Enabling industrial internet of things-based digital servitization in smart production logistics

ABSTRACT

Digital servitization (DS) enabled by the Industrial Internet of Things (IIoT) is essential for securing long-term competitiveness in manufacturing. The literature identifies the need for developing data models for multichannel communications across IIoT devices that fulfil the DS vision. This is crucial for avoiding isolated systems based on proprietary solutions and for promoting data sharing and interoperability across existing and future DS applications. Accordingly, this study proposes a data model for multichannel communication that facilitates IIoT-enabled DS for smart production logistics (SPL). We present three findings from a case study focussing on material handling in a manufacturing company. First, this study provides a model with four modelling profiles, including IIoT devices, databases, and services for multichannel communication. Second, it shows how the proposed modelling profiles transfer information across monitoring, control, optimisation, and autonomous decision services. Third, it presents the operational benefits of applying the proposed data models, including improvements in the delivery, makespan, and energy of material handling. These findings are essential for capturing, processing, and transferring information across products, services, and software databases in IIoT-enabled DS for SPL. They are relevant for manufacturing managers and academics and improve our understanding of IIoT-enabled DS's deployment for SPL in manufacturing.

KEYWORDS:

• IIoT-enabled digital servitization
• smart production logistics
• systems modelling language
• case study
• material handling

1. Introduction

Digital servitization (DS) enabled by the Industrial Internet of Things (IIoT) is essential for providing customised value propositions based on higher-quality services and securing long-term competitiveness in manufacturing (Hsuan, Jovanovic, and Clemente 2021). DS refers to the application of digital technologies to transform the business models of manufacturing companies from product- to service-centered ones (Tronvoll et al. 2020; Grubic and Jennions 2018). DS presents distinct benefits to manufacturing companies, including new market entry opportunities, higher scalability at marginal cost, and closer customer interaction (Sklyar et al. 2019). Against this backdrop, the literature identifies IIoT as indispensable for realising the critical features of DS. Specifically, manufacturing companies developing DS must acquire and use digital data streams from multiple sensors to monitor, control, optimise, and make autonomous decisions (Kohtamäki et al. 2019). Accordingly, IIoT provides the connection of all industrial assets, including machines and control systems, with the information systems and business processes essential for understanding the flow of material and information within factories (Jovanovic, Sjödin, and Parida 2021).

Smart production logistics (SPL) is an area of increasing interest in IIoT-enabled DS. SPL refers to applying IIoT for achieving the active perception, response, and autonomous decisions in production logistics (Y. Zhang et al. 2018). Its purpose is securing increased visibility, control, adaptation, and prognosis in the movement of materials and information (Su and Fan 2020; Guo et al. 2021). Accordingly, SPL includes all activities focussed on the transfer of materials and information within the physical limits of a factory (Tu, Lim, and Yang 2018; Qu et al. 2016). SPL is distinct from but contained by efforts for applying IIoT in a supply chain involving all internal and network-wide material, part, and product flows along the complete value-added chain (Song et al. 2021). Achieving IIoT-enabled DS in SPL is essential for the manufacturing industry for two reasons. First, acquiring and using data enabled by IIoT across products, services, and software ensure the reliable delivery of materials and information to customers (Rabetino et al. 2018). Second, the services include monitoring, control, optimisation, and autonomy of material and information flow (Mosch, Schweikl, and Obermaier 2021). These services are indispensable for increasing visibility, assessing operational risks, and responding to customer needs (Naik et al. 2020).

Realizing IIoT-enabled DS in SPL remains challenging despite recent efforts. The literature identifies three features that make IIoT-enabled DS of SPL challenging compared with that of other fields. First, the increased need for SPL for services providing real-time visibility crucial for planning and controlling spatially and temporally distributed activities (Winkelhaus and Grosse 2020; Trappey et al. 2017). Second, realising the compatibility of IIoT devices with existing infrastructure ensuring a long-term solution (Tran-Dang et al. 2020). Third, managing a large variety of protocols, data formats, and physical sensing resources in SPL operation (Song et al. 2021). In particular, the literature identifies the need for developing data models facilitating communication across multiple IIoT devices (Rymaszewska, Helo, and Gunasekaran 2017). This is crucial for avoiding isolated systems based on proprietary solutions and for promoting data sharing and interoperability across existing and future applications in DS (Sisinni et al. 2018). Pan et al. (2020, 2021) propose data models for the synchronisation of real-time production logistics as a first step in this direction. Additionally, Şahinel et al. (2021) introduce an ontology-based data model including IIoT middleware. C. Liu et al. (2019) present an Open platform communications united architecture (OPC UA) data model to standardise and communicate data across machines and applications. Jiang et al. (2021) design an interconnection scheme linking logical, geometric, and data models to optimise a manufacturing system. These studies constitute an important contribution to establishing the communication of IIoT devices. However, these studies do not consider the critical aspect of DS. Namely, proposing data models supporting communications between multiple IIoT devices and products, services, and software enables the transition from product- to service-oriented manufacturing (Kohtamäki et al. 2019). This is crucial for collecting logistics data across diverse IIoT devices, monitoring status, detecting early warnings, optimising resources dynamically, and controlling and executing autonomous responses to unexpected events for service-oriented manufacturers (Chen et al. 2021; Ding et al. 2020; Trappey et al. 2017).

To meet this need, this study proposes a data model for multichannel communication that facilitates IIoT-enabled DS in SPL in the manufacturing industry. We apply the principles of model-driven engineering (MDE) and develop data models for multichannel communication in Systems modelling Language (SysML). This choice is informed by the need to describe the data structure, behaviour, requirements, and constraints for IIoT-enabled DS in SPL. This study focuses on material handling in SPL because of the extensive research efforts applying IIoT in this field and its significance for increasing the competitiveness of manufacturing (Winkelhaus and Grosse 2020; Cao, Li, and Wang 2019; Granlund and Wiktorsson 2014).

The study makes three contributions to the literature. First, it presents three modelling profiles, including IIoT devices, databases, and services for multichannel communication. This is essential for capturing, processing, and transferring information across products, services, and software to realise IIoT-enabled DS in SPL. Second, the proposed modelling profiles are applied to transfer information across monitoring, control, optimisation, and autonomy services in the execution of material handling tasks by automated guided vehicles (AGVs). Third, the study outlines the operational benefits of applying the proposed data model through a case study of the Swedish automotive industry, including improvements in the delivery, makespan, and energy consumption of material handling. Finally, it presents managerial implications critical for manufacturing companies, including guidelines for the deployment of IIoT-enabled DS in SPL.

The remainder of this paper is organised as follows. Section 2 presents related studies. Section 3 proposes a model for multichannel communication and its corresponding architecture. Section 4 presents our empirical results. Section 5 discusses the implications of this study, and Section 6 concludes.

2. Related studies

This section describes the current understanding of DS, IIoT in SPL, and existing architectures and data models for multichannel communication in SPL.

2.1. Digital servitization

DS constitutes a novel paradigm in manufacturing that differs from the manufacturing of products or the development of digital services that support a product (Chen et al. 2021; Tronvoll et al. 2020; Pirola et al. 2020). DS includes digitalisation of the manufacturer's entire service business model (Ritter and Pedersen 2020) and proposes an outcome provision where solution providers sell outcomes instead of services (Visnjic, Neely, and Jovanovic 2018). A critical feature of DS is to support customers in achieving their goals, instead of supporting the products themselves (Naik et al. 2020). Recent studies identify five aspects in which DS differs considerably from product service offerings in manufacturing (Mosch, Schweikl, and Obermaier 2021). First, there is a higher scalability of digital services (Hasselblatt et al. 2018). Second, physical products are replaced with digital services that facilitate outcome-based contracts (Vendrell-Herrero et al. 2017). Third, there is a need for data-driven platforms that manage decisions and strategic actions (Sklyar et al. 2019). Fourth, it facilitates cooperation between stakeholders (Benitez, Ayala, and Frank 2020). Fifth, novel capabilities lead to changes in power structures in supply chains (Coreynen, Matthyssens, and Van Bockhaven 2017; Porter and Heppelmann 2014).

The literature establishes two essential characteristics for realising DS in manufacturing. First, a transition from product- to service-oriented manufacturing, where IIoT enables integrated and mutually supportive products, services, and software (Kohtamäki et al. 2019). Accordingly, products in DS require IIoT connectivity to exchange data with other products or services offered to manufacturers, operators, and customers (Frank et al. 2019). Services refer to digital and ubiquitous activities that are essentially intangible, perishable, and include a well-defined functionality that enhances the value proposition supporting customer goals (e.g.availability, reliability, or performance) (Naik et al. 2020). Software is a set of systems with the generative capacity to produce unforeseen possibilities for creating value originating from data (Autio et al. 2018; Kohtamäki et al. 2019), and transforming manufacturing to support new business models (Saadatmand, Lindgren, and Schultze 2019; Rajala et al. 2019; Hsuan, Jovanovic, and Clemente 2021). Second, the literature on DS underscores the importance of developing control, optimisation, and autonomy services that increase value to customers (Porter and Heppelmann 2014). Monitoring refers to the capability of products to provide real-time information about their location and status (Ardolino et al. 2018). Control concerns the use of embedded software to manipulate resources remotely (Cimini et al. 2021). Optimisation involves the use of digital technologies to enhance the operational performance of equipment on the factory floor (Lerch and Gotsch 2015). Autonomy includes the self-diagnosis, coordination, and autonomous operations of equipment (Qu et al. 2016).

