Keywords

1 Introduction

Production Systems are ever-changing, adapting to the market, circumstances and newly available technologies. In the last decades, globalisation and individualisation have changed the planning and operating of production systems, increasing the number of product variants and shortening product lifecycles. Therefore, industrial companies are forced to shorten planning processes and to be more flexible in their production. As a result, production systems are changing from complicated to complex systems [1]. Technologies often associated with Industry 4.0 such as smart automation based on Cyber-Physical Systems (CPS), the Internet of Things (IoT), and Artificial Intelligence (AI) [2] can help to handle the resulting complexity, but create new challenges, especially concerning organisational and human aspects. The following paper will analyse challenges and changes along the product lifecycle and illuminate work in Smart Production Systems.

The term “Smart Production Systems” is always connected, sometimes even used synonymously, with CPS and Industry 4.0. The term Industry 4.0 was first shaped in 2011 as part of the high-tech strategy in Germany marking the beginning of the fourth industrial revolution [3]. Each preceding industrial revolution has brought radical changes. The first industrial revolution at the end of the 18th century enabled industrial production in factories based on water and steam power. The second revolution using electrification to enable mass production started at the end of the 19th century. In the 1970s, the third revolution was driven by the introduction of electronics and Information and Communication Technologies to automate production [1, 2]. Each of the revolutions not only brought technological change but also changed production systems with all aspects including organisational and human factors fundamentally [4]. The first revolution completely changed society from producing at home and in small workshops to the production in factories creating a new working class. The introduction of Scientific Management and work division in the second revolution has changed work until today. The following revolution due to the automation of processes changed work adding more surveillance tasks and requiring a more systematic approach [5].

Currently, the term Industry 4.0 characterizes the emergence and usage of new digital technologies in industrial production, including (Fig. 1): Big Data and Analytics, Autonomous Robots, Simulation, Horizontal and Vertical Integration, IoT, Cybersecurity, Cloud Computing, Additive Manufacturing and Augmented Reality [6, 7].

Fig. 1.
figure 1

Technologies of Industry 4.0 (own illustration)

Full size image

The goal of Big Data is to ensure real-time decision-making and finding patterns in a large amount of data from different sources to optimise production. Although robots have long been used in manufacturing, they are becoming more cooperative, flexible and autonomous, interacting with each other and with humans [6, 8]. Simulation will be used more extensively and can access real-time data to perform more efficient testing so that processes and settings can be improved and optimised before production starts. Therefore, it can help to reduce a waste of time and improve product quality [9]. The Horizontal and Vertical System Integration seeks to connect the entire organisation including all departments and functions as well as to integrate suppliers and customers so that companies are able to interact and connect. IoT refers to objects such as sensors, smartphones, and any machines or devices that facilitate data transfer. In manufacturing, this is referred to as the Industrial Internet of Things (IIoT) [6]. With increasing connectivity, cybersecurity becomes more important because as networking increases, so does the security risk. The information systems and production lines are at increased risk of cyberattacks and companies need to protect their systems. The cloud is an important topic for the contribution of networked system integration to the transformation of Industry 4.0. The main goal here is to increase efficiency by lowering product lifecycle costs and achieving optimal resource utilisation by managing customer-oriented work with variable demand [9]. The next technology is additive manufacturing, which is mainly used for the production of small series of customised products. In the process, efficient, decentralised additive manufacturing systems reduce transport routes and inventories [6, 8]. The last technology describes augmented and virtual reality (AR & VR). AR tools support the integration of computer-based imaging of a real environment with additional and valuable information [10].

All the described technologies and especially a comprehensible combination of these technologies are essential to uncover new potentials in the product lifecycle. To realise this potential not only the technology is needed but also an adaption of organisational and human factors, therefore changing work as a whole. The following examples for the implementation of new technologies and the accompanying change of work are given in the context of recent research projects of the “Institute of Production Systems (IPS)” and the cooperating “RIF Institute for Research and Transfer e. V.”.

