Analysing Industry 4.0 technology-solution dependencies: a support framework for successful Industry 4.0 adoption in the product generation process

Technologies generally represent “a system of knowledge about how to manipulate nature in a logical scheme to achieve a functional transformation” (Betz 2011). Although the term technology thus describes broadly the ability and knowledge to achieve such a transformation (Gochermann 2022), often based on scientific insights, it is usually associated with hardware and software, such as additive manufacturing and big data (Nosalska et al. 2020). Purposefully combined, compatible technologies provide new functions and enable the adoption of i4.0 in terms of i4.0 tech-solutions (Kang et al. 2016).

In addition to the continuously growing variety of technologies, a key challenge of systematically managing technologies is a varying degree of abstraction in literature, e.g. ranging from specific, such as deep learning, to abstract, such as general artificial intelligence. A clear taxonomy is key to ensure a unified understanding of what is actually described, managed or applied. To manage this complexity, such a taxonomy needs to be able to incorporate and clearly structure technologies on different levels of abstraction to differentiate between single technologies and groups of technologies. Due to the fast-paced technological development, another requirement is to allow for future updates, i.e. adding, deleting, and modifying technologies and groups. This includes a robust mapping with i4.0 solutions to allow for adding, deleting, and modifying technologies, groups, and tech-solutions without affecting the rest of the mapping framework. A range of existing studies have categorised i4.0 technologies in different ways and with different foci. This paper builds on this wealth of prior knowledge and combines these insights in a taxonomy that fulfils the requirements of allowing varying levels of abstraction, future updates, and a robust mapping of i4.0 technologies with i4.0 solutions.

Bueno et al. (2020) use a rough five-item technology categorisation of internet of things (IoT), cyber-physical systems (CPS), additive manufacturing (AM), cloud manufacturing, and big data and analytics/AI, and map these onto different capabilities and applications for production scheduling and control. This provides a systematic overview of technology potentials but lacks a detailed technology categorisation and consideration of other product generation phases. Zheng et al. (2021) provide a broader view, both in terms of considering different product lifecycle phases and presenting a more detailed analysis of 10 technology types, which adds to the five categories of Bueno et al. (2020): IoT, CPS, AM, cloud technologies, big data and analytics, AI, blockchain, simulation and modelling, visualisation technologies, and automation and industrial robots. Dalmarco et al. (2019) describe nine technology groups: simulation, big data and analytics, cloud computing, cyber-physical systems, cybersecurity, collaborative robotics, augmented reality, additive manufacturing, and systems integration. Although they explain that these technologies are linked to developing, manufacturing, operating and servicing activities, the specific links to the activities and to each other are not clear. In this respect, the framework of Lopez-Gomez et al. (2017) analyses interdependencies of technologies. It uses a data perspective to structure i4.0 technologies into four groups including subgroups: data generation and capture (incl. CPS), data transmission (incl. network infrastructure), data conditioning, storage and processing (incl. big data and cloud computing), and data application (incl. advanced manufacturing capabilities). However, the framework does only mention a limited number of technologies for each group. Lee et al. (2015) also use a data perspective to build their hierarchical 5C CPS architecture: smart connection level (e.g. sensors), data-to-information conversation level (e.g. smart machine components), cyber level (e.g. digital twins), cognition level (e.g. collaborative diagnostics and decision making), and configuration level (e.g. self-optimising systems). Although the hierarchical levels building on each other are a powerful concept, they focus on a production system as ultimate goal and put less focus on smart individual machines or product development phases. Calabrese et al. (2020) analyse the relationship between i4.0 goals, technologies, and barriers, based on a structure of nine technology classes reflecting different i4.0 use case areas: production line, smart worker, smart equipment, computing, sharing, smart product, data analytics, network, and cyber-security technologies.

