Since Industry 4.0 technologies were introduced, manufacturing processes have continued to evolve with, for example, Artificial intelligence (AI), the industrial Internet of things (IIoT), robotics, digital twin technology, and other cutting-edge innovations to empower intelligent manufacturing processes. Nonetheless, there remain obstacles to integration within and optimization of systems of systems. An open-source micro-service architecture can alleviate these hurdles. To demonstrate interoperability between technologies, we have chosen to conduct our research using a physical training factory from Fischertechnik with diferent digital twins (DTs). In this paper, one digital replica of the physical factory is implemented using a three-dimensional CAD tool (Siemens NX-MCD), while the second is an information-centric digital twin created using Eclipse Ditto. The work is focused on creating digital twins of individual sections of the full factory to achieve interoperability, incorporating open-source frameworks to achieve interoperability and systems binding at run-time, and adapting multiple protocol communications in twins. This work is in progress and foresees two scenarios as its further development. The first is the digital twin's ability to facilitate predictive maintenance. Secondly, the use of an information-centric digital twin in the loop to constantly monitor the production process in the event of connectivity issues.
dustrial asset maintenance issues. Being introduced in From mechanization (IR1.0) to electrification (IR2.0), dig- the early 2000s and later revisited by NASA in 2012, the italization (IR3.0), and IT-driven smart solutions (IR4.0), term “digital twin" became a staple in industry 4.0 [4]. the industrial revolution (IR) roadmap has witnessed It is defined as a digital copy of an object, system, or changes drastically [1]. Adding smart technologies, es- process that communicates with its physical counterpart pecially during the 4th revolution, is a way to achieve in real time to serve specific applications [ 5]. It is a tool its fundamental design principles, such as decentraliza- to bridge the gap between the physical and digital worlds. tion, horizontal and vertical integration, interoperability, Integrating the digital world into the loop of a production smart factory, smart product, product & service individ- line, the need for manual labor goes down, and mobile ualization, real-time capability, service orientation, and management becomes possible [6]. A digital twin has sevvirtualization [2]. Technology like the industrial internet eral definitions and interpretations, but one thing needs of things (IIoT), big data and artificial intelligence (AI), to be consistent, the digital model needs to be integrated. simulation & modeling (digital twins), additive manufac- Without integration between the two worlds, what is turing, cloud data & computing, autonomous robotics, left, is just a digital model. As digital twin technology augmented & virtual reality (AR/VR), cybersecurity, and grows in popularity, more work must be put into making so on are all contributing to the increasing intelligence of it flexible and adaptable to meet the needs of diferent smart factories [3]. With these technological paradigms industries. Some of the identified bottlenecks are model and design principles in mind, this study has focused creation, interoperability, and data synchronization [7]. on improving the interoperability challenges by combin- Interoperability is one of the important aspects of Ining the physical and digital twins of a physical training dustry 4.0. It is the ability of two or more systems or factory. components to exchange information and to use the information that has been exchanged [8]. With the usage TOKuTluP, 2F0in2l3a: nAdnnual Symposium for Computer Science 2023, June 13-14, of a broad range of protocols and technologies in manu* Corresponding author. facturing sectors, interoperability between the various $ sarthak.acharya@oulu.fi (S. Acharya) industrial components has grown to be a significant chal0000-0001-8774-9433 (S. Acharya) lenge [9]. Many industries use the proprietary solution © 2023 Copyright for this paper by its authors. Use permitted under Creative Commons License to achieve integration between diferent components in CPWrEooUrckReshdoinpgs IhStpN:/c1e6u1r3-w-0s.o7r3g ACttEribUutRion W4.0oInrtekrnsahtioonpal (PCCroBYce4.0e).dings (CEUR-WS.org) lfexibility, DT-scalability, DT-reusability, and many more [16] are combinedly taken into consideration to find the research gaps related to digital twins. In this article, the main research concern has focused on the interoperability aspects of digital twins. A six-step approach, suggested by Deloitte Insights [5], was followed to deploy the digital twin of the physical factory, also presented in ifgure 1. The work presented in the paper covers the create, communicate, and aggregate steps (marked green in the figure 1), whereas the other steps are still in progress (marked yellow in the figure 1). However, the complete action of interoperability (Act) is not achieved yet and hence marked as ’red’ in the figure. The main highlights of the work are as follows: • Development & Demonstration of digital twin of
a manufacturing unit. • Incorporation of the open-source digitalizing framework (Eclipse Arrowhead framework) to achieve interoperability and systems binding at run-time. • Adaptability of multiple protocol communica
tions • Creating digital twins of individual sections of the entire assembly to demonstrate interoperability among digital twins in the future. the factory. Thus, the requirement for open-source interoperability frameworks and platforms has emerged.