2 Digital Planning of Smart Production Systems

One part of the product lifecycle that Industry 4.0 is fundamentally changing is the planning of production systems. As described, not only production systems themselves but also the planning process become more complex because new technologies and requirements need to be considered and new competences and job profiles are required. Furthermore, to keep pace, the organisation of the planning process itself is changing as outlined in the following sections.

2.1 Agile Planning Methods

The growth of the product and its process complexity combined with the cost pressure have placed well digitalised and organised companies ahead into success. With the goal to shorten the product development process (PDP), the product design and production system design needs to be paralleled for efficient time and cost. Simultaneous Engineering has been a helpful method to reduce the PDP, however, space for reconciliation is still present. While mature software systems (such as computer-aided x (CAx), product lifecycle management (PLM) systems, manufacturing execution systems (MES), etc.) are universal tools in design and production areas, isolated solutions remain in the task of production planning. The opportunity lies in connecting isolated solutions to a smart and holistic system.

In the product development process, established IT systems focus not only on end-to-end data management and distribution (e.g. PLM, ERP) but also on operational project management tasks, including the control of planning processes (e.g. time and resource planning, quality gates). In the field of data management, distribution, and use, neutral data exchange formats (e.g. JT, STEP, XML) have become established for uniform description and integration in the industrial context. The networking of services and products provide impulses for future planning approaches. For example, through end-to-end data management, the product development phase can be assessed, therefore evaluating the ergonomics and production time is possible at an early stage and can be considered for further planning initiatives. Moreover, as the data for digital design increases, production scenarios can be simulated beforehand using digital human models to ensure more robust planning. Therefore enabling structured data management will develop more consistent planning [11]. Current system landscapes show perceptible media discontinuities between the leading data-holding systems of product design, the process-oriented project management and collaborated solutions between companies. The challenge lies in creating a holistic dataset, combining and storing accessible knowledge. Putting forth a methodical system in organising new digital work can enhance the technological change and helps to overcome complex challenges [11, 12].

For companies to overcome current volatile markets, adaptability needs to be considered in the design of production systems. This requires that the possibility of redesigning systems comes at a low financial cost and operating the production system efficiently without additional resources [13]. Agile working methods help customer-oriented systems react swiftly to unpredictable and constant changes through distributive innovations [14]. At a micro level, the method aims to change the work processes, structures, and conditions to a time-saving, goal-oriented complex operation. Expected competences demand employees to adapt to their new availability and the production system’s expanded accessibility [15].

The basis of digital, agile work is the formation of interdisciplinary teams characterised by removing barriers between different disciplines and their experts’ collaboration [14]. Successful collaboration can be achieved when comprehensive transparency is presented at the base of the process. The “Backlog” is a communicative tool that provides overview information to keep all team members informed about all tasks at all times [14]. Simple digital tools to organise the work enhance the effects of agile methods. Digitalisation offers numerous approaches to automating workflows, simplifying interfaces, or increasing flexibility [16]. The characteristics of Digital Work are as follow: agility, spacious, voluntary-based, instructing leadership, high dynamic, short-cycle, and digitally supported [17]. The benefits that follow these characteristics are: better management of changing priorities, increase in productivity, faster development and delivery of products, more precise overview of projects, increase in transparency, and higher cooperation quality with the IT department in the company [18].

Digitalisation has been available for use for a decade already. However, it was not until COVID19 that many companies pushed themselves to set up the digital landscape. During the pandemic, collaboration software has helped to create a digital work environment for distributing the workload. Planning any kind of project, contrary to traditional ways, can be achieved regardless of the team members’ location. Nevertheless, digitalisation also comes with an overloading of sensory information and the stress level has been proven to be higher due to constant accessibility. Furthermore, the new distance can cause companies to lose their employees’ attachment [17]. Still, digitalisation in Germany, especially for small and medium-sized enterprises (SMEs) in rural areas, can provide more opportunities worldwide with the availability of skilled workers and improved work-life balance [17]. The goal of the agileKMU project is to develop a digital platform for agile collaboration to design (smart) production systems as shown in Fig. 2. The platform supports the agile collaboration with a goal-oriented provisioning of planning data, a custom workflow management system, and flexible visualisation concepts.