Recent studies provide critical insights regarding IIoT-enabled DS in SPL. For example, Chen et al. (2021) investigate the process of DS as manufacturing companies making a shift from provision of standard products and services to smart solutions. This study highlights the importance of DS for collecting and monitoring data in logistics, which is essential for facilitating early warning and rapid response to unexpected events in logistics tasks. Accordingly, problems that surface in SPL (e.g. warehousing) may be identified quickly, and suppliers can be contacted directly for an immediate solution. Kohtamäki et al. (2019) discuss DS business models and underscore the importance of delivery services for logistics that lead to increased visibility and reduced transaction costs. Zheng et al. (2019) and Pirola et al. (2020) analyse the importance of mutually supporting DS products, services, and software to ensure the reliability and correct functioning of the flow of materials and information. Finally, Struyf et al. (2021) posit that DS is critical for increasing the visibility of material and information flow in a factory and developing relational intimacy and informational openness between manufacturers and customers, which is essential for increased agility.

2.2. Industrial internet of things in smart production logistics

SPL is a term distinct from but related to additional ones frequently found in the literature including smart logistics and IIoT in logistics. Smart logistics involves applying digital technologies including IIoT, identification technologies, artificial intelligence, and cyber-physical systems (Su and Fan 2020). The purpose of smart logistics includes increasing visibility, controllability, predictions, and adaptation to changes in the flow of materials or information (Ding et al. 2020). Importantly, the scope of smart logistics covers logistics activities throughout a supply chain and beyond the physical limits of a factory (e.g. freight transportation, warehousing, customer service, planning, and delivery; Barreto, Amaral, and Pereira 2017). Therefore, SPL is distinct from but contained within smart logistics, including its efforts in applying IIoT in a supply chain and involving all internal and network-wide material, parts, and product flows as a part of the complete value-added chain (Song et al. 2021). IIoT in logistics refers to the use of a global Internet-based information architecture for industrial systems facilitating real-time information including multiple levels and dimensions (Y. Zhang, Zhu, and Lv 2018; Qu et al. 2016). IIoT in logistics is the foundation for numerous applications to achieve active perception, response, and autonomous decisions (Winkelhaus and Grosse 2020; Tu, Lim, and Yang 2018) and is one of the digital technologies that is critical for SPL.

The literature presents IIoT as a promising technology for Industry 4.0 (Xu, He, and Li 2014), offering the opportunity to build SPL and services by leveraging the growing ubiquity of IIoT devices (Sisinni et al. 2018; Y. Zhang et al. 2018). However, IIoT-enabled DS in SPL still faces challenges in current practices. Specifically, the DS of SPL relies on mutual support among three elements: products, services, and software (Paschou et al. 2020; Hsuan, Jovanovic, and Clemente 2021).

Numerous research efforts have been made to apply IIoT devices to products (Mörth et al. 2020; Ding et al. 2020). Various approaches to facilitate real-time data acquisition of manufacturing resources, production logistics process control and monitoring, and decision-making during material handling have been presented (Y. Zhang et al. 2015; Tu, Lim, and Yang 2018; Qu et al. 2016; S. Liu et al. 2020a). Real-time data collection and sharing of production logistics resources is a fundamental element for the interaction of physical parts (products/devices) and cyber parts (services/software) (S. Liu et al. 2020b; Zafarzadeh et al. 2019). Within this context, the configuration of IIoT-enabled devices includes a production logistics system with the capabilities of active perception, interaction, and communication (S. Liu, Zhang, and Wang 2018). In parallel, advances in emerging communication technologies facilitate efficient data transmission and sharing among production logistics resources, where a communication channel between physical devices and digital modules is deployed (Khan et al. 2020).

Data availability has, therefore, allowed the adoption of IIoT-enabled industrial applications for efficient and flexible production logistics operations (Huang et al. 2019). Some examples are a cost-effective IIoT solution based on the dynamic analysis of the production logistics shop floor (Qu et al. 2017) and a smart logistics vehicle enabled by the IIoT to allow real-time interaction with logistics tasks (S. Liu et al. 2019). In addition, DS in the production logistics system aims to transform the product-centric business model to a service-centric model (Tronvoll et al. 2020). Therefore, there has been an increasing interest in the field of IIoT-driven services for high-quality production logistics (Winkelhaus and Grosse 2020; Jeong, Flores-Garcia, and Wiktorsson 2020).

Services originating from IIoT devices are often associated with data value-added services, optimisation, autonomous functions, and control execution. Data value-added services focus mainly on data processing, data modelling, and data management by deploying advanced analytic tools (Zafarzadeh et al. 2019). In the last decade, many studies have focussed on the optimisation of production logistics systems, especially for logistics tasks and production resources. Further, various approaches and frameworks that focus on scheduling optimisation, task allocation, and logistics planning have been developed (Winkelhaus and Grosse 2020). In parallel, a comprehensive study of optimisation algorithms that fulfil the need for optimal solutions is discussed in Lohmer and Lasch (2021), which covers traditional methods, heuristics/meta-heuristics algorithms, and artificial intelligence-based approaches. The deployment of digital technologies in IIoT-based solutions provides control of the execution of tasks and production operations. However, the deployment of digital services in production logistics systems also needs to be supported by software modules (Kohtamäki et al. 2019). Therefore, the combination of products, services, and software within IIoT-enabled solutions is a promising tool for SPL.

2.3. Data model for multichannel communication

Data models are crucial for connecting physical objects, sensors, and services enabled by the IIoT. Data models are responsible for sensing, processing, transmission, storage, and analysis of data enabled by IIoT in manufacturing (Cecil, Albuhamood, and Cecil-Xavier 2018). Moreover, recent studies show that data models in IIoT-enabled manufacturing are essential for describing information about manufacturing processes, production plans, historical occurrences, and simulations (Jiang et al. 2021). Importantly, the literature posits that developing data models for IIoT-enabled manufacturing requires a systems understanding of factories, including properties, functionalities, hardware descriptions, runtime environment of the devices, and communication interfaces (Şahinel et al. 2021).

The literature provides diverse modelling specifications and standards that support the development of data models for IIoT-enabled manufacturing. Examples include Web Ontology Language (OWL), AutomationML, Asset Administration Shell (AAS), and SysML. OWL is a standard language describing concepts, relationships between entities, and categories including classes, instances, properties, and relations (Hitzler, Krötzsch, and Rudolph 2010). OWL is well suited for complicated knowledge modelling, allowing logic inference based on predefined reasoning rules, which makes it possible to acquire implicit knowledge (Shi et al. 2018). For example, Şahinel et al. (2021) apply OWL to develop a data model of human agents for IIoT middleware, enabling the use of information in cyber physical systems (CPS). Bao et al. (2020) apply OWL in an ontology-based modelling method for a digital twin in an aircraft assembly, focussing on the reduction of data entries. AutomationML is an open XML-based format aimed at interconnectivity and data exchange between different tools in manufacturing (Schmetz et al. 2020). It integrates different task-specific formats and ontologies, allowing the modelling, storage, and exchange of engineering models (Schmetz et al. 2020; Schroeder et al. 2016). H. Zhang, Yan, and Wen (2020) apply AutomoationML to model information and integrate physical resources in a digital twin. Monostori et al. (2016) combine AutomationML and OPC UA for modelling factories and their components together with skills, topology, interfaces with and relations to others, geometry, kinematics, and even logic and behaviour of CPS. AAS describes an asset electronically in a standardised manner with the purpose of exchanging data across devices, production orchestration systems, or engineering tools (Chilwant and Kulkarni 2019). Platenius-Mohr et al. (2020) apply AAS for information exchange between a real-world application in an industrial context and a digital twin. Park, Son, and Noh (2020) apply AAS to develop a multilevel reference model describing physical objects and their functionalities in a CPS. Subsequently Arm et al. (2021) apply AAS to establish communication across machines and negotiate the priorities of production according to pre-specified rules in two assembly lines.

This study applies SysML to propose a data model for multichannel communication facilitating IIoT-enabled DS for SPL. SysML is a general-purpose modelling language for MDE that provides structural, behavioural, and parametric semantics for analysing systems (Holt and Perry 2018). There are three reasons for the choice of SysML. First, there is a pressing need to achieve interconnectivity across products, services, and software to achieve service-oriented manufacturing (Hsuan, Jovanovic, and Clemente 2021). This need is combined with the realisation that manufacturing systems were not developed to do so (Tronvoll et al. 2020) especially with the emergence of SysML based on MDE principles for addressing changes in the interconnectivity of systems involving hardware and equipment, software, data, people, and facilities (Friedenthal, Moore, and Steiner 2015). Second, a standardised language is applied to capture the structure, behaviour, requirements, and constraints of information flow (Candell et al. 2019). Third, SysML facilitates visualisation and communication of information flow across stakeholders, which is essential for developing end-to-end services in a DS context (Ferraris, Fernandez-Gago, and Lopez 2020).

2.4. Identified research gaps and research questions

The literature increasingly acknowledges the importance of IIoT-enabled DS for SPL. Studies posit that achieving IIoT-enabled DS in SPL is critical for realising new processes and capabilities, improving value creation and capture, increasing customisation efficiency, and securing order delivery (e.g. right quantity, cost, time, place, condition, customer, and product) (Ardolino et al. 2018; Kohtamäki et al. 2019; Hohmann and Posselt 2019; Tiwari 2020). However, current research efforts overlook an essential aspect, namely, developing data models for multichannel communication facilitating IIoT-enabled DS in SPL in the manufacturing industry. Accordingly, this study contributes to the understanding of IIoT-enabled DS for SPL in the manufacturing industry by answering the following research questions (RQ) that originate from applying data models for multichannel communication.

