed on the shop floor to see and cover the actions in real-time. Fourth, during this thesis project, a minor unrelated side-project was performed which enabled me to interact with operators in their daily work, allowing for data collection by participant observation. Physical artefacts and archival records were not available and therefore not considered.
18
The interviews that were conducted were focussed on the operators. The results provide a baseline for the perception of operators on autonomy and knowledge characteristics in the current state of the CPS. A semi-structured interview method was applied to allow for replicability and to have some flexibility in the form of new questions that may emerge during the interview. The validated work Design Questionnaire of Morgeson & Humphrey (2006) was used to measure the autonomy and knowledge constructs. We used WDQ questions that were translated in Dutch for a PhD project. The translated questionnaire can be found in Appendix I – Interview protocol operators. The questionnaire includes 32 questions based on a 7 point Likert scale and 11 open questions. The questionnaire took around 30 minutes to finish. Six out of 18 operators of Hormone department were interviewed. Their age varied between 23 and 53 years with an average of 44, which is representative for the manufacturing site. The employees have worked at the company for an average of 13 years, of which 7 in the Hormone department. All respondents had been working with the CPS since the implementation 1.5 years ago.
The results of the interviews were discussed and validated with high-skilled staff. Short case study interviews (Yin, 2014) were used, directed towards high-skilled staff. These semi-structured interviews were focused but maintain an open element in order to gain new insights. The interviews were used to corroborate previous findings and were therefore performed in a later stadium. Multiple interviewees were engaged to test the genuineness of the views. An interview protocol was developed prior to the interview to identify interviewees, design a set of questions, explain project aims and other formalities to the interviewee. Interviews were recorded and transcribed to enable data comparison at a later stage.
The direct observations included side-walks during production and included the observation of consultation sessions between operators and automation staff in some cases when issues arose. This provided additional information about the study of the topic, the role of operators and the impact of the implemented CPS.
19
as work methods, decision-making issues, characteristics of their job or procedures and provided insight about the production process and work of operators. 3.2 Data analysis In contrast to quantitative data, which can use statistical software to analyse collected data, the qualitative data involves more text. This rich data was analysed using the qualitative data-analysis tool Atlas TI. The interviews were recorded and the audio records were repeatedly listened to, to ensure that all information was processed into transcripts. A pattern matching logic was applied to compare the predicted pattern with an empirical pattern. Pattern matching is especially relevant if the descriptive conditions are defined prior to the study (Yin, 2014). In this study, the descriptive conditions are the deskilling and upskilling scenarios due to CPS implementation and the resulting implications on job autonomy and knowledge characteristics. The similarity in the empirical and predicted patterns increase the internal validity (Yin, 2014).
20
21 Figure 5. Example of tablets in a blister pack
4. Case description
4.1 General information The case company is part of an international pharmaceutical organisation with over 70.000 employees, of which 500 are part of the production site that is studied. The product portfolio of the site includes high volume low variety generic products such as hormone pills, vaccines and lactulose. In most cases, generic products are available after patent production dates expire and are associated with lower profit margins. The production process of the site is dominated by automated packaging lines that enable high volumes. In this study, we focus on the production department that packs hormone related products. The hormone production includes two production lines. The products and production layout of both lines will be discussed briefly.
Products
Two main products are produced in the hormone department which are called dydrogesterone and estradiol. Both products are made-to-order under two brand names and sold over the entire world. Products may contain customer-specific content such as the language of the package and information leaflet or the location of printed 2D matrix codes (see Figure 4), but also differences in the content of information that is associated to the 2D matrix code due to regional legislation differences. As a result, the variety of packaging possibilities increases exponentially and is rather complex. Production flow The hormone department contains two packaging lines. One is a fully automated packaging line which processes in particular large orders. The second line is a semi-automated packaging line, including a manual stacking station and is dedicated for smaller orders. Both packaging lines process pharmaceutical tablets into blisters (see Figure 5). The process flow of the manufacturing line is shown in Figure 6.