Fig. 2.
figure 2

Digital platform to facilitate distributed collaboration in the agile design of smart production systems, own illustration based on [19]

Full size image

It is assumed that the integration of the digital platform will improve collaboration for SMEs, especially in the design of production systems. Refining the tasks for the Kanban board carries the most significant challenge and benefit at the same time. Adequate effort estimations and task refinements represent initial barriers. The increased transparency of processes in the project significantly simplifies communication between employees. In addition, the task-related data connection streamlines collaboration with other departments. This enables project staff to increase the proportion of their core tasks and reduce the proportion of their secondary tasks (e.g. administrative and organisational processes). Research concepts include collaborative work in VR to enrich collaboration with additional information or connect distributed teams.

2.2 The Change of Planning Processes for and with New Technologies

VR can support agile work in interdisciplinary teams, facilitating not only members working together from multiple locations. With the help of suitable software and hardware, VR can create a controllable experience, expanding collaboration and information dissemination beyond web and video conferencing.

The transformation of cardboard engineering, together with VR-supported collaboration, lays the foundation for the expansion of interactive methods in work system design in order to meet shortened product life cycles and customer requirements. VR technology has been increasingly utilised in planning smart production systems to visualise workstations or construct them with digital cardboard engineering [20, 21].

However, it should be noted that the requirements for effective collaboration and functional distribution of roles in virtual cardboard engineering show a high degree of complexity compared to physical interaction [21]. Working as a team in this virtual space is referred to as Collaborative Virtual Environment (CVE). Not only do CVEs allow spatially separated people to interact, but they also enable the interaction between those people and virtually simulated objects [22]. CVEs represent a framework for successful virtual collaboration, which must be detailed for the concrete consideration of virtual cardboard engineering with the help of defined scenarios. In classical cardboard engineering, direct, face-to-face communication is supported by whole-body movements, gestures, and joint interaction on a prototype, enabling the formation of a shared understanding of the subject of discussion. VR offers the possibility to partially realise this vital support of communication by implementing a realistic avatar [22]. Another influencing factor on collaboration is, among others, the simultaneous use of different media. Besides the possibility of all participants viewing the virtual environment through VR headsets, screen projections can visualise the process for non-VR users. As a result, possible scenarios of VR applications emerge.

There are four different workshop scenarios while using VR applications as a collaborative tool, each presenting a specific impact based on the focus of the task. In scenarios one and two, all participants are equipped with VR headsets; while participants in scenario one work together in one physical space, scenario two separates them into groups and allows them to work remotely. Participants in scenarios three and four are divided into VR users and non-VR users, while the non-VR users can only view the work system through transfer screening. The difference lies, again, in the physical space in which the workshop takes place [23]. The defined scenarios support interactive collaboration in VR so that the same results and discussion content can be achieved compared to a physical meeting of an interdisciplinary working group. However, it must be noted that the use of VR, other than being a collaborative tool, also impacts tasks and authorisations, which can be recorded with a role construction.

The participants’ roles are defined through task- and authorisation-oriented concepts. Task-oriented roles allow VR and non-VR users to take up different charges of the cardboard engineering workshop. On the other hand, authorisation-oriented roles restrict the functionality of VR applications for each user distinctively. Therefore, to ensure an efficient VR working experience for smart production systems and the future of new product integration, workshop requirements should be collected and included in a development process before conducting virtual cardboard engineering [23]. For example, motion-economic evaluations of the process based on the virtual execution of the work method require new competences with respect to the rule-compliant application of conventional evaluation methods and the transfer performance to a real process.