The comparison of these framework shows distinctive similarities, where Bueno et al. (2020) can be seen as a subset of Zheng et al. (2021). In respect of a consolidated i4.0 technology taxonomy, a central category are manufacturing technologies, which enhance manufacturing capabilities and their level of automation and autonomy, and can be more generally addressed as production line (Calabrese et al. 2020), abstractly as data application (Lopez-Gomez et al. 2017), or more specifically as AM or robots (Dalmarco et al. 2019; Zheng et al. 2021). Computing technologies (Calabrese et al. 2020) correspond with data conditioning, storage and processing (Lopez-Gomez et al. 2017) and include cloud technologies (Dalmarco et al. 2019; Zheng et al. 2021). They enable the sharing and provision of computing power and resources across business processes. They also improve the exchange and processing of data across company boundaries. While data technologies also overlap with data conditioning, storage and processing, their focus is more on data analytics (Calabrese et al. 2020; Dalmarco et al. 2019), including big data, AI and blockchain (Zheng et al. 2021). They enable advanced analysis, management and secure storage of data that forms the heart of i4.0 (Inkermann et al. 2019). Closely linked are simulation technologies (Dalmarco et al. 2019), which also include modelling (Zheng et al. 2021) and computing, and represent a form of data application (Lopez-Gomez et al. 2017). They combine technologies for the simulation and the optimization of business activities, ranging from technologies like ERP, MES and CIM/CAx novel technologies like advanced simulation for predictive control and problem solving (Modrák and Mandul’ák 2009). While human interface technologies are not explicitly addressed, they correlate with smart worker, smart product and smart equipment (Calabrese et al. 2020) in respect to data generation and capture and data application (Lopez-Gomez et al. 2017), including visualisation (Zheng et al. 2021). They can range from partially immersive augmented, where real-world elements are combined with the virtual reality, to fully immersive ways to visualise data and images. Network technologies combine sharing and network (Calabrese et al. 2020) and IoT (Zheng et al. 2021) in the sense of data transmission (Lopez-Gomez et al. 2017) as well as integration (Dalmarco et al. 2019). They enable the i4.0-inbuilt networking of machines and humans in real time (Bauernhansl et al. 2014; Lee et al. 2015) and show strong links to cloud-based computing technologies. Sensor technologies are not explicitly mentioned but are critical is respect to data generation and capture (Lopez-Gomez et al. 2017) as they connect the physical and digital side of i4.0 through capturing real-world data (Szász et al. 2021). While this enables smart i4.0 systems, these novel sensors are smart themselves as they are autonomous interconnected systems. An increasingly crucial category but only mentioned by Calabrese et al. (2020) and Dalmarco et al. (2019) are cyber security technologies. They protect hardware and software i4.0 infrastructure from unauthorised access, tempering or attacks (Bajic et al. 2021). Despite their relevance, the analysed literature surprisingly did not state specific technologies. Figure 1 provides an overview of the consolidated i4.0 technology taxonomy (with more details in Appendices 1 and 2). Its three-level structure allows to allocate technologies on different levels of abstraction and differentiate specific technologies from technology groups. It also supports its time-robustness as new technologies can be added to existing sub-categories. As cyber-physical systems (CPS) combine several technology categories, they cannot be mapped to a single category and are, therefore, not mentioned in the taxonomy.

Technology-solutions (tech-solutions) are “an implementation of people, processes, information and technologies in a distinct system to support a set of business or technical capabilities that solve one or more business problems” (Gartner 2022). I4.0 tech-solutions are the combination and realization of different technologies within a concrete i4.0 adoption scenario, i.e. a specific application and purpose, such as combining additive manufacturing with artificial intelligence to enable optimised material selection for 3D printing (Pedota et al. 2023). A set of technologies can be combined to various tech-solutions based on different adoption purposes and strategies. The use of i4.0 tech-solutions represents the adoption of i4.0.

Following the idea of i3.0/CIM from CAD design to produced product (Marri et al. 1998), this research uses a holistic lifecycle phase-spanning view on i4.0 tech-solutions across the product generation process, which can be structured into four phases (Fig. 2, and Table 2) or groups based on Naefe and Luderich (2016) and Schneider et al. (2020).

The DMM uncovers a varying use of technologies across the product generation process. While production process tech-solutions show a broad use of technologies, production planning and product development focus on Data and Simulation. The focus of product planning lies on Data and Network technologies. This reflects the characteristics of each phase along with the level of concretisation and maturity of the associated product.