RQ 1: How do data models for multichannel communication in IIoT-enabled DS for SPL facilitate capturing, processing, and transferring of information across products, services, and software?

RQ 2: How can data models for multichannel communication support monitoring, optimisation, control, and autonomy services for IIoT-enabled DS in SPL?

RQ 3: What are the operational benefits of applying data models for multichannel communication in IIoT-enabled DS for SPL?

3. Proposed architecture and data model

This section describes a three-layered architecture, a data model, and services for the multichannel communication of IIoT-enabled DS in SPL for material handling.

3.1. Three layered architecture for smart production logistics

Establishing an architecture that supports multichannel communication of IIoT devices is crucial for realising DS in SPL. This study applies two main requirements for developing an IIoT-enabled DS architecture in SPL for material handling. The first is real-time collection of data from logistics resources, a means of networking and communication, and applications ensuring end-to-end services (Rymaszewska, Helo, and Gunasekaran 2017; Naik et al. 2020; Boehmer et al. 2020). The second requirement is realising product-service-software systems leading to output-based resource efficiency facilitated by monitoring, control, optimisation, and autonomy (Kohtamäki et al. 2019; Cimini et al. 2021; Ardolino et al. 2018). The study applies a three-layered IIoT architecture (IEC 2016), comprising sensing, network, and application layers, to extend the understanding of IIoT architectures in SPL for material handling (Y. Zhang, Zhu, and Lv 2018; Wang, Zhang, and Zhong 2020; Mörth et al. 2020; Guo et al. 2021; N. Zhang 2018). Figure 1 presents the proposed architecture for IIoT-enabled DS in SPL.

Figure 1. Proposed architecture for Industrial Internet of Things-enabled digital servitization in smart production logistics.

The sensing layer is the lowest level of an IIoT architecture and responsible for collecting and transforming data from resources or the environment into a digital setup. This layer includes material handling resources (e.g. parts, warehouses, or staff) that are made smart by attaching multiple IIoT devices. IIoT devices include sensors, actuators, and automatic identification systems, including cameras, radio frequency identifying devices (RFID) sense points, and ultra-wide bandwidth (UWB) tags. IIoT devices transmit multichannel and real-time data. Finally, the physical layer contains information about the manufacturing system, including its process, layout, and material handling tasks. This information may be contained in the enterprise resource planning (ERP) systems represented by the manufacturing system database in Figure 1.

The networking layer establishes the multichannel communication essential for achieving IIoT-enabled DS in SPL. This layer captures, processes, and transfers information across products, services, and software to conduct material handling tasks. Capturing includes acquisition and encapsulation of real-time data from IIoT devices on the factory floor. Processing refers to transformation of data to information and its standardisation, which is crucial for adding value to information and facilitating its use in SPL. Transfer of information involves mobility, authentication, authorisation, and management of information across products and services. Importantly, the networking layer facilitates transfer of information from IIoT devices in the sensing layer to databases for their subsequent use in the applications layer, represented by the real-time database in Figure 1.

The application layer is the top layer of the proposed architecture for realising IIoT-enabled DS in SPL and provides services for specific needs. This study proposes four services for material handling tasks executed by AGVs. The optimisation service acquires a list of tasks for material handling and produces an optimal schedule for the delivery of material handling tasks. Then, a control service determines the shortest route for executing the material handling schedule. Next, the autonomy service calculates and executes the movement of the AGV during material handling. Finally, a monitoring service reports the AGV status and material handling tasks. The monitoring service reports deviations between the scheduled and executed tasks in material handling. If deviations occur, the optimisation service generates a new delivery schedule for material handling.

3.2. Data model for digital servitization in multichannel communication

The study applies SysML to propose a data model for multi-channel communication in IIoT-enabled DS for SPL in material handling. SysML is designed to support specialised domains by applying extension mechanisms, named stereotypes, which are grouped into packages called profiles (Friedenthal, Moore, and Steiner 2015). Accordingly, the proposed data model for multichannel communication in IIoT-enabled DS for SPL is composed of three modelling profiles. The IIoT-enabled DS for SPL Product Profile includes the functions, devices, and activities of IIoT devices. The IIoT-enabled DS for SPL Database Profile presents information contained in databases including tasks in material handling, optimised schedule for material handling, layout, routes followed by AGVs, status of AGVs, status of tasks in material handling, and AGV location. The IIoT-enabled DS for SPL Service Profile consists of a monitoring, optimisation, control and autonomy service for material handling. Figure 2 presents the proposed data model for multichannel communication in IIoT-enabled DS for SPL in material handling.

Figure 2. Proposed data model for multichannel communication in Industrial Internet of Things-enabled digital servitization for smart production logistics.

A data model for multichannel communication including IIoT modelling profiles, data bases, production logistics tasks, and services.

IIoT-enabled DS for SPL Product Profile includes the characteristics, functions, and activities of IIoT devices. A block generates the stereotype for the functions of IIoT devices, shown in Figure 3. It contains the attributes and parameters, including device name, its functionality, type of data, location of a device, and layout. The interface of the function of an IIoT device includes a connector and a port. The stereotype of devices in IIoT relates to sensors and actuators for material handling. Sensors include the attributes of identifying and tracking an object. Sensors are classified into cameras or UWB tags and RFID sense points. Actuators include attributes of identifying and tracking an object, and controlling an object. AGVs are classified as actuators. The attributes of both sensors and actuators include the identification number of each device, its communication protocol, manufacturer, and type of device, including a choice (e.g. a choice between sensor and actuator). Activities of IIoT devices include collecting data and controlling resources. The values and parameters for collecting data comprise position, time, and battery. Similarly, the values and parameters for controlling resources involve status and mission name specifying the actions of the controlling resource.

Figure 3. Proposed industrial internet of things digital servitization for smart production logistics function profile.

Function profile for IIoT-enabled digital servitization describing the functions, parameters, connectors and stereotypes for smart production logistics.

IIoT-enabled DS for SPL Database Profile includes stereotypes for tasks in material handling, optimised schedule for material handling, layout, routes followed by AGVs, status of AGV and tasks in material handling, and AGV location. These stereotypes specify values, parameters, and types of data supporting the monitoring, optimisation, control, and autonomy services. As an example, we present the IIoT-enabled DS for SPL database profile for material handling tasks in Figure 4. Its attributes include ID, sequence ID, order sequence, task type, pickup time, material handling time, consumption time, from location, to area, to location, transfer order type, and state. IIoT-enabled DS for SPL Service Profile comprises a monitoring, optimisation, control, and autonomy service for material handling. The following section presents a detailed description of these services, data needs, and their contribution to IIoT-enabled DS in SPL for material handling.

Figure 4. Proposed industrial internet of things digital servitization for smart production logistics database profile.

Database profile for IIoT-enabled digital servitization describing the parameters for material handling tasks in smart production logistics.

3.3. Services supporting Industrial Internet of Things-enabled digital servitization in smart production logistics

This section describes the monitoring, optimisation, control, and autonomy services in material handling for the IIoT-enabled DS in SPL. These services originate from the communication and interoperability of information facilitated by the proposed data model. We describe the content and flow of information across products, services, and databases that facilitate each service. Further, we present the contribution of each service to realising IIoT-enabled DS for SPL in material handling. Figures 5 presents the monitoring, optimisation, control, and autonomy services in material handling for IIoT-enabled DS in SPL.

Figure 5. Monitoring, optimisation, control, and autonomy services in material handling for the Industrial Internet of Things-enabled digital servitization for smart production logistics.

Example of IIoT-enabled digital servitization for smart production logistics in material handling including monitoring of status, optimisation of schedule, control of routes, and autonomy in the execution of tasks.

The monitoring service informs the real-time status of AGVs and material handling tasks. It presents real-time information, including the status of AGVs, comprehends the current position, distance travelled, condition (e.g. moving, stopped, charging, or maintenance), areas of concentration for material handling activity (e.g. heap maps), and patterns followed by AGVs during the execution of material handling tasks (e.g. spaghetti diagrams). Real-time information about the condition of tasks for material handling includes order information about the task under execution, distance of the AGV from its current location to the point of delivery, estimated time of arrival, and shortest route between tasks in material handling.

The monitoring service depends on the information exchanged among products, services, and software databases in three ways. First, it requires real-time information from the AGVs connected to the IIoT devices. This information includes the name of AGVs, time, longitude, and latitude transmitted by UWB tags and RFID sense points. Second, the monitoring service needs information from software databases necessary for displaying the layout of the factory and includes the distance, position, and dimensions of machines, material racks, crossways, AGV paths, and dimensions of the factory. Third, it applies information from other services. For example, it requires information from the optimisation service to present the chronological sequence of scheduled tasks in material handling for the delivery of parts. The monitoring service contributes to IIoT-enabled DS in SPL by increasing the visibility of AGVs and tasks in material handling. Additionally, the monitoring service is essential for increasing flow of information across different levels of seniority in a logistics organisation, departments, and customers, and detecting deviations that jeopardise the delivery of materials and information.

The control service manipulates resources remotely and corrects the conditions for the delivery of material handling. In particular, the control service determines the shortest route for delivery of parts. To do so, the control service first acquires the position of the AGV and its assigned material handling task, including the location for the confinement and delivery of parts. Next, it calculates the shortest distance between the location of confinement and delivery and estimates the time of travel between these locations. Additionally, the control service corrects the route for the delivery of material when there are changes in scheduled material handling tasks.