The interaction between role concepts and facets of collaboration can significantly impact media-supported interaction. In addition to using developed scenarios one of the most crucial factors to ensure the future success of cardboard engineering in the digital and virtual world is an extensive and consistent data basis as discussed at the beginning of the chapter.

2.3 The Emerging Role of Data Science and IoT in Engineering Smart Production Systems

The handling of data in its different forms and usages is one of the main challenges in Industry 4.0 concerning all technologies. The value-adding use of constantly increasing and available amounts of data is becoming a decisive competitive factor and is the basis for intelligent products, processes, and production technology. In mechanical and plant engineering and especially in the design of smart production systems, new roles and requirements emerge. Machine Learning (ML) as the basis of AI poses great challenges for companies, as the demand for experts, so-called Data Scientists, significantly exceeds the offer of available young talents. Furthermore, these experts rarely have the required domain knowledge – the core competences of manufacturing companies. In this context, the new job description of the Citizen Data Scientist as a link between the most important disciplines of information technology, domain knowledge, and data science arises. The shift to increasing software-heavy and data-analytics-driven production system design affects large companies, which often have access to talent pools and resources to build up expert departments, but also small and medium-sized enterprises, whose digital sovereignty is progressively threatened. The ML2KMU project is therefore developing an interdisciplinary role model for the implementation of ML initiatives in the manufacturing industry with a special focus on SME and mechanical and plant engineering as the core of production system design [24]. Building on an assessment of the basic competences required for the individual roles, the focus is on in-service competence development. For this purpose, a platform is being developed from which individual development measures for individual employees in project teams can be derived in a targeted manner. In addition to the methodological changes presented above, especially professional qualifications place new demands on future work. In addition to ML skills such as mastering statistics, data integration and pre-processing, training and evaluation of machine learning methods as well as deployment and maintenance of models (especially when smart production systems with data-based services are handed over to customers), the focus is increasingly on IoT skills. This requires employees who, in addition to the classic programming of shop floor IT such as NC, robot or PLC programming up to the integration of the SCADA and MES level, also have the skills to transfer smart production systems into IoT platforms. All in all, three system worlds with the same perspective are emerging: classic document-based planning using digital planning tools with PDM and PLM backbones will continue to exist but must be able to react more flexibly and quickly to rescheduling and also take into account digital components of later production systems. The second system world of automation technology is also becoming more important as manual work in later production systems is increasingly replaced by automation technology. In addition, there are new requirements for the interoperability of the shop floor components themselves and, in particular, an opening of the automation pyramid to enable intelligent data use, e.g., by using protocols such as OPC UA or MQTT. The third perspective includes IoT competences, as smart production systems must be mapped on uniform structures and platforms in the sense of the digital twin in order to enable data-driven services. Examples are Siemens Mindsphere and PTC Thingworx, but hyperscalers such as Amazon or Microsoft (as well as numerous other solution providers) also offer corresponding modules. Overall, the effort and complexity of the planning phase are significantly increased, but at the same time, new opportunities for intelligent data use arise in the operating phase of smart production systems.

3 Operating Smart Production Systems

New technologies open up the possibilities to manage the rising complexity in production systems and improve work for employees. As presented in the following, some technologies such as data usage are the same as in the planning phase although partly used with another focus. Other technologies such as robots, exoskeletons or assistant systems are linked more directly to operating production systems.