23
Task of operators
The main function of the operators is to keep the line running. This includes the replenishment of packages and information leaflets into the buffers and to react to minor issues such as for instance the removal of a jammed information leaflet in the machine. The operators have to perform setups and line clearance procedures in cases of order changes for both lines. In addition, the semi-automated line requires operators to stack products manually and scan bundles, cases and pallets manually in aggregated serialization batches. 4.2 Serialization and aggregation Future serialization legislation in the pharmaceutical industry requires that products can be traced in an aggregated level (see Figure 7). Each item requires a unique identifier (UI). Every actor in the supply chain must be able to trace the origin of the items. The aim of serialization is to counterfeit fraud. Current serialization is often only required at a batch aggregation, which allows counterfeit. Yet, legislation in the near future requires serialization to be present at an aggregated level including the item, the bundle, the case and the pallet. This aggregated serialization requires practices including recording, authenticating and sharing of records before dispatching. Serialization on an aggregated level has an impact on the manufacturing procedures, the required technology, the role of operators and the role of supporting staff. Some countries such as China and Russia already require aggregated serialization. As a result, a few production lines in the case plant are designed and operational to facilitate the aggregated serialization process, including the hormone department.
Serialization is essentially a practice that allows the traceability of individual items in the supply chain. The main goal is to have real-time information visibility regarding the whereabouts of items. Applications in the supply chain are able to read individual items
24 through unique identifiers, providing the basis for visibility of items in all stages within the production environment and supply chain. Serialization requires the vertical integration of various IT systems, from data collection of components (sensors) on single devices up to the communication of data with regulatory agencies and other actors in the supply chain.
Production equipment that is installed to facilitate serialization can evaluate data generated at the production line, enabling data analysists and consequently optimize the process flows in real time. The required equipment to facilitate serialization includes high-speed inkjet printers and camera readers. In addition, information is connected to other parts in the production line, the plant and the supply chain which are relevant elements in the era of CPSs and serialization. To facilitate data generation and traceability, a serialization system structure is required which is explained in the next section. 4.2.1 Serialization system structure The systems that require integration and communication include the Master Tracking System (MTS), the Global Tracking System (GTS) and the Notification Tracking System (NTS). The relation between the systems is shown in Figure 8 and will be explained briefly.
The Global Tracking System (GTS) is the core of the tracking system. The company ERP system issues work orders into the GTS. The GTS allocates orders into the production lines including serialization rules and product recipes with the correct settings of printers and cameras. Once the production processes start, the GTS receives production information from all the lines, including statistics, and serialization codes. This data is stored in the site database and accessible from the ERP system of the plant. The data is shared with the notification Tracking System (NTS) allowing for communication of data with regulatory agencies, contract manufacturing organizations and other entities in the supply chain.
25 on the configurations of the manufacturing line various modules are available to facilitate aggregated serialization on automated, semi-automated or manual manufacturing lines. 4.2.2 Operational implications Implications for the production line Equipment is needed during the labelling, case packing and palletising process that was not required before. This includes new printing equipment and cameras. Printing is used to print data matrix codes on individual items, but also on bundles, cases and pallets. Cameras and scanning equipment are used to recognise and verify 2D data matrixes. Rejection systems are in place to remove non-conforming items. The machines are connected to the network to interact with the production line, to sense downstream machines and drive upstream machines. Each machine provides and manages real-time production data. In addition, the technology that is used in the equipment enables the ability to identify reject typologies which can assist operators, supervisors and managers in detecting the reject causes in the upstream production process. Yet, the additional equipment and components increase the failure probability and equipment downtime. In addition, more setups are required to
26 configure the equipment and its components. As a result, serialization is believed to initially decrease OEE figures by 8-10% (Spence & Malhotra, 2017). Implication for the workers The printed codes and production data are stored in real-time in a database. Features include the real-time remote support and visualization through networks and mobile connections. VPN connections inside the machine enable remote access and control, which can be used for offsite support. This enables the ability to transfer machine configuration tasks from the operator to off-site support employees.
The introduction of serialization and aggregation has led to an increasingly digital environment for the operators to work in, such as an increased interaction with computer screens and real-time visualisation of machine performance on screens. In addition, the introduction of serialization and aggregation require detailed standard operating procedures for workers to support the process. Training programs are in development for operators to allow them to work with the system and its procedures. Yet, it is not exactly known what the implications are on the knowledge characteristics as well as the autonomy of the operators.
27
5. Results
The results of the study are shown in this chapter. First, a description of the state of the CPS within the company is shown. Second, an overview of the results of the autonomy and knowledge characteristics of the interviewees is shown and explained.