The control service requires information originating from products, services, and software databases to contribute to the execution of material handling by an AGV. For example, this service applies real-time information about the position of the AGV originating from IIoT devices to update the shortest delivery route. In addition, the control service utilises information from software databases, including the optimal order sequence, and combines this information with that of the position, time, and location of the AGV to its next delivery task. Additionally, the control service applies software databases' information containing the distance, position, and feasible routes in a layout. It supports IIoT-enabled DS in SPL for material handling in several ways. The connectivity of the control service enabled by IIoT devices allows changes in the behaviour of AGVs in material handling. Specifically, the AGV does not travel on fixed routes, which may lead to increased cost, energy consumption, or makespan, jeopardising the time delivery of materials and information. Instead, the control service proposes and adapts the route for picking and delivery of scheduled material handling tasks in real time in line with the preference of the shortest distance. Accordingly, the control service allows the customisation of performance, directing the manner in which resources execute material handling tasks.

The optimisation service involves the use of digital technologies to enhance the operational performance of equipment on the factory floor. Specifically, the optimisation service arranges a set of material handling tasks in chronological order and minimises the energy consumption, makespan, and late deliveries of these tasks. To achieve this, the optimisation service requires information from the products, services, and software databases. It applies information from the monitoring service, including information about the status of material handling tasks. The monitoring service detects a deviation in the scheduled tasks and informs the optimisation service of the need for a new schedule. Further, the optimisation service requests information about the factory layout, including the distance, position, and dimensions of machines, material racks, crossways, AGV paths, and dimensions of the factory. Additionally, the optimisation service applies software databases' information containing the characteristics of the AGV movement, including its name, speed, and acceleration.

The optimisation service supports IIoT-enabled DS in SPL for material handling in two ways. First, the optimisation service enables the optimised utilisation of individual resources assigned to the execution of material handling tasks. This is important for satisfying customer needs, including on-time delivery, enhancing operational performance, and understanding trade-offs. Second, the optimisation service supports proactive actions, including determining the consequences of changes to the number of AGVs in material handling and clarifying the effects of changes in customer orders.

The autonomy service facilitates the independent execution of material handling tasks by AGVs. The autonomy service requires the transfer of information that describes the behaviour of AGVs during the execution of material handling tasks. This information includes a chronological list of locations and motion profiles executed by an AGV (e.g. time, start and end position, velocity, and acceleration) corresponding to scheduled material handling tasks. Furthermore, the autonomy service helps an AGV scan, identify, and avoid obstacles in its path.

The autonomy service applies information from products, services, and software databases to execute the movement of the AGV in material handling. This service acquires the real-time position of AGVs from IIoT devices, including UWB tags and RFID sense points. Additionally, this service utilises information from the software databases, including the movement profile of the AGV, sequence of steps to set in motion and stop the AGV, and layout of the factory. Finally, the autonomy service acquires information from the optimisation service specifying the AGV's start and finish positions when executing material handling tasks.

The autonomy service supports IIoT-enabled DS in SPL in three ways. It enables the self-coordination of information among products, services, and software, leading to independent execution of tasks in material handling by AGVs. This is crucial for improving on-time delivery of materials in the context of shorter manufacturing times. Additionally, the automatic transfer of information about the execution of material handling tasks is crucial for realising self-coordination with other products or systems adjacent to SPL, such as maintenance responsible for AGVs, or coordinating orders with suppliers and end customers. Furthermore, the autonomy service combines the monitoring, control, and optimisation services enabled by the IIoT. Combining these services allows the flow of information and materials in a supply chain without the intervention of staff facilitating operations from remote locations, reducing the need for operators in low-value adding activities.

4. Empirical results

This study applies the proposed data model for multichannel communication facilitating IIoT-enabled DS in a laboratory environment based on a case study of a Swedish manufacturing company.

4.1. Case description

The manufacturing company is an original equipment manufacturer in the automotive industry. The case involves the material handling tasks of picking and delivering materials in the manufacturing of transmission components. Manually operated forklifts deliver supplies from the warehouse to manufacturing stations and transfer work-in-progress parts or finished products from earlier to later stages of the manufacturing process. The manufacturing company experiences problems when executing material handling tasks. Currently, forklift drivers execute material handling tasks by driving on fixed routes at established times during a shift. Drivers rely on experience when locating supplies, parts, and finished products, a situation that is prone to errors and is time consuming. Management identifies that current practices lead to failed on-time deliveries, increased makespan, and unnecessary movements of forklifts.

Management has identified four aspects that are critical for improving the current material handling practices. First, monitoring the status of forklifts with information including travelled distance, real-time location, and operational conditions (e.g. en route, idle, or charging). Second, optimising the delivery schedule of material handling tasks to minimise late deliveries, makespan, and energy consumed in material handling. Third, controlling and providing routes that lead to a shorter travelling distance for material handling. Fourth, automatically executing tasks in material handling with AGVs.

4.2. System implementation

This section describes the implementation of the proposed data model for multichannel communication in a proof-of-concept prototype in a laboratory environment. Figure 6 presents the prototype. This section describes the hardware, networking, databases, and software for its implementation.

Figure 6. Proof-of-concept prototype in a laboratory environment.

Layout of laboratory environment including 10 stations, six Radio Frequency Identification Device sense point, an Ultra Wide Band Width tag, and one automated guided vehicle.

The hardware supporting the proof-of-concept includes physical devices for executing material handling tasks and IIoT devices for acquiring and transmitting the real-time status of an AGV. This study utilises a Mobile Industrial Robot (MiR) AGV to execute material handling tasks. Parts are manually placed in the AGV upon arrival at a manufacturing cell or buffer. Additionally, the AGV may deliver or pick parts from a warehouse or travel to a charging station. The material handling tasks are confirmed with a graphical user interface. The study applies IIoT devices, including UWB tags and RFID sense points, to acquire and transmit the real-time status of the AGV. A UWB tag transmits real-time data including the position of the AGV to six RFID sense points three times every second. The RFID sense points process the data and provide information, including time, location, and status of the AGV (e.g. en route, arrival at destination, idle, and charging). A RESTful application programming interface (API), including the RFC 2616 protocol, transfers data from the RFID sense point to a cloud server. The RESTful API is chosen for its lower bandwidth requirement for accessing and using data.

Networking involves capture of data from IIoT devices and processing and transfer of information. This study applies Node-RED as a middleware for capturing data, processing, and transferring information. Node Red is chosen as the middleware for its ease of use through visual programming and multiple connections to APIs, hardware, and online services, among others. Node-RED captures and processes information by converting raw data from IIoT devices into information about the position of the AGV. To do so, it calls the API from the cloud server containing the data from the RFID sense points and retrieves a JSON object. Then, it establishes and maps the parameters for determining the position of the AGV (e.g. time, AGV name, longitude, latitude, speed, and acceleration) from the JSON object.

The middleware contributes to transferring information in three ways. First, it contributes to transferring information from IIoT devices, including the position of the AGV, to the databases. Information about the position of the AGV is critical for registering its movement, comparing the execution to the scheduling of material handling tasks, and determining deviations. Transferring information, including the position of the AGV to databases, is resource-consuming. Therefore, the transfer of information includes a comparison to determine whether the AGV is in motion or stationary. Node-RED transfers information about the position of the AGV to a database object exclusively when the AGV is in motion and but not when the AGV is stationary. Second, the middleware facilitates the transfer of information from the IIoT devices, including the position of the AGV, to the services. This includes real-time information, which is essential for monitoring the execution of material handling tasks and the status of the AGV. This study accomplishes this by combining the middleware with a data streaming bus (e.g. Apache Kafka). This choice is explained by the open-source platform and the handling of the multiple API calls necessary to reduce the burden on the Apache Kafka servers. Third, the middleware transfers information from the services to the physical devices, including the movement of the AGV. This information determines the movement of the AGV during the execution of material handling tasks and includes the name of the AGV, start and finish locations, speed, acceleration, and path. Node Red transfers the information from the autonomy service to the AGV with a JSON object.

The information originating from the IIoT devices and services in the proof-of-concept is stored in a database. This study utilises the relational database management system, MariaDB, for this purpose. This is because it is a free and open-source software, which contributes to a lower cost, and the use of declarative methods that facilitate the specification of data and queries. The proof-of-concept stores its databases in a Linux-based server 4 GB RAM, 120 GB SSD. The proof-of-concept is applied to six database tables. The first data table specifies the layout of the proof-of-concept, including the coordinates of all physical objects, distances between objects, and routes. The second lists the material handling tasks that the AGV executes. The third is the optimal schedule of the material handling tasks. The fourth is the optimal route proposed by the control service for delivering material handling tasks. The fifth is the routes followed by the AGV during the execution of the material handling tasks. The final table includes the results of the optimisation service, including the makespan, energy consumption, and delivery.

Finally, this study applies four software for generating the monitoring, optimisation, control, and autonomy services. The monitoring service includes the use of the commercially available software Griffin Enterprise Positioning (https://www.hd-wireless.com/products/geps/). Figure 7 presents a screenshot of the monitoring service for material handling for the proof of concept. Additionally, the study applies Matlab for developing optimisation service. Then, the control services apply the Dijkstra algorithm for path finding programmed in Python. Figure 8 provides a screenshot of the control service in the proof of concept. The autonomy service is developed with the MiR software. Figure 9 shows a screenshot of the autonomy services in the proof of concept.

Figure 7. Screenshot of the monitoring service for material handling for the proof of concept.

Spaghetti diagram and status of an automated guided vehicle during execution of material handling tasks in a proof of concept laboratory environment.

Figure 8. Screenshot of the control service in the proof of concept.

Route optimisation for automated guided vehicle during the execution of material handling tasks in a proof of concept laboratory environment.