It also reflects the origin of i4.0 as a manufacturing concept and key R&D area. Although a holistic life-cycle-spanning consideration of i4.0 would be important and initial scientific signs in this direction can be observed (e.g. Arromba et al. 2021; Ramírez-Durán et al. 2021), a broad tech-solution of i4.0 beyond manufacturing has been limited to date. This is also evident in our DMM analysis although it has to be treated with care as we only analysed review articles. Due to the retrospective character of reviews, this might cause a time delay of reported i4.0 tech-solutions. Still, this does not affect the general pattern of i4.0 technologies being under-utilised in non-manufacturing related product generation process phases. This represents a major gap as opportunities and synergies to other technology-and-innovation-management approaches are not exploited. For example, i4.0 and open innovation show strong similarities in respect to underlying IT technologies, networked features and organisational openness, which would be beneficial to build integrated industry ecosystems if combined.

The sum of links of each DMM column indicates how frequently specific i4.0 technologies are used in i4.0 tech-solutions. Three groups—low, medium and high frequency of use—can be differentiated, using the maximum value of 39 to derive three evenly spaced groups. Knowing the groups can support companies in their investment decisions: high frequency can indicate mature technologies with limited adoption risks (cf. Núñez-Merino et al. 2020), which might also be central enabling technologies (cf. Pedota et al. 2023), i.e. representing future-proof investments that are applicable for a large variety of tech-solutions.

In general, technologies with a high frequency can be an indicator of potential i4.0 “base” or “platform” technologies, i.e. technologies that are fundamental prerequisites for any i4.0 tech-solution. Besides CIM/CAX systems and AI, these might include medium frequency technologies such as big data analytics and industrial networks. However, some analysis results are counter-intuitive and need further investigation in the future: this includes the low frequency of social networks, which have great potential for customer or user engagement in the sense of open innovation in the product development phase as well as maintenance or after sales. But this might be due to the focus on manufacturing of i4.0. Another example is mobile internet, which has the potential of a broad application—especially in the context of 5G-eabled fast external and in-factory communication. This might be due to the analysis focus on links stated in literature. It is also important to point out that no specific security technologies were mentioned in the analysed literature despite their critical role for i4.0 (Fuller et al. 2020). Future research needs to explore the underlying reasons. These might include an under-reporting as cyber security technologies could be generic or not i4.0-specific. Given that effective cyber security not only requires technologies but also processes, mechanisms and trainings, a broader analysis scope could be beneficial as well.

The DMM also allows insights into the relative complexity of i4.0 tech-solutions through its row sums. A higher number of required technologies per tech-solution can indicate the associated efforts, such as integrating or managing and maintaining different technology types. Aside from the absolute number of links, their distribution across different technology groups is also relevant. While technologies from one group, such as AI and big data analytics, show large similarities, using technologies from different groups can increase integration efforts and also affect individual innovation cycles, e.g. varying software vs. hardware lifecycles.

Deriving three evenly spaced and whole-numbered groups based on a maximum value of 11, the DMM shows three complexity groups. These involve technologies from at least two technology groups but with no tech-solution using technologies from all seven technology groups:

In general, the DMM row sums and relative complexities provide an indication of the resulting R&D and implementation efforts, and costs of i4.0 tech-solutions. Highlighting the required technologies for different i4.0 tech-solutions increases the transparency of required preconditions. This supports a well-informed investment decision including a detailed assessment of adoption benefits and risks of specific i4.0 tech-solutions.

The i4.0 tech-solution-technology DMM also allows initial insights into dependencies between technologies themselves. Looking at DMM rows of different tech-solutions, patterns of technologies become visible which tend to occur together.

A prominent example is augmented reality, which typically occur in combination with data, (advanced) sensors and simulation. This is reasonable as augmented reality requires AI for sophisticated models and simulations that are projected onto a real-world environment using specific interface technologies like headsets (Alcácer and Cruz-Machado 2019). Similarly, additive manufacturing systems occur in combination with CIM/CAx as such digital models are a prerequisite for the actual manufacturing step (Alcácer and Cruz-Machado 2019). Another example are cobots in combination with advanced sensors to detect humans and objects, or advanced simulation based on CIM/CAx models and AI to enable autonomous and safe collaboration with humans and other machines via industrial networks (Kamble et al. 2018).