5.1 CPS classification case company
Connectivity
Data is being pushed into the system and production results are confirmed in real-time during and after production. The equipment enables remote control of machine tools to assess machine parameters. In addition, sensors are able to recognize items and communicate the status of items which are pushed in real-time to the IT system. Data and production results are shared in real-time with operators through interactive computer screens, increasing the cooperation with human workers and machines. These elements result in a higher level of horizontal integration due to the CPS implementation.
28
on the shop floor. As a result, the second stage of the CPS maturity model, transparency, is not existing in the production process. Transparency is a prerequisite to reach predictive capacity and finally adaptability. Therefore, these stages are not present in the CPS which lead to the conclusion that the maturity of the CPS can be considered as low. 5.2 Overview of interview results The results of the interviews with operators are revealed in this section. The results of the interviews were discussed and validated with the production manager, production engineer and the automation engineer. New insights from these discussions are also revealed in this section if considered relevant.
Table 1 shows the results of the survey included in the interviews and a general perception of CPS implications on each measure. Interviewees were asked to answer questions based on a Likert scale where 1 represents strongly disagree and 7 represents strongly agree. The purpose of the quantitative measures is to create a baseline, enabling future comparison for changes in the future or comparison with other industries. In addition, questions were asked how the CPSs affected each measure. The results were returned per interviewee via email for confirmation. The average score of the knowledge characteristics was 4.7 while autonomy was scored as 3.6. The pharmaceutical industry is heavily regulated, restricting freedom of actions and changes in every hierarchy within the company, which explains the difference in the scores.
In general, interview results showed an enriching and positive effect on the knowledge characteristics. The autonomy is negatively affected by the implementation of the CPS. More details are shown in the next paragraph.
29 Job characteristics Mean score of current job (1 = strongly disagree, 7 = strongly agree) Standard deviation Perceived effect of the CPS on job characteristics Job Complexity 3.5 0.5 + Information processing 5.5 0.6 ++ Problem-Solving 4.7 0.7 - Skill variety 5.2 1.0 + Specialization 4.6 1.0 - Work scheduling autonomy 2.8 1.4 No effect Decision-making autonomy 4.2 1.7 - Work methods autonomy 3.7 1.5 - Table 1. Overview of interview results (n=6) 5.3.1 Knowledge characteristics Job Complexity
Respondents revealed unanimously that the job became more complex due to the CPSs introduction. Additional tasks need to be performed that require operators to focus on their work. As one respondent explained: “It has become more complex. You have to check more and recalculate everything. For example, the screen may tell me that an order exists of 22 pallets, but you have to check that before you can start, with my calculator in front of the computer. Previously it was easier. Previously, you indicated that you have a full pallet, a receipt came out and you put it on the pallet.”
30
rejected by the system due to reading issues that are hardly visible. A PLC signal will reject physical items to be reworked by operators. The item receives the status Reject in the database. In order to reprocess these products, operators need to manually reassign a status to these products in the virtual environment in order to reprocess the product, increasing the complexity of the job.
Currently, there is no uniform legislation for the traceability requirements for different countries resulting in the need for different software versions. This is resulting in different production standards and procedures. Standards are a key success factor in CPS communication (Herman et al., 2015), but the diversity of standards increases the complexity for the design and functionalities of the CPS. The different standards increase the perceived complexity for operators and decrease the ability to fix issues. “With Russian orders, it [the standard] is worthless because too many things are wrong with the system. Such as identifying items so you can trace the bundle back to the case in order to make a new print. That is possible with China but not for Russia.” Information processing
31 that, and to open a troubleshooting screen if necessary. […] Currently, it is sometimes hard to find the information you need on the screen.” Problem-solving Problem-solving is an important issue within the case company since the implementation of the CPS requires more IT-related skills and knowledge which not all operators possess. Before the introduction of the CPS, it was sufficient to possess mechanical skills to configure machines and fix issues. Yet, after the introduction more IT related skills were required of operators to configure machines, by adjusting parameters in the software interface. The inability to intervene in the production process by properly configuring the machine parameters has resulted in major downtime of the line due to the inability of operators to intervene. To increase the intervention ability, a few operators were trained and authorized to perform interventions in the CPS by adjusting camera settings. The ability to intervene on the line is enhanced by real-time visualisation of machine performance on screens. The visualisation increases the ability to recognise problems and to act accordingly. In addition, the ability to get remote assistance of IT staff enables operators to solve problems while IT guides them from a distance.
In some cases, the CPS requires an intervention which can be solved by the operator, such as a box that is missing on the pallet. Operators are required to find the box either physically or virtually (i.e. an error in the data transfer leading to a mismatch between two connected databases, the box is processed in one database, but non-existing in the other).