Figure 9. Screenshot of the autonomy services in the proof of concept utilising MiR software.

Automatic execution of tasks in material handling including the use of an automated guided vehicle in a proof of concept laboratory environment.

4.3. Optimisation results

This section presents empirical data drawn from the SPL proof-of-concept for validating the proposed framework involving IIoT-enabled DS for SPL in material handling. The results consider three key performance indicators (KPIs): material handling delivery, makespan, and energy consumption. A real-time data-driven dynamic optimisation method is used to obtain the optimal solutions for SPL. The detailed optimisation procedure is presented in the Appendix an the notations used in the study are summarised in Table A1.

The proof-of-concept demonstrates the efficiency of the proposed data-driven dynamic optimisation method for SPL. Table A2 (Appendix) indicates the position of the stations in the layout of the proof-of-concept. The material handling tasks in each cycle are initialised to create task sets. Table A3 (Appendix) presents the tasks for material handling in the proof-of-concept, and it shows 10-cycle tasks with five tasks in each cycle, the task number, type of task (value adding or non-value adding), pick and delivery stations, time for pick-up and delivery of material, and time window constraints. The value adding and non-value adding types of tasks indicate the transfer of parts/tasks and empty pallets, respectively. The data-driven dynamic optimisation method is used to select tasks satisfying Equation (A1(A1) $\begin{array}{rl}f\left({\mathbf{F}}_i\right)& =\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{r}\mathrm{c}\mathrm{c}\mathrm{o}\mathrm{s}\left[\right({\mathbf{T}}_i{\mathbf{F}}_j\left)/\right(|{\mathbf{T}}_i||{\mathbf{F}}_j|\left)\right]\\ \mathrm{s}.\mathrm{t}.{\mathbf{W}}_i& \le {\mathbf{S}\mathbf{W}}_j,{\mathbf{V}}_i\le {\mathbf{S}\mathbf{V}}_j,i\in [1,n],j\in [1,m]\end{array}$(A1) ) as pre-optimised subsets. The tasks in the intersections within more than one AGV are re-optimised by Equation (A3(A3) $f=min(w_1\ast C\ast D+w_2\ast T+w_3\ast E)$(A3) ). Finally, the optimal results meeting the optimisation objective are obtained as shown in Table A4 (Appendix). This includes the optimal sequence of the tasks and the actual time of picking up and delivering the tasks from pick-up and delivery locations for all tasks in each cycle.

The value of the makespan, battery consumption, and delivery time for the optimal results are shown in Table A3 (Appendix). The optimisation considers a hard time window for achieving on-time delivery. Staff benefit from the optimal results because these include the location, time, and sequence for picking and delivering tasks. Consider the optimal task sequence in Cycle 1. This includes tasks 1, 4, 5, 2, and 3 with a makespan of 325 s, delivery time of 206 s, a total travelled distance of 18.16 m, and energy consumption of 162.50 W (the battery consumption per unit is assumed as 1.8 kW/hour; Guo et al. 2021). Results also show that the AGV had the lowest battery consumption and required the shortest distance while finishing tasks in Cycle 2. By contrast, the delivery of the tasks in Cycle 7 generated the highest objective value, and the makespan, distance, and battery consumption were 365 s, 20.39 m, and 182.50 W, respectively. The observation results show that the proposed optimisation method can offer optimal scheduling solutions for efficient SPL. Consider the importance of on-time delivery at takt time. Our results show that the proof-of-concept prototype in a laboratory environment completes all tasks in material handling with a makespan lower than that of the takt time. Figure 10 presents the results from the data-drive optimal schedule of material handling tasks in SPL.

Figure 10. Results from the data-drive optimal schedule of material handling tasks in smart production logistics.

Results of optimised schedule of tasks in material handling comparing makespan and takt time.

5. Discussion

This study proposes a data model for multichannel communication that facilitates IIoT-enabled DS for SPL in the manufacturing industry, and applies the principles of SysML. In particular, the study addresses the need for information sharing and interoperability facilitated by the IIoT across products, services, and software, ensuring the reliable delivery of materials and information to customers. Additionally, it focuses on the acquisition and use of data from multiple IIoT devices to monitor, control, optimise, and execute the autonomous decisions necessary in service-oriented manufacturing. Furthermore, the study presents the operational benefits of applying the proposed data model in material handling. The findings are particularly relevant for manufacturing managers and academics and improve our understanding of the deployment of IIoT-enabled DS for SPL in manufacturing.

5.1. Facilitating the capture, process, and transfer of information across products, services, and software (RQ1)

Capturing, processing, and transferring of information across products, services, and software is a critical capability for manufacturers pursuing IIoT-enabled DS (Tronvoll et al. 2020). In particular, manufacturers must capitalise on the cooperation across products, services, and software for transitioning from product- to service-oriented manufacturing (Kohtamäki et al. 2019). However, the literature presents meagre contributions to understanding the resources and activities that facilitate this transition (Naik et al. 2020). Addressing this gap, this study provides new insights into the importance of data models for multichannel communication facilitating the capture, processing, and transferring of information across products, services, and software in IIoT-enabled DS for SPL.

The key contributions of this study include presenting data models for the multichannel communication of IIoT devices based on SysML. This finding is essential for establishing the properties, functionalities, and environment for the communication and interoperability of products, services, and software for IIoT-enabled DS in SPL (Rymaszewska, Helo, and Gunasekaran 2017). For example, the proposed data model applies the concept of stereotypes and defines three modelling profiles'a product, database, and service profile – for IIoT-enabled DS for SPL. Thus, the proposed data models depict how the system model specifies the hardware and software components in IIoT-enabled DS for SPL. This is critical for establishing component interconnections, interfaces, interactions, associations and performance of resources (Friedenthal, Moore, and Steiner 2015), a situation that prior studies identify as problematic because of the diverse activities, resources, IT systems, and formats of information in SPL (Song et al. 2021).

The reusability of the proposed data model for multichannel communication is an additional advantage. Modelling profiles of SysML extend the modelling language by applying distinct properties, rules, and relationships in new domains (Friedenthal, Moore, and Steiner 2015; Obermeier, Braun, and Vogel-Heuser 2015). This study proposed three modelling profiles – a product, database, and service profile – for IIoT-enabled DS for SPL for material handling aimed at capturing, processing, and transferring information across products, services, and software. In particular, the study applied AGVs for material handling as actuators and RFID as sensors. However, future studies applying IIoT devices transmitting multichannel and real-time data in material handling can apply the proposed modelling profiles. Accordingly, new applications of material handling concerned with capturing, processing, and transferring information across products, services and software can add or remove IIoT devices.

Applying SysML presents benefits distinct from those put forward by existing research efforts for IIoT-enabled DS in SPL. For example, prior studies about IIoT-enabled DS give precedence to pathways manufacturers could adopt for realising IIoT-enabled DS (Hsuan, Jovanovic, and Clemente 2021). However, the literature does not disclose the tools that support staff in the design of critical functions and interfaces facilitating the transfer of information across products, services, and software. This study reveals how a formalised modelling language (e.g. SysML) contributes to achieving this. Doing so is crucial for enhancing communications across staff responsible for the design of IIoT-enabled DS, reducing development risks, improving quality, and leveraging models during downstream lifecycle phases (Holt and Perry 2018). Consider the specification of attributes for both sensors and actuators in the proposed modelling profiles. These identify the number of each device, its communication protocol, the manufacturer, and type of device. Accordingly, applying SysML for developing data models for multichannel communication enhances specification, design quality, and consistency, reuse of the specification and design artifacts, and communication among the development team (Wang et al. 2021).

This study presents several key findings regarding the importance of data models in the multichannel communication of IIoT devices. A multichannel communication system facilitated by IIoT devices is adopted to sense and collect data on manufacturing resources and devices in the physical layer, which is composed of cameras, RFID, sensors, and AGVs. Therefore, real-time production logistics tasks can be collected to form a data model of the production logistics systems, and the real-time location information of the AVGs is acquired from the developed real-time location system. In addition, an RFID-enabled material handling system can track the information of the materials and visualise the data of the logistics task on the AGVs. Communication modalities between the physical and digital parts of production logistics systems are deployed for data transmission to the defined data model. Then, digital services driven by the data model are developed for dynamic optimisation of SPL and reliable material handling operations. Within this context, the processed data such as logistics time and constraints are defined as the input of the multi-objective optimisation model, and then the optimal results are automatically transmitted to the AGVs for the delivery of the logistics tasks. Accordingly, our findings extend those of previous studies focussing on the use of IIoT devices in SPL (Qu et al. 2016; Y. Zhang et al. 2015; Tu, Lim, and Yang 2018), by addressing what information can be communicated across various IIoT devices in service-oriented manufacturing and how this can be achieved.

5.2. Supporting monitoring, optimisation, control, and autonomy services (RQ2)

This study highlights the importance of data models for aligning data originating from the factory floor with information supporting monitoring, optimisation, control, and autonomy services. These services are indispensable for increasing visibility, assessing operational risks, and responding to customer needs and unexpected events in the context of service-oriented manufacturing (Rymaszewska, Helo, and Gunasekaran 2017; Jovanovic, Sjödin, and Parida 2021). Prior research efforts apply the IIoT and develop independent monitoring, optimisation, control, and autonomy services in SPL (Zhao et al. 2020; Zheng et al. 2019; Trappey et al. 2017). Unlike prior research, a key contribution of this study is presenting data models for multichannel communication that facilitate the integration and interconnection of once isolated activities of SPL. Thus, this study presents an operation of material handling that enhances the value proposition supporting customer goals (e.g. availability, reliability, or performance) as opposed to prior efforts focussing on the physical movement of material in a factory (Mosch, Schweikl, and Obermaier 2021; Vendrell-Herrero et al. 2017). For example, real-time-information-enabled data models have the capacity to monitor material handling processes. The optimisation modules with smart algorithms are executed to achieve dynamic decision-making for logistics scheduling with the data input. Further, routing solutions are generated to allow the AGVs to deliver optimal logistics tasks and execute control of logistics processes efficiently.