3.1 Connectivity and Internet of Things in Operating Smart Production Systems

The technological foundations for the development of smart production systems were created by the exponential increase in the performance of the available hardware in terms of memory, computing, and transmission capacity, described by Moore’s Law, with simultaneous miniaturisation of the hardware and falling prices. Supported by the flexibility of programmable software, this has led to the convergence of previously separate developments in the field of global digital networking and embedded systems based on programmable controllers, single-board PCs, and system-on-a-chip (SoC) [25, 26]. Already today, the far-reaching possibilities of global digital networks are combined with the potential of embedded software-based systems within modern production systems, creating the basis for mapping, monitoring, and controlling physical processes [26]. Embedded systems not only enable plants and components to communicate in an increasingly detailed and fast manner, but also products become capable of communication. In addition, cloud computing enables a scalable provision of computing and storage capacities in addition to comprehensive, platform-based networking of objects. This makes it possible, for example, to carry out computationally intensive analysis processes in the evaluation of large volumes of data that would not be feasible on the basis of embedded systems or internal company IT structures, or to provide data and services on demand [27]. The focus is on the Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS), and Software-as-a-Service (SaaS) operating and business models. With IaaS, only the basic resources such as network, storage space, and computing power are provided. The user has control over the operating system and the applications. PaaS starts one level higher and, in addition to the components mentioned under IaaS, provides middleware that, for example, balances the distribution of the load. This form is aimed at developers who need a platform to publish applications (e.g. ML-based services) without having to maintain the operating system. SaaS, as the closest view to the user, enables solution providers to provide services to end users without having to provide the technical infrastructure and without bearing the responsibility for installation and updates. In sum, smart production systems are emerging that can be understood as the further development of mechatronic systems. The underlying disciplines of mechanics, electronics, and computer science, therefore, play an increasingly important role not only in planning, but also in operating, maintaining, and improving smart production systems [28]. The physical design on the one hand and, on the other hand, the digital representations of production systems mapped using IoT platforms in Iaas, PaaS, and SaaS structures must be continuously maintained and serviced. In addition to planning, new competences are required, extending from classic automation and programming to IoT platforms, network and communication technology, and data science. Competence requirements for the use of digital twins are also emerging on the user side. This could be, for example, the configuration of individual dashboards, but also the configuration of services and the derivation of actions from digital recommendations.

3.2 Flexible Production Systems

One of the new possibilities of operating smart production systems is the demand-oriented and flexible provision of resources for production orders, summarised in concepts like Factory-as-a-Service (FaaS). To address smaller and more flexible batch sizes in emerging competition fields, smart factories strive to unite flexible manufacturing with high transparency of production processes and thus bring about new demands on communication between all actors at field, customer, and supplier level. The efficiency of this interaction is a key success factor for globally positioned production networks. Production markets increasingly require internet-based business platforms to compare capabilities of smart factories with customer specifications, creating a market not only for products but also for production processes. In this environment, modular production systems will be particularly successful, as they can easily adapt to a broad product portfolio and different production volumes. The research project PHASE shows possibilities to build such a platform, which is based on permanently up-to-date digital twins of the production systems and especially the capabilities and configurations of the respective components (Fig. 3). Building on the same foundations, the CSC research project focuses on the technical documentation of flexible, changeable production systems over the life cycle, which is both a planning and a maintenance task [29]. This is another example of new services that are only made possible by digitisation, since manual maintenance (as well as the reconciliation of production capacities in the previous example) quickly reaches its limits of flexibility and economic efficiency. With regard to the change in work, it can also be seen here that new digital competences are required in the creation, usage, and maintenance of digital twins and services in the context of smart production systems. On the other hand, they enable new options that were not possible with formerly physical manual work.

Fig. 3.
figure 3

Flexible Hybrid FaaS concept within the research project PHASE (own illustration).