Issues that prohibit the operator to solve issues are often software related and are akin to software bugs. The introduction of the CPS requires organizations to solve issues especially during the introduction of the CPS (Hirsch-Kreinsen, 2016). The case revealed that often high-skilled specialized IT staff is required since operators are not able to solve issues. The amount of issues however decreased over time (expert 3), which indicates that the system becomes more stable and reduces the dependency on IT staff in the near future.
32 its virtual cyber twin. Issues like these decrease the ability of operators to solve problems since they are highly dependent on others. Thus, although there are elements that enable the ability to solve problems in the CPS, operators perceived that they could not solve issues themselves. Most issues prohibit the operator to solve problems due to dependency on others resulting in a decreased perception of the problem-solving ability. Yet, considering that the CPS is operational for 1.5 years, developments in the future may stabilize the technological infrastructure resulting in fewer errors and a decrease of interference of the IT department (expert 3). Skill variety Operators respond unanimously that the CPS required them to use additional skills related to working with the computer screen, the IT system and the interface of the digital environment. Yet, the perception is that emerging tasks due to the CPS do not require high-level skills, in particular due to the level of routine that is required of the operators. The digitization requires operators to focus more on non-mechanical work. This is reflected in the authorization of some operators to adjust the configuration of the cameras and printing units, resulting in a shift of mechanical to digital problem-solving abilities. As one operator explained: “Serialization and aggregation results in less mechanical work. So, you are busier with the computer to see where the errors can be. To solve them all is of course not entirely possible. But you learn more and more.”
33 require some new knowledge but are perceived as generic. Considering that the diversity of required skills increases, the impact on the specialization is negative. Conclusion knowledge characteristics
The main conclusion is that CPS seems to have an enriching and positive effect on the knowledge characteristics. Overall, operators perceived the increasing complexity and processing of information as challenging, resulting in a sense of fruition. Operators did like the digital elements of their work and the increased variety of skills needed to work in a digital environment, though some perceived it as a complicating factor for their elder colleagues. Operators perceived a decreasing effect on the level of specialisation since the new tasks in the CPS were generic, and only some operators were trained to perform specialized tasks. Yet there were some negative associations with the implementation of the CPS, such as the increased dependency on IT staff. The inability to solve issues due to this dependency was associated with frustration for some.
5.3.2 Autonomy
Work Scheduling autonomy
Work scheduling autonomy received the lowest score in the survey. The sequence of production orders is determined by logistics and is rarely influenced by the production department. Before each shift, the operators receive an update of the production planning status from the previous shift. The sequence of orders is fixed and operators have no influence to change this sequence. The introduction of the CPS had no influence over the work scheduling autonomy.
Work methods autonomy
34
information availability also increased the management control ability, lowering the perceived work method autonomy of operators. As one operator explained: “You must now write down how many cases you have produced. Previously no one mentioned it if you missed a case, but now you have to be right.”
In addition, it is expected that the work method autonomy will decrease in the future due to digitization (expert 1, expert 3). For instance, by the projected implementation of Electronic Batch Records (EBR) that require operators to strictly follow and fill quality samples procedures in digital batch records, which will stop the line if not done correctly or not on time. Decision-making autonomy The standard deviation in the survey responses was the highest of all constructs measured for the construct of decision-making autonomy (see Table 1). The difference can be explained by the different perceptions of operators towards the freedom to make decisions concerning quality issues and line malfunctioning. Quality issues and corresponding decisions are highly standardized and procedural, leading to limited decision-making autonomy for the operator. Yet, operators can decide themselves how to fix minor issues on the line or whether they need maintenance support. With the introduction of CPS however, more collaboration and consultation between operators and team leaders is required in order to determine actions. As one operator explained: “With every minor malfunction you have to contact the supervisor, which was not necessary before. [..] And with serialization you have to approach the supervisor to jointly discuss which follow-up actions are being taken. The supervisor calls maintenance service more often. The responsibility shifts because of this.” As a result, the decision-making autonomy has decreased during the introduction of the CPS.