Additionally, the proposed data model for multichannel communication offers a reliable tool for efficient data interaction across monitoring, optimisation, control, and autonomy services for IIoT-enabled DS in SPL. The proposed data model is unlike that of prior research efforts in IIoT-enabled SPL. For example, prior publications focus on distinct services enhancing operational performance including monitoring, optimisation, control, or execution of material handling tasks in SPL (Mörth et al. 2020; Guo et al. 2021). Yet, to the best of our knowledge, studies on SPL remain absent of data models for multichannel communication, supporting the transfer of information across an IIoT-enabled architecture for SPL and simultaneously addressing monitoring, optimisation, control, and autonomy services. This is important for efficiently collecting large sets of data and processing information (Qu et al. 2016), which are vital in managing the increasing number of IIoT-devices in SPL economically and competitively (Winkelhaus and Grosse 2020).

Moreover, this study reveals that applying SysML to data models for multichannel communication facilitates the explicit description of data structures, behaviours, requirements, and constraints of modelling profiles for IIoT-enabled DS in SPL. This perspective is unique and unlike that of recent studies that apply SysML to describe the organisation and decomposition of data models for achieving digital services (Wang et al. 2021), or focus on the design of smart resources in SPL including material handling (Aloui et al. 2021). These findings are essential for establishing communication across devices, databases, and services and designing architectures that support IIoT-enabled DS (Jovanovic, Sjödin, and Parida 2021). The IIoT-enabled DS proposed in this study is difficult to express intuitively. It is provided in an abstract result, such as a service, thus it is difficult to understand the relationship between the components and specifications of the system. However, through MDE methods such as SysML, abstract results can be intuitively expressed as graphical results. This helps understand the data model and system architecture when designing a system for DS. In addition, many people can understand the data model and system architecture expressed in SysML without needing a separate notation or explanation because SysML is a standard modelling language. IIoT-enabled DS involves various stakeholders such as IIoT technology providers, hardware and software developers, and service users; therefore, it is important to interpret the model from a common perspective. This study's results, therefore, complement recent research efforts applying SysML for IIoT-enabled manufacturing (Wang et al. 2021; Thramboulidis 2015), and extend our understanding of the use of SysML in the domain of IIoT-enabled DS, which is essential for achieving our goal. This is important because IIoT-enabled DS is a novel research domain, and the need to enhance the sharing of information across IIoT devices and adjacent systems in manufacturing or through supply chains is likely to increase (Sisinni et al. 2018).

5.3. Operational benefits (RQ3)

This study identifies the operational benefits of applying data models for multichannel communication in IIoT-enabled DS for SPL based on the conceptual proposal in Section 3 and the proof-of-concept and simulation results in Section 4. The results show that realising services, including monitoring, optimisation, control, and autonomous decisions, improves the makespan, delivery, and energy consumption in material handling. The literature suggests that IIoT-enabled DS contributes to increased operational efficiency, resource allocation, and transparency (Coreynen, Matthyssens, and Van Bockhaven 2017). However, extant understanding remains conceptual in nature and lacks empirical evidence about smart solutions and advanced services in IIoT-enabled DS facilitating increased performance (Kohtamäki et al. 2019). To this extent, this study presents a novel finding leading to an outcome-based solution with increased operational performance in material handling. Therefore, our study contributes to clarifying the importance of achieving IIoT-enabled DS for supporting operations at a facility by delivering the right material in the right quantities to the right location at the right time. Accordingly, the implications of presenting operational benefits of applying data models for multichannel communication are critical for establishing outcome provisions in IIoT-enabled DS (Visnjic, Neely, and Jovanovic 2018). This finding suggested that solution providers (e.g. AGV or forklift manufacturers) realising data models for multichannel communication in IIoT-enabled DS may support the sale of outcome-based offerings necessary for transitioning from product to service-oriented manufacturing (Lerch and Gotsch 2015).

5.4. Managerial contributions

This study provides direct managerial implications for enhancing the success of IIoT-enabled DS in SPL. Primarily, it underscores the importance of systemic thinking in establishing data models for multichannel communication. Our findings imply a call for standardisation and routines when securing communication and interoperability among IIoT devices for DS. In doing so, managers may avoid isolated systems and promote the sharing of information between existing and new applications in IIoT-enabled DS for SPL. Furthermore, this study describes a generic representation for capturing, processing, and transferring information enabled by the IIoT originating from multiple sensors in material handling. Our findings are relevant to factory managers, production planners, and development engineers responsible for logistics departments in the manufacturing industry. Accordingly, this study offers guidance on how to realise the communication and interoperability among products, software, and services in IIoT-enabled DS for SPL in material handling. By applying these findings, managers can increase visibility, assess operational risks, and enhance interventions to improve the flow of materials and information.

IIIoT-enabled DS is a nascent field and providing guidance for its successful implementation in manufacturing is crucial for fully capitalising on the advantages of service-oriented manufacturing (Schroeder et al. 2016; Pirola et al. 2020; Zambetti et al. 2021). Our study contains several recommendations for managers that may enhance the deployment of IIoT-enabled DS in SPL. The findings reveal that applying extant knowledge about SPL may be critical for realising the communication of information across products, services, and software. For example, manufacturing companies may benefit from adopting established IIoT architectures in SPL, supporting the capture, processing, and transfer of information. This is essential for developing monitoring, optimisation, control, and autonomy services crucial for IIoT-enabled DS.

6. Conclusions

This study aimed to develop a data model for multichannel communication that facilitates IIoT-enabled DS for SPL in the manufacturing industry. It applied the proposed data model to a Swedish manufacturing company in the automotive industry, focussing on the execution of material handling tasks by AGVs. This study contributes to the extant understanding in three ways. First, the study presents three modelling profiles, including IIoT devices, databases, and services for multichannel communication. This is essential for capturing, processing, and transferring information across products, services, and software essential to realising IIoT-enabled DS in SPL. Second, the proposed modelling profiles are applied to transfer information across monitoring, control, optimisation, and autonomous decision services in the execution of material handling tasks by AGVs. Third, the study presents the operational benefits of applying the proposed data model in a case study for the Swedish automotive industry, including improvements in the delivery, makespan, and energy consumption of material handling.

This study has three limitations. The first is the application of the proposed data model to material handling tasks executed by AGVs. Recent studies (Mörth et al. 2020; Wang, Zhang, and Zhong 2020; C. Liu et al. 2019), have argued that increased automation of material handling improves operational performance in SPL involving conveyor systems, robots, or interfaces with human operators. Figure 5 presents the logic of the proposed data model for achieving this goal. Therefore, validating our results against additional resources, including automation in material handling, is important. The validation of the results could include adopting the proposed solution in different scenarios (e.g. additional AGVs) for verifying the effectiveness of the model. Additionally, future work could focus on verifying the effectiveness of the proposed data model using static analysis methods and dynamic testing. Another limitation is the deployment of the proposed data model as a proof-of-concept in a laboratory environment. This study represents a first step toward facilitating IIoT-enabled DS for SPL. However, future research could focus on empirical research in manufacturing to investigate the communication and interoperability of the proposed data model for multichannel communication with legacy systems. Exploring this topic is crucial for facilitating the transition from product- to service-oriented manufacturing necessary in the context of IIoT-enabled DS. Finally, IIoT-enabled DS offers manufacturers a competitive advantage, including new business models integrating suppliers, factories, and customers, and complementing traditional business offerings. Future research could investigate how capabilities facilitated by multichannel communication in IIoT-enabled DS for SPL contribute to a transformation in the manufacturing supply chain, leading to new business models and processes supporting customers to achieve their goals.

Data availability statement

The data that support the findings of this study are available from the corresponding author, Erik Flores-García ([email protected]), upon reasonable request.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors would like to acknowledge the support of Swedish Innovation Agency (VINNOVA). This study is part of the Cyber Physical Assembly and Logistics Systems in Global Supply Chains (C-PALs) project. This project is funded under SMART EUREKA CLUSTER on Advanced Manufacturing program.

Notes on contributors

Erik Flores-García

Erik Flores-García is a researcher in the Department for Sustainable Production Development at KTH Royal Institute of Technology. He earned his Ph.D. in Innovation and Design (2019) and MSc (2013) from Mälardalen University. His research interests include applying simulation, digital twins, and CPS in production logistics.

Yongkuk Jeong

Yongkuk Jeong is an Assistant Professor in Production Logistics at KTH Royal Institute of Technology, Sweden. He received his Ph.D. in Engineering from Seoul National University, South Korea, in 2018. He has experience in production and logistics simulation with various manufacturing companies. His research interests are focussed on digitalisation in production logistics, manufacturing simulation, and sustainable production systems. He has published 30+ scientific publications, including International Journal of Production Research (IJPR) and International Journal of Advanced Manufacturing Technology (IJAMT).

Sichao Liu

Sichao Liu received the bachelor's and master's degrees in Mathematics and Mechanical Engineering from the Northwestern Polytechnical University, Xi'an, China, in 2014 and 2017, respectively, and the Ph.D. degree in mechanical engineering from KTH Royal Institute of Technology, Stockholm, Sweden in 2022. His research interests are focussed on human-robot collaboration, robotics/brain robotics, robot learning, multimodal robot control, digital twin and smart production logistics.