A special case are digital twins. These are essential for cyber-physical systems, which are described as a key element of i4.0. Digital twins are sometimes seen as a technology and often used in an ambiguous way as an umbrella term (Murray et al. 2019; Schleich et al. 2017) for what Fuller et al. (2020) differentiate into digital model, digital shadow and actual digital twin. Thus, we do not consider them as a single technology but rather as a group of tech-solutions, which build on use case-specific combinations of technologies, such as computing, data analytics and simulation technologies (Liu et al. 2021).

Although these links between technologies already provide valuable insights into which technologies could or should be combined, enabling and hierarchical dependencies have to be analysed in a next step. This will allow for identifying technology chains and networks and support answering questions like which technologies are required to enable a specific technology. This will enable more technological transparency and contribute to a better understanding and managing of complexity in the context of i4.0 adoption.

To answer the question of how the i4.0 tech-solution-technology framework could be used, it was applied to an existing cyber-physical system (CSP), an i4.0 nano-brewery at the University of Technology Sydney in Sydney, Australia (Fig. 3) to conceptually verify it as part of the DRM’s Prescriptive Study (Blessing and Chakrabarti 2009). The brewery’s autonomous process management ensures constant output quality as well as process flexibility and parameter prediction to quickly prototype new recipes. Additional aspects include the upscaling of recipes from lab to industrial scale and the exchange of recipes across global production sites tackling concept drift etc. To address the latter, a partner research institute at the <will be added for accepted publication> possesses an identical physical twin brewery system.

Although the DMM does not specifically list cyber-physical breweries yet, the listed “smart machining” (S4.21) fits well due to its focus on manufacturing and smart processes. “Smart factories” (S4.20) are less suitable due to their focus on an entire socio-technical factory system. The DMM allows the following insights:

The sub-category of Autonomous Equipment is a key aspect for the brewery, which needs to autonomously react to changing ingredients quality like water and malt, and environmental factors such as temperature or humidity, in order to adjust process parameters and ensure a constant beer quality. Other manufacturing technologies like cobots for material handling might be added later but are not part of the core system. An important basis of the brewery is Cloud Computing, in this case, using Amazon Web Services to run RapidMiner AI Hub to analyse process data and build machine learning models to enable an autonomous brewery. Cloud Manufacturing is relevant to share beer recipes between the two global physical twin brewery systems. Although the basis for this is the analysis of big amounts of data, the brewery uses Artificial Intelligence as it does not only include descriptive and diagnostic but also predictive and prescriptive factors. Similar to cobots, advanced human interfaces could be beneficial in the future but are not core element or requirement for the current brewery tech-solution. Due to its encapsulated setup, the brewery does not interact with other manufacturing systems. Thus, Industrial Networks do not apply in this case—despite the DMM entry. Instead, IoT middleware is essential to connect the digital twin components with the underlying different hardware parts of the brewery. This includes actuators, such as valves and heating elements, and traditional sensors, such as temperature and flow sensors, as well as Advanced Sensors, such as electronic nose sensors to assess ingredients and brew quality. However, Computer Integrated Manufacturing along with Factory Management technologies play a minor role for this research brewery—despite the DMM entry—which is mainly due to it being a lab and not an industry brewery line. Nevertheless, they will become important when moving from a lab to an industrial production. This includes upscaling of production volumes and careful management of supplies and logistics to ensure a reliable and economic production.

In respect to the resulting investment decisions when planning such a system, the high column sums of both associated technology sub-categories (Table 2) indicate that this would be a beneficial and future-oriented investment if a company does not have existing CIM or ERP systems. In general, the high row sum indicates a high level of technological complexity of the CPS, which is also reflected in the large variety of required enabling technologies.

From a product development perspective, the DMM can provide guidance around what technology groups will be relevant when designing and developing a new tech-solution. Even though the listed tech-solutions might differ slightly to the one to be developed, it can highlight what might be important, allowing the R&D team to discuss if and how those technologies apply to the new tech-solution. However, the brewery example also showed that some of the listed tech-solutions are more specific than others. Therefore, future research should also investigate how a sub-structure of tech-solution clusters could find a suitable level of abstraction to balance general applicability and level of detail.

In summary, the retrospective application of the i4.0 tech-solution-technology framework to this brewery confirms its conceptual plausibility and usability. It helps validating and sense-checking specific system design decisions of the brewery. In addition, it also indicates aspects that are currently not considered and can therefore trigger a systematic reflection and discussion if and why these aspects might be relevant—right now or in the future.