Full size image

An example of production systems that provide such flexibility is the usage of cobots, small, light and affordable robots, which no longer need to be caged for safety reasons but can interact and collaborate with humans. The research project SUPPLY focuses on integrating cobots into flexible manpower lines. Depending on the required quantity, one or more cobots can be integrated into the assembly increasing the output without requiring additional staff. Especially short-term increases in demand can be covered more easily. To realise this potential, the products need to be designed to be handled by humans as well as cobots. Not all steps in assembly can be automated today, therefore humans remain an essential part of assembly systems. But increasing the steps cobots can be used for, especially repetitive or non-ergonomic ones increases the flexibility of production systems while relieving humans. However, the successful implementation of those systems requires technological advances as well as new qualifications. Beginning with the product design, designers need to take into consideration the capabilities of cobots, e.g. designing suitable grip surfaces as well as reducing flexible components [30]. Also, assembly system planners need to be qualified to integrate this new technology effectively into production systems. User-friendly simulation software, such as has been used in the research project KoMPI can enable non-robotic experts to plan and verify different scenarios. Simulation can also help to reduce the effort and time needed to teach the different scenarios to the physical cobot. Further development is to directly create robotic programs from simulation [31]. However, the specific language of every robot manufacturer poses a challenge. Another possibility to teach robots and especially cobots more easily is learning by demonstration [32]. The possibilities to teach robots by guiding them manually or using a trace pen are already implemented [33]. To directly capture the human movement in the production process and transfer it to the cobots is still a research focus [34]. The integration of cobots also changes the work on the shop floor. Human workers are working side by side without a separating fence. Whereas before, problems with an industrial robot could only be handled by robotic experts, now small problems will need to be handled by line workers to avoid standstill. More user-friendly control panels are facilitating this new task for line workers. Whereas teaching the robots today is handled by experts, the new teaching approaches open up the possibility to transfer this task to line workers while simultaneously reducing set-up times.

3.3 Worker Assistance Systems

However, not all non-ergonomic tasks in flexible production systems can be easily automated by using cobots. In such cases like e. g. non-stationary activities exoskeletons present a novel solution [35]. Exoskeletons are defined as wearable mechanical structures that relieve the musculoskeletal system during certain movements or postures [36]. Depending on the mode of operation, active and passive systems can be differentiated. Active exoskeletons have an energy source that supports or amplifies human movements. Passive systems on the other hand do not feature an external energy supply but use passive possibilities like springs to store and release energy [37].

An alternative classification of exoskeletons is based on the supported body regions. The most common systems on the market are trunk- and shoulder-arm support systems while full-body solutions are hardly commercially available [38]. Whereas exoskeletons have been widely investigated for applications in the military or healthcare, their usage in production environments has only recently gained increasing intention with the release of several specific systems for industrial use cases [39]. Due to this novelty, there are still a number of open questions concerning the application of exoskeletons. The research project SyNExo, therefore, addresses the systematic introduction of such systems in production and logistics. Initial results of an explorative interview study indicate that besides the associated costs, the technology acceptance of operators is currently an important adoption barrier. For example, exoskeletons may be rejected due to movement restrictions in secondary activities like walking or sitting at a computer desk [40]. Thus, in addition to investigating the long-term effects of exoskeletons further research is required on the multi-faceted construct of technology acceptance to fully unravel their potential to prevent musculoskeletal complaints.

While cobots and exoskeletons provide support regarding physical stress, operators in future smart production systems are also faced with highly dynamic environments leading to mental stress. The increasing number and fast change of product variants increase system complexity and cognitive demands for operators [41]. For that reason, it is not only necessary to consider physical loads but also to examine cognitive loads in the design of manufacturing processes [42]. Therefore, it is necessary to distinguish between two types of operator assistance systems, which are highlighted in Table 1 [43]. Energetic assistance systems like cobots or exoskeletons serve to reduce the physical stress on human workers by providing physical support when e.g. lifting loads. In contrast to this, the purpose of informational assistance systems is to lower mental stress by presenting necessary information according to the operators’ needs. Exemplary technological applications for this category are AR glasses, tablets or projection-based systems.