Magnus Wiktorsson

Magnus Wiktorsson is a Chair Professor of Production Logistics at KTH Royal Institute of Technology. His research interest concerns how complex production logistic systems can be described and predicted. The application areas are within manufacturing industry and his research is based on digitisation within supply chains and the need for transformation into environmentally sustainable production. Magnus Wiktorsson has a professional background with experience from business, government and universities. He graduated from the Royal Institute of Technology in Stockholm with an MSc in Systems Engineering (1995) and a PhD in Manufacturing Engineering (2000). Prof Wiktorsson is the head of department of Production Engineering at KTH as well as the department of Sustainable Production Development at KTH.

Lihui Wang

Lihui Wang is a Chair Professor at KTH Royal Institute of Technology, Sweden. His research interests are focussed on cyber-physical production systems, human-robot collaborative assembly, brain robotics, and adaptive manufacturing systems. Professor Wang is actively engaged in various professional activities. He is the Editor-in-Chief of International Journal of Manufacturing Research, Journal of Manufacturing Systems, and Robotics and Computer-Integrated Manufacturing. He has published 10 books and authored in excess of 600 scientific publications. Professor Wang is a Fellow of Canadian Academy of Engineering (CAE), International Academy for Production Engineering (CIRP), Society of Manufacturing Engineers (SME), and American Society of Mechanical Engineers (ASME). He is a registered Professional Engineer in Canada, and was the President (2020–2021) of North American Manufacturing Research Institution of SME, and the Chairman (2018–2020) of Swedish Production Academy.

Appendix

Optimisation procedure

The real-time information of logistics resources from SPL are defined as the control inputs. The information of AGVs and tasks are represented by matrices $\mathbf{F}$ and $\mathbf{T}$. The AGV and tasks are divided into n sub-regions on the basis of their current locations. Then, n tasks and m AGVs subsets are formed as $\mathbf{T}=({\mathbf{T}}_1,{\mathbf{T}}_2,\dots ,{\mathbf{T}}_n{)}^T$, and $\mathbf{F}=({\mathbf{F}}_1,{\mathbf{F}}_2,\dots ,{\mathbf{F}}_m{)}^T$. In this case, tasks and AGVs in the same region are divided into the identical subsets, and form the pre-optimised sets (${\mathbf{T}}_i,{\mathbf{F}}_j$), $i\in [1,n]$, and $j\in [1,m]$. For (${\mathbf{T}}_i,{\mathbf{F}}_j$), the current location and destination could be constructed as task vector ${\mathbf{T}}_i({\mathbf{C}\mathbf{P}}_i,{\mathbf{N}\mathbf{D}}_i)$ and vehicle vector ${\mathbf{F}}_j({\mathbf{C}\mathbf{P}}_j,{\mathbf{N}\mathbf{D}}_j)$. A dynamic pre-optimisation model for tasks is proposed shown as the following steps. Step 1:construct the optimisation function that is to select the minimal angle of the task vector and AGV vector; (A1) $\begin{array}{rl}f\left({\mathbf{F}}_i\right)& =\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{r}\mathrm{c}\mathrm{c}\mathrm{o}\mathrm{s}\left[\right({\mathbf{T}}_i{\mathbf{F}}_j\left)/\right(|{\mathbf{T}}_i||{\mathbf{F}}_j|\left)\right]\\ \mathrm{s}.\mathrm{t}.{\mathbf{W}}_i& \le {\mathbf{S}\mathbf{W}}_j,{\mathbf{V}}_i\le {\mathbf{S}\mathbf{V}}_j,i\in [1,n],j\in [1,m]\end{array}$(A1) Step 2: AGV vector ${\mathbf{F}}_j({\mathbf{C}\mathbf{P}}_j,{\mathbf{N}\mathbf{D}}_j)$ and task vector ${\mathbf{T}}_i({\mathbf{C}\mathbf{P}}_i,{\mathbf{N}\mathbf{D}}_i)$ are substituted into Equation (A1(A1) $\begin{array}{rl}f\left({\mathbf{F}}_i\right)& =\mathrm{m}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{r}\mathrm{c}\mathrm{c}\mathrm{o}\mathrm{s}\left[\right({\mathbf{T}}_i{\mathbf{F}}_j\left)/\right(|{\mathbf{T}}_i||{\mathbf{F}}_j|\left)\right]\\ \mathrm{s}.\mathrm{t}.{\mathbf{W}}_i& \le {\mathbf{S}\mathbf{W}}_j,{\mathbf{V}}_i\le {\mathbf{S}\mathbf{V}}_j,i\in [1,n],j\in [1,m]\end{array}$(A1) ). If ${\mathbf{N}\mathbf{D}}_i=‘‘\mathrm{f}\mathrm{r}\mathrm{e}{\mathrm{e}}^{″}$, then free ${\mathbf{T}}_i$.

Step 3: tasks mapped with the obtained minimal value of the above-mentioned function is allocated to AGV j. The pre-allocated set for each task is built as follows. (A2) ${\mathbf{T}}_i=({\mathbf{F}}_1,{\mathbf{T}}_2,\dots ,{\mathbf{F}}_x),{\mathbf{V}}_i\le {\mathbf{S}\mathbf{V}}_j,i\in [1,n],x\in [1,m]$(A2) Step 4: pre-allocated sets are classified into three classes, namely Class 1: x = 0, Class 2: x = 1, and Class 3: $x\ge 2$.

Step 5: a multi-objective optimisation function for tasks in Class 3 is formulated as follows. (A3) $f=min(w_1\ast C\ast D+w_2\ast T+w_3\ast E)$(A3) Constraints: (A4) $\begin{array}{rl}& T{s}_i\le TF{s}_{ij},TF{c}_{ij}\le T{c}_j,i\in [1,n],j\in [1,m]\end{array}$(A4) (A5) $\begin{array}{rl}& TF{c}_{ij}-TF{s}_{ij}\le T{d}_i,\mathrm{b}\mathrm{u}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{r}\in (0,\propto )\end{array}$(A5) (A6) $\begin{array}{rl}& W_i\le S{W}_j,V_i\le S{V}_j,i\in [1,n],j\in [1,m]\end{array}$(A6) (A7) $\begin{array}{rl}& \sum _{i=1}^n{w}_i=1\end{array}$(A7) (A8) $\begin{array}{rl}& D=D_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}+D_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A8) (A9) $\begin{array}{rl}& T=T_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}+T_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A9) (A10) $\begin{array}{rl}& E=E_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}+E_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A10) (A11) $\begin{array}{rl}& E_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}=V_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}\ast D_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}\end{array}$(A11) (A12) $\begin{array}{rl}& E_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}=V_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\ast D_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A12) where Equation (A8(A8) $\begin{array}{rl}& D=D_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}+D_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A8) ) denotes the total distance of the AGV delivering all the tasks, and it is the sum of the distance with and without the tasks ($D_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}$ and $D_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}$). Equation (A9(A9) $\begin{array}{rl}& T=T_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}+T_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A9) ) describes the total delivery time of the AGVs during finishing all the tasks. $T_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}$ is the time used for delivering the tasks, and $T_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}$ is the running time of the AGV without tasks. Equation (A10(A10) $\begin{array}{rl}& E=E_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}+E_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A10) ) is the total battery consumption of the AGV that is composed of the battery from full and empty loads. Equations (A11(A11) $\begin{array}{rl}& E_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}=V_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}\ast D_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}\end{array}$(A11) ) and (A12(A12) $\begin{array}{rl}& E_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}=V_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\ast D_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}\end{array}$(A12) ) are battery consumption of the AGVs without and with task delivery, where $V_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}$ and $V_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}$ are battery consumption per unit, respectively.

Step 6: the AGV with the minimal value of Equation (A3(A3) $f=min(w_1\ast C\ast D+w_2\ast T+w_3\ast E)$(A3) ) in a global optimisation is responsible for distributing ${\mathbf{T}}_i$. Other AGVs are free.

Step 7: tasks in Class 1 return to Step 1. The task in Class 2 is loaded and distributed to the optimised AGVs.

Step 8: End.

Table A1. Notations for optimisation model.

$\mathbf{T}$	task set
$\mathbf{F}$	AGVs set
${\mathbf{T}}_i$	task i
${\mathbf{F}}_j$	AGV j
${\mathbf{C}\mathbf{P}}_i$ and ${\mathbf{C}\mathbf{P}}_j$	current position of task i and AGV j
${\mathbf{N}\mathbf{D}}_i$ and ${\mathbf{N}\mathbf{D}}_j$	destination of task i and AGV j
${\mathbf{W}}_i$	weight of task i
${\mathbf{V}}_i$	volume of task i
${\mathbf{S}\mathbf{V}}_j$	surplus weight of vehicle j
${\mathbf{S}\mathbf{W}}_j$	surplus volume of vehicle j
${\mathbf{w}}_i$	weight of objective variables
$T{s}_i$	starting delivery time of task i
$TF{s}_{ij}$	the starting time of AGV j with task i
$TF{c}_{ij}$	the completing time of AGV j delivering task i
$T{c}_j$	the completing time of AGV j
$T{d}_i$	the needed delivery time of task i
D	total distance of AGV
$D_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}$	distance of AGV with tasks
$D_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}$	distance of AGV without tasks
T	the total delivery time of AGV
$T_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}$	the time used for delivering the tasks for the AGV
$T_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}$	running time of the AGV without tasks
E	the total battery consumption of AGV
$E_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}$	the battery consumption of the AGV with tasks
$E_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}$	the battery consumption of the AGV without tasks
$V_{\mathrm{f}\mathrm{u}\mathrm{l}\mathrm{l}}$	the battery consumption per unit for the AGV with tasks
$V_{\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{t}\mathrm{y}}$	the battery consumption per unit for the AGV without tasks

Table A2. Position of stations in proof-of-concept prototype in a laboratory environment.