Conclusion Autonomy
Operators perceived the autonomy as rather low. They have some autonomy over the decisions that they can make in particular during line malfunctioning. The autonomy concerning work methods and work scheduling is limited. Operators know that regulations and procedures are required in the industry, limiting the autonomy of operators. The introduction of the CPS did decrease the autonomy of operators due to an increase of
35
6. Discussion
The consequences of CPSs on the role of workers is a point of debate and there are two dominant schools of thought: those who believe that technological developments lead to an upskilling of workers, and those who believe it will lead to deskilling of workers. The aim of this study was to investigate the impact of CPS on knowledge characteristics and operator autonomy, making a contribution to the upskilling/deskilling debate. This chapter will compare the key findings with prior research.
6.1 Theoretical implications
The conceptual model in this study is built upon the predictions that CPSs will alter the required knowledge characteristics and autonomy of operators. Lee et al. (2015) imply that the CPS autonomy increase when CPS maturity evolves. Yet even for a low level of CPS maturity, we found that the autonomy of operators is affected negatively due to an increase of procedures, standardized work methods and an increasing need for mutual decision-making. The autonomy decreased due to CPS implementation by enforcing a kind of managed work which led to a deskilling effect. The expectation is that the decision-making autonomy and work method autonomy will decrease even further in the future due to a stronger determination of procedures by the more mature CPS. The increasing exposure to digital technologies of operators results in required know-how that is often not in the possession of operators. As a consequence, the modification of the CPS is managed by high-skilled employees. These developments are mentioned by Dworschak and Zaiser (2014) in their automation scenario and is associated with deskilling of non-specialized general skilled employees. An important element in this scenario is the deterioration of autonomous worker decision-making and work methods, which is reflected in the case results.
36 and must define what data means from displays. The increased cognitive demands suggested an upskilling effect on the operators. In addition, more in-depth IT knowledge is required to understand the networking of machines and functionality of the system, to respond or initiate actions when malfunctions occur. The results indicate that competences are required for parameterisation and usage of specialised software skills to operate machinery, as predicted by Dworschak and Zaiser (2014).
Zuboff (1985) argues that the introduction of information technology is applied to automate operations, removing human efforts and their skills. Automating leads to the generation of new information, which Zuboff refers to as the process of informating. The CPS system in the case company is designed to automate the process, reducing the autonomy of operators. Yet, the increasing automation results in more data generation, enabling more monitoring, controlling and interpreting data which was done by the operator and had an upskilling effect on the operator. This explains the differences of the perceived effects on the knowledge characteristics and the autonomy. Consequently, this case does suggest a mixture of upskilling and deskilling effects. 6.2 Practical implications The effects of new information technologies such as CPS can have significant effects on the requirements of employees. Organizations should consider the dispersion of roles and responsibilities between departments, such as the consideration who will need the knowledge to control and maintain equipment and perform interventions. Also, the authorisation of tasks that need to be performed should be considered as the introduction of CPSs increasingly requires performing activities in a digitized environment.
38
7. Conclusion
This study aims to increase the understanding of the impact of Cyber-Physical Systems on knowledge characteristics and worker autonomy. Based upon the up- and deskilling debate, a conceptual model is developed to guide the research and associate important elements for job design in the context of CPS developments. To investigate the effect of the CPS implementation, an in-depth case study was performed using a qualitative research method to answer the research question: How do CPSs affect the autonomy and knowledge characteristics for operators?
7.1 Answering the research question
The results indicate that the knowledge characteristics of operators are in general positively affected by the introduction of the CPS, with increasing complexity, more information processing and an increase in skill variety. Problem-solving is overall affected in a negative way since many problems are not possible to solve for operators due to the dependency on automation staff. However, it is expected that this dependency will decrease in the future since the amount of issues with the CPS have decreased over time and is expected to decline more. The level of operator autonomy is negatively affected by the CPS introduction. Work method autonomy and decision-making autonomy reduced and work scheduling autonomy remained unaffected in the studied case. 7.2 Limitations and future research Multiple methods were used to create reliable, valid and reproducible results. Yet, there were some limitations that need to be addressed. First, the number of operator interviewees is limited in this explorative study with a sample size of 6 in order to gain in-depth knowledge about perceptions and beliefs from each subject. Future studies may incorporate more participants to validate the findings of the questionnaire statistically.