For the clarification of the role of DT in several sectors, profuse consortiums and committees like Industrial Internet Consortium (IIC) [10], Digital Twin Consortium [11], and Platform Industry 4.0 [12] provided a definition.
Followed by the denfiition, several aspects that must be considered are information models, connectivity to the physical twin, data ingestion techniques, APIs, security, In this paper, the background and motivation for the and interoperability of DT. Industry 4.0 associated orga- research work were mentioned in section II. Section III nizations [13, 14] pivoted on industrial entities’ adoption has demonstrated the experimental setup. Results and of DT technologies based on existing industry 4.0 stan- discussion related to the research work were outlined dards. There is an inexorable need for Open standards in section IV. Section V has summarised the work and and open source implementations as well as the adoption proposed two scenarios as the extension of this research of these standards to expedite DT development within or work. across diferent enterprises [ 14]. Many open-source communities such as Apache Software Foundation, Eclipse 2. Background IoT, Linux Foundation, and many more have gained the role of a valuable technology supplier to the smart man- 2.1. Creation of Digital Twins ufacturing and software industry. Present practice in the industry is that DT platforms are built as closed sys- 2.1.1. Graphical Type tems (not open source) limiting the overall primacy of Digital twins are widely used throughout the industry. smart manufacturing [15]. Therefore, this article steers However, the meaning of the word varies a lot depending the discussion of open-source DT solutions to make them on the context. One rule that applies to all digital twins more accessible to a broad academic and industrial re- is the interaction with the physical twin. A graphical search community. There are multiple options available digital twin adds, in addition to the aspects of a digital to build/deploy DTs as cloud enterprise solutions such as twin of another kind, the visual simulation of its physical Azure DTs, AWS DTs, and IBM DTs; industrial solutions twin. such as Bosch’s DT solution, and GE Predix; and vendor Properties such as gravity, collision, mass, and inertia agnostic OSS frameworks such as Eclipse Ditto, Swim need to be considered. Depending on the main objective OS, iModel JS. of the digital twin the range of inclusion amongst the
Studies on industry 4.0 practices, existing interop- physical properties will vary. Having no connection to erability scenarios, and the introduction of various the real world the model can not be considered a digital open-source platforms, in the context of DTs, have twin and will instead exist solely as a virtual model of its opened various attributes such as DT-modularity, DT- physical counterpart [17]. interchangeability, DT-accuracy, DT-interoperability, DT
erability between the physical twin and DT with the goal of its scalability and re-usability. Simulating a digital twin in real-time requires a lot of computational power and will hence demand simplifications within the model [18].
There are many diferent programs suited to creating a digital twin, all with each its own niche [19]. Ansys has properties that benefit computational mechanics while NX and Catia have their main use when working with mechanical construction. NX MCD is a good mechatronics application in NX that allows the model to connect to external signals from varying sources and has therefore been used in this use case [20].
Because of technological advancements, manufacturers are no longer limited by the physical design of the machines. Instead, hybrid simulation environments give an advantage in visualizing overall processes and end products before production. Cost-cutting and eficiencyboosting strategies are crucial for businesses of all stripes in today’s highly competitive global market. DTs, or High-Fidelity virtual protocols, are gaining popularity due to the following aspects: 1. Real-time monitoring enables quick task planning & building accurate testing environments. 2. Predictive Maintenance of the real assets. 3. Allow the operators to get hand-in-experience on virtual machines before handling the real machines. 4. Pre-diagnosis and fault detection of goods before their production. 5. Optimization of the performances of physical assets.