Station	$X\left(m\right)$	$Y\left(m\right)$
2	7.00	10.00
3 in	7.20	6.00
3 out	9.20	6.00
4 in	7.20	7.50
4 out	9.20	7.50
5 in	7.20	9.00
5 out	9.20	9.00
6	10.60	9.30
7	13.60	9.30
8	13.00	6.60
9	15.00	6.00
10	16.00	8.00

Table A3. Tasks for material handling in the proof-of-concept.

Cycle	Task	Task type	From station	To station	Window time start (s)	Material handling time (s)	Window time end (s)	Duration (s)
1	1	VA	(2)	(10)	0	50.00	230.00	230.00
	2	VA	(10)	(3) in	80.00	50.00	310.00	230.00
	3	NVA	(7)	(8)	0	27.00	400.00	400.00
	4	VA	(10)	(4) in	100.00	42.00	330.00	230.00
	5	VA	(10)	(5) in	100.00	37.00	330.00	230.00
2	6	VA	(3) out	(2)	400.00	30.00	630.00	230.00
	7	VA	(2)	(10)	500.00	50.00	730.00	230.00
	8	NVA	(10)	(9)	400	22.00	800.00	400.00
	9	VA	(10)	(5) in	450.00	37.00	680.00	230.00
	10	VA	(4) out	(2)	500.00	44.00	730.00	230.00
3	11	VA	(2)	(3) in	820.00	15.00	1050.00	230.00
	12	VA	(2)	(10)	800.00	50.00	1030.00	230.00
	13	NVA	(2)	(9)	800.00	35.00	1200.00	400.00
	14	VA	(10)	(3) in	850.00	50.00	1080.00	230.00
	15	VA	(10)	(4) in	950.00	42.00	1180.00	230.00
4	16	VA	(2)	(10)	1,200.00	50.00	1430.00	230.00
	17	VA	(3) out	(2)	1,350.00	30.00	1580.00	230.00
	18	NVA	(6)	(9)	1200.00	34.00	1600.00	400.00
	19	VA	(4) out	(2)	1200.00	44.00	1430.00	230.00
	20	VA	(5) out	(2)	1300.00	32.00	1530.00	230.00
5	21	VA	(10)	(3) in	1610.00	50.00	1,840.00	230.00
	22	VA	(10)	(4) in	1700.00	42.00	1930.00	230.00
	23	NVA	(7)	(8)	1600.00	27.00	2000.00	400.00
	24	VA	(10)	(5) in	1680.00	37.00	1910.00	230.00
	25	VA	(2)	(10)	1770.00	50.00	2000.00	230.00
6	26	VA	(3) out	(2)	2010.00	30.00	2,240.00	230.00
	27	VA	(4) out	(2)	2050.00	44.00	2280.00	230.00
	28	NVA	(9)	(7)	2000.00	20.00	2400.00	400.00
	29	VA	(10)	(3) in	2100.00	50.00	2330.00	230.00
	30	VA	(5) out	(2)	2100.00	32.00	2330.00	230.00
7	31	VA	(2)	(10)	2400.00	50.00	2630.00	230.00
	32	VA	(2)	(10)	2430.00	50.00	2660.00	230.00
	33	NVA	(8)	(2)	2400.00	37.00	2800.00	400.00
	34	VA	(10)	(3) in	2470.00	50.00	2700.00	230.00
	35	VA	(10)	(4) in	2570.00	42.00	2800.00	230.00
8	36	VA	(3) out	(2)	2800.00	30.00	3030.00	230.00
	37	VA	(4) out	(2)	2870.00	44.00	3100.00	230.00
	38	NVA	(6)	(7)	2800.00	27.00	3200.00	400.00
	39	VA	(3) out	(8)	2950.00	20.00	3180.00	230.00
	40	VA	(5) out	(2)	2900.00	32.00	3130.00	230.00
9	41	VA	(10)	(3) in	3200.00	50.00	3430.00	230.00
	42	VA	(10)	(5) in	3210.00	37.00	3440.00	230.00
	43	NVA	(8)	(6)	3200.00	31.00	3600.00	400.00
	44	VA	(10)	(5) in	3370.00	37.00	3600.00	230.00
	45	VA	(2)	(10)	3300.00	50.00	3530.00	230.00
10	46	VA	(10)	(4) in	3630.00	42.00	3860.00	230.00
	47	VA	(5) out	(2)	3600.00	32.00	3830.00	230.00
	48	NVA	(9)	(6)	3600.00	34.00	4000.00	400.00
	49	VA	(3) out	(2)	3700.00	30.00	3930.00	230.00
	50	VA	(4) out	(2)	3750.00	44.00	3980.00	230.00

Table A4. Optimal results of data-driven dynamic optimisation of tasks in material handling (VA: value added task, NVA: non value added task).

Cycle	Task	Task type	From station	To station	PickUp(s)	Material handling time (s)	Delivery (s)	Makespan (s)	Distance (m)	Battery consumption (W)
1	1	VA	(2)	(10)	52.00	50.00	102.00	325.00	18.16	162.50
	4	VA	(10)	(4) in	102.00	42.00	144.00
	5	VA	(10)	(5) in	186.00	37.00	223.00
	2	VA	(10)	(3) in	260.00	50.00	310.00
	3	NVA	(9)	(8)	350.00	27.00	377.00
2	6	VA	(3) out	(2)	537.00	30.00	567.00	263.00	14.69	131.50
	7	VA	(2)	(10)	567.00	50.00	617.00
	9	VA	(10)	(5) in	617.00	37.00	654.00
	10	VA	(4) out	(2)	684.00	44.00	728.00
	8	NVA	(10)	(9)	778.00	22.00	800.00
3	11	VA	(2)	(3) in	916.00	15.00	931.00	284.00	15.87	142.00
	12	VA	(2)	(10)	946.00	50.00	996.00
	14	VA	(10)	(3) in	996.00	50.00	1046.00
	15	VA	(10)	(4) in	1096.00	42.00	1138.00
	13	NVA	(2)	(9)	1165.00	35.00	1200.00
4	16	VA	(2)	(10)	1275.00	50.00	1325.00	325.00	18.16	162.50
	19	VA	(4) out	(2)	1352.00	44.00	1396.00
	17	VA	(3) out	(2)	1426.00	30.00	1456.00
	20	VA	(5) out	(2)	1488.00	32.00	1520.00
	18	NVA	(6)	(9)	1566.00	34.00	1600.00
5	21	VA	(10)	(3) in	1,650.00	50.00	1700.00	350.00	19.55	175.00
	24	VA	(10)	(5) in	1750.00	37.00	1787.00
	22	VA	(10)	(4) in	1824.00	42.00	1866.00
	25	VA	(2)	(10)	1893.00	50.00	1943.00
	23	NVA	(9)	(8)	1973.00	27.00	2000.00
6	26	VA	(3) out	(2)	2048.00	30.00	2078.00	337.00	18.16	168.50
	27	VA	(4) out	(2)	2122.00	44.00	2166.00
	29	VA	(10)	(3) in	2216.00	50.00	2266.00
	30	VA	(5) out	(2)	2298.00	32.00	2330.00
	28	NVA	(9)	(9)	2357.00	20.00	2377.00
7	31	VA	(2)	(10)	2435.00	50.00	2485.00	365.00	20.39	182.50
	32	VA	(2)	(10)	2535.00	50.00	2585.00
	34	VA	(10)	(3) in	2585.00	50.00	235.00
	35	VA	(10)	(4) in	2677.00	42.00	2719.00
	33	NVA	(8)	(2)	2763.00	37.00	2800.00
8	36	VA	(3) out	(2)	2890.00	30.00	2920.00	290.00	16.20	145.00
	37	VA	(4) out	(2)	2964.00	44.00	3008.00
	40	VA	(5) out	(2)	3040.00	32.00	3072.00
	39	VA	(3) out	(8)	3102.00	20.00	3122.00
	38	NVA	(6)	(9)	3153.00	27.00	3180.00
9	42	VA	(10)	(5) in	3306.00	37.00	3343.00	289.00	16.15	145.50
	41	VA	(10)	(3) in	3380.00	50.00	3430.00
	45	VA	(2)	(10)	3445.00	50.00	3495.00
	44	VA	(10)	(5) in	3495.00	37.00	3532.00
	43	NVA	(8)	(6)	3564.00	31.00	3595.00
10	46	VA	(10)	(4) in	3680.00	42.00	3722.00	320.00	17.88	160.00
	47	VA	(5) out	(2)	3751.00	32.00	3783.00
	49	VA	(3) out	(2)	3813.00	30.00	3843.00
	50	VA	(4) out	(2)	3887.00	44.00	3931.00
	48	NVA	(9)	(6)	3966.00	34.00	4000.00

Figure A1. Industrial Internet of Things digital servitization for smart production logistics device profile.

Architecture for Industrial Internet of Things-enabled digital servitization in smart production logistics including a sensing, networking, and applications layer.

Figure A2. Industrial Internet of Things digital servitization for smart production logistics activity profile.

A data model for multichannel communication including IIoT modeling profiles, data bases, production logistics tasks, and services.