43
Appendices
Appendix I – Interview protocol operators Goedemiddag/goedemorgen, Allereerst bedankt dat je tijd hebt genomen voor dit interview. Ik wil graag even kort toelichten wat de reden van dit interview is. Voor mijn onderzoek aan de Rijksuniversiteit Groningen ben ik bezig met het in kaart brengen van de impact van serialisatie op het aantal taken van operators, de benodigde vaardigheden van operators en de impact op de vrijheid van werken. Omdat jij onderdeel bent van de ploeg waarin serialisatie/aggregratie is geïmplementeerd ben ik benieuwd naar jouw ervaringen. Er zullen een aantal vragen volgen om de impact op jouw werk te meten. De resultaten worden anoniem verwerkt. Er zijn geen goede of foute antwoorden te geven. Voel je vrij om om verduidelijking te vragen als de vraag onduidelijk is. Het doel is om uiteindelijk een advies uit te brengen aan Abbott zodat Abbott de invoering van serialisatie verder kan optimaliseren. Daarnaast zou ik het gesprek graag op willen nemen als jij dit goed vind. Ik zal daarnaast aantekingen maken tijdens het gesprek, tenzij je daar bezwaar tegen hebt. De vragen bestaan voor een deel uit gesloten vragen. Deze kun je beantwoorden door middel van een schaal, variërend van helemaal oneens tot helemaal eens.
Helemaal
oneens Oneens Gedeeltelijk oneens Neutraal Gedeeltelijk eens Eens Helemaal mee eens
44 Algemene vragen Hoe lang ben je werkzaam bij Abbott? Hoe lang op deze productielijn? Hoe lang werk je met Serialisatie & Aggregratie? Knowledge Characteristics
Definition Definitie Job Complexity
Job complexity refers to the extent to which the tasks on a job are complex and difficult to perform. De mate waarvan taken in het werk complex en lastig zijn om uit te voeren. 1. Ik kan in mijn werk maar één ding tegelijk doen (reversed code). 2. De activiteiten die ik doe zijn routinematig en niet ingewikkeld (reversed code). 3. Mijn baan bevat naar verhouding ongecompliceerde taken (reversed code). 4. De taken die ik doe zijn moeilijk. (reversed code). 5. Hoe heeft serialisatie en aggregratie invloed op de complexiteit van voor jouw werk?
Definition Definitie Information Processing
The amount of information processing needed at work reflects the degree to which a job requires attending to and processing data or other information Informatatieverw erking betreft de mate waarin het werk vraagt om monitoren en verwerken van data of andere informatie. 1. In mijn werk moet ik op veel dingen tegelijk letten. 2. Mijn werk vereist dat ik veel moet nadenken. 3. In mijn werk moet ik veel dingen in de gaten houden. 4. In mijn werk moet ik veel informatie verwerken. 5. Hoe heeft serialisatie en aggregratie invloed op de hoeveelheid van informatieverwerking voor jouw werk?
Definition Definitie Problem-Solving
45
Definition Definitie Skill Variety
Skill variety reflects the extent to which a job requires an individual to use a variety of different skills to complete the work Variatie in vaardigheden gaat over de mate waarin het werk vraagt om verschillende vaardigheden om het werk uit te voeren. 1. Voor het werk heb je veel verschillende vaardigheden nodig. 2. Om het werk uit te voeren moet ik veel van mijn vaardigheden gebruiken en inzetten. 3. Het werk vereist dat je beschikt over complexe vaardigheden of vaardigheden van een hoger niveau. 4. Om het werk uit te voeren zijn verschillende vaardigheden nodig. 5. Hoe heeft serialisatie en aggregratie invloed op de verschillende vaardigheden voor jouw werk?
Definition Definitie Specialization
46
Autonomy
Definition Definitie Work Scheduling Autonomy
The freedom to control the scheduling and timing of work Vrijheid in het plannen van mijn werk 1. Ik kan mijn eigen werkplanning maken. 2. Ik kan zelf de volgorde bepalen waarin ik mijn taken doe. 3. Ik kan zelf de volgorde van mijn werk indelen. 4. Hoe heeft serialisatie en aggregratie invloed op jouw vrijheid in het plannen van werk?
Definition Definitie Decision-Making Autonomy
The freedom to make decisions at work Vrijheid in het maken van beslissingen 1. Ik kan mijn werk naar eigen inzicht en initiatief invullen. 2. Ik kan veel beslissingen nemen in mijn werk. 3. Ik mag veel beslissingen nemen in mijn werk, zonder eerst met mijn baas te overleggen. 4. Hoe heeft serialisatie en aggregratie invloed op jouw vrijheid in het maken van beslissingen?
Definition Definitie Work Methods Autonomy
47