Table 1. Classification of operator assistance systems based on Bornewasser and Hinrichsen [43]
Full size table

One promising use case for informational assistance systems is operator training. Especially in manual assembly, product variety leads to more complex tasks, as many different assembly procedures have to be managed by operators [44]. Conventional training methods like demonstrations or paper-based instructions are not sufficient to ensure the necessary flexibility of workers [45, 46]. Informational assistance systems provide novel opportunities for technology-mediated learning by providing timely and context-specific information on assembly processes. Potential benefits of this approach have already been demonstrated in several studies. Assistance systems can reduce execution times especially at the beginning of training and help to maintain a higher process quality by preventing errors [47, 48]. Despite these promising results, the lack of knowledge about the wide range of available technologies as well as related high-potential use cases remain open issues. In order to promote the dissemination of assistance systems in industrial practice, methods for a structured selection of application scenarios and the user-centred implementation need to be developed [49]. The latter also includes suitable approaches for integrating product variants by linking them to the corresponding IT infrastructure [50].

4 Improving Smart Production Systems

As the past has shown, simply planning and then operating a production system does not lead to success. Especially the dissemination of lean management has stressed the importance of continuous improvement as a success factor [51]. The following sections show, how Industry 4.0 can support continuous improvement.

4.1 From Classic Improvement Methods and Roles to Digital Shop Floor Management

Methods from the field of lean management are increasingly merging with developments in Industry 4.0, as new possibilities for data collection and evaluation are emerging. The GaProSys 4.0 project, for example, addresses the development of a selection system for methods of a “Holistic Production System 4.0” for SMEs [52]. In the following, new possibilities in the widespread methods of shop floor management, bottleneck analysis, value stream analysis, and design are presented.

Classic shop floor management (SFM) addresses the economic operation of production systems and value streams as well as systematic continuous improvements by employees. The goals are to create an organisational framework that empowers employees to engage in continuous improvement processes (CIP) to improve and stabilise processes in manufacturing [53]. The core elements are on-site communication and leadership, transparency through visual management, and the promotion of a structured problem-solving process. While these tasks are conventionally carried out through shop floor management boards, which include (often manual) records of unit numbers and key figures of quality or productivity, they are not always carried out on the shop floor. Due to the increasing opening of shop floor IT up to the implementation of Industrial IoT platforms and digital twins, shop floor management can be supported digitally.

The advantages are time savings and error avoidance in data collection through connection to ERP/MES systems, possible use as a communication and information tool across departmental boundaries, and automation of write-ups. However, the challenges lie firstly in the development and maintenance of such a digital tool, and especially in its application. For the employees, this results in new requirements to set up, maintain and evaluate dashboards. However, if these competences are given, new possibilities arise for the flexible configuration of interesting visualisation forms for specific problems, e.g. for data-based error and error cause analysis.

On a system level above specific problem-solving in individual processes, value stream analysis (VSA) and value stream design (VSD) are widespread methods for visualising and modelling material and information flows and are used to analyse existing and design future value streams [54]. They help to map and improve material and information flows as well as value creation processes in a transparent and simple way. In the conventional form, in which value streams are mapped with pen and paper, they represent a snapshot of the current state. However, the method reaches its limits, especially in the case of a high number of variants and flexible process routes up to shared resources, as they occur in the development of CPS. Dynamic value stream analysis (DVSA), which enables a time-dynamic view, can help. An overview of the general methodological approaches of the DVSA and their implementation to date can be found at reference [55]. The DVSA makes it possible to monitor different product routes, cycle times, the real behaviour of inventories and lead times, and to analyse static dynamic and dynamic bottlenecks. It is often implemented at the MES level, but another trend also moves towards IoT platforms and the implementation of data-driven services. Figure 4 shows the implementation of DVSA services, including dynamic bottleneck analysis by IPSO Factory.

Fig. 4.
figure 4

Dynamic value stream and bottleneck analysis (IPSO Factory)

Full size image

For the employees, however, this also poses the challenge of commissioning and permanently operating the platforms and additional IT systems. This results in new digital competence requirements. On the one hand, new roles are emerging for the maintenance of productive shop floor IT systems. On the other hand, process experts are required to be able to operate and configure the systems. Production staff must also be qualified to use the systems, not only to interpret visualisations on the dashboard but also to create them in a customised way.

4.2 Industrial Data Science Changes the Possibilities and Requirements of Improvement Processes

Conventional CIP processes are based on the application of lean methods, e.g. the identification and elimination of waste. In quality management (QM), however, the requirements are more stringent, as it must be ensured that only good parts are delivered to customers. Therefore, data-driven approaches are already in use in this domain, especially those that address the identification of errors and error causes. The design of inspection and rejection strategies and the application of statistical methods from the field of Six Sigma have already led to production in the ppm range (parts per million) in many factories. On the one hand, the successes are based on very high inspection volumes and the rejection and reworking of defective parts; on the other hand, conventional statistical methods such as correlation analyses reach their limits with multivariate and often cross-departmental data sets. In particular, in quality management, the methodological toolkit of the product and process improvement team is therefore extended to include the application of machine learning (ML) as part of AI. The most common applications in QM are advanced fault diagnosis and root cause analysis based on large amounts of data as well as the prediction of quality metrics based on product and process parameters in order to reduce the load on inspection gates or to be able to proactively interact in process control. One example is the “QU4LITY” project. The goal of the EU project is the realisation of application-oriented and data-driven Zero Defect Manufacturing (ZDM). An essential aspect of the project is the reduction of defect costs of automatic process control in the production of electronic components by reliably anticipating the resulting product quality already during the production process. Based on the anticipated product quality, corrective actions are to be derived for the subsequent production process, so that the test efficiency is increased in the entire value stream. The latter objective requires the development and integration of virtual sensors into the production processes to ensure detailed and digital documentation.

In general, this results in new heterogeneous competence requirements for the workforce. Cooperation between domain experts, management, IT, and data science has established itself as a success factor for data science projects. The main challenge is that the corresponding competences must either be acquired externally or built up internally. The ML2KMU research project presents a role model in which competence profiles are defined. Furthermore, the AKKORD project offers a platform for identifying on-the-job data science training courses [56]. In addition to the development of technical framework conditions such as IT architectures and IoT or data science platforms, the development of new digital competences and the orchestration of specialist teams are of great importance in order to be able to realise the potential of digitalisation.

5 Conclusion and Outlook

As described above, Industry 4.0 is changing work in all areas of production, from production planning to operation and improving smart production systems. In production planning new consistent software tools as well as VR open possibilities for faster and more accurate planning e.g. using simulation and VR to test new systems intensively and improving them, before buying a physical system. New technology also allows working in interdisciplinary teams from different locations reducing travel time and cost as well as making experts better available. Nevertheless, to realise this potential, new ways to organise the work in an agile way are required. Employees need new skills to work with new technology but also need to learn to organise work differently. To address these changes new job descriptions arise. However, professional training and study programmes are only slowly changing, not yet able to meet the demand for experts. As described one of the most prominent examples are Data Scientists. Adequately using huge amounts of data and realizing the potential of Industry 4.0 is one of the most fundamental challenges. Competences in connection to working with data are often summarised as Data Literacy. Data Literacy not only includes knowledge needed by Data Scientists, but today a certain data understanding is required in all areas and career paths [57]. Equally, new competences are necessary concerning IoT, commissioning or operating platforms and networks.

However, not only work and competences in planning functions and jobs with higher education are changing but also working on the shop floor requires new skills. As described e.g. working alongside or even in collaboration with robots requires some basic knowledge of automation. On the one hand, new connectivity, smart machines, and higher automation facilitate work, on the other hand, new sources of error arise that need to be recognised and corrected. Additionally, the required flexibility and short reaction times are challenging. Informational assistance systems can help to manage this situation, but also require training the workforce in the correct usage. Due to the speed of the change in Industry 4.0 continuously and lifelong learning gains increasing importance. Knowledge acquired in formal education will be outdated within a few years forcing companies to new and continuous ways of learning.