tools such as NVIDIA Omnipresent, Microsoft Azure, Eclipse Ditto, and many more [21]. Rather than focusing on 3D/2D visualizations, such platforms make The real challenge for an original equipment maninformation-centric models that mimic the attributes and ufacturer (OEM) is to find a standardized method for features of a physical object or process. Such digital twins assembling various physical components from multiple can also be created using the application programming vendors. If digital twins of each part can interact the interfaces (APIs) [22]. Some of the digital twins were same way they do in the real world, the existing manubuilt using an open sensor manager (OSEMA) and open facturing system will be much more robust, advanced, platform communication unified architecture (OPC-UA)- time-consuming, and error-prone. As a result, the interGraphQL wrapper [23]. In some cases, document-based operability of DTs is critical for developing such multidigital twins are also created to control smart factories disciplinary engineering competence. Interoperability [24]. The advantages of these digital twins over CAD- here means that data and information can be transferred based twins include their flexibility, modularity, and scal- from the physical twin to the digital twin, from the digital ability. In this work, Eclipse Ditto [25], is used to realize twin to another digital twin, and from the digital twin to the information-centric twin of the factory. the real assets, as depicted in the figure 2. PT, in figure 2,
Most industries and component suppliers use propri- is the Physical Twin. etary engineering tools and information processing systems for product development in IIoT systems. Intercommunication or interoperability among diferent systems 2.2. Need of DT Interoperability or applications is thus required to exchange mutual infor- 2.3. Eclipse Arrowhead Framework mation. So far, only a few works of literature addressing interoperability issues have been discovered. In literature The Eclipse Arrowhead Framework is an open-source, [26], a customized mapping model is proposed, which local cloud-based industrial framework that provides intranslates the proprietary ABB Ability digital twins to the teroperability solutions in Industry 4.0.[28]. This is built Asset Administration format. It targeted to enable inter- on SOA (Service Oriented Architecture) and promotes operable digital twins by transforming the information late binding, loose coupling, cyber security, and multimodels. It demonstrated a real-world application using stakeholder integration. The framework allows run-time ABB devices as a use-case with some facets of interoper- communication between systems within a local cloud or ability achieved through files and APIs using model trans- between systems registered in diferent local clouds. To formation. The implementation, however, still needs to facilitate interaction between systems, it provides three achieve complete digital twin interoperability. Recently, mandatory core systems. The mandatory core systems literature [27] has proposed twins interoperability arti- are: facts to establish communication between products and production systems. Nevertheless, the question remains: ‘How to represent products, components, production resources, and relevant infrastructures virtually by their digital twins and make them inter-operate.’
currently being ofered, 2. the Authorization system that controls system-tosystem authorization at a detailed level for secure service exchange, 3. the Orchestrator that enables the consumer appli- signals and mapping these to the external sources of sigcation to discover the required service endpoint nals the digital twin is bound to work like the physical at run time. counterpart [6].
3.1. About the Factory
presence of objects and/or functionalities that are harder to determine than simple movements. The factory includes both sensors and actuators that somehow need to be represented in the digital twin [29].
framework also ofers supporting core systems, such as the Gateway and Gatekeeper systems.
Gmbh [29], includes four stations, the high-bay warehouse, the sorting line, a vacuum gripper robot, and a 3.1.1. Sensors multi-process station. Parts of the physical factory have been disassembled to allow measurements to be taken. The factory holds two diferent kinds of sensors, color The measurements are brought into the NX modeling sensors, and light barrier sensors. The color sensor that application and applied to sketches. 3D models are ex- this particular factory holds is located in the Sorting Line truded and generated from the sketches, this way the section of the factory [10]. The goal of the sensor is to digital model will hold the same proportions as the phys- determine in which pocket the product should end up. ical counterpart [6]. The light barrier sensors act as position sensors to update
The individual parts created in the modeling applica- the position of the product and make sure that there still tion are brought together with the help of the NX assem- is a product in play. bly application. By creating an assembly the parts can be copied and constrained together. The assembly is divided 3.1.2. Actuators into four sub-assemblies representing the four stations Actuators in the factory are used to mark certain posiin the physical factory. By dividing the main assembly it tions of the machines. An example of the use of actuators is easier to maintain the safety of the assembly and not is to home a machine. There is a certain position that is constrain it in a wrong fashion [6]. always marked with actuators [10]. The machine moves
By bringing the assembly to NX MCD, NX allows the until it notices an actuator being pushed, it then knows user to assign communication with external sources of that the machine in question is home and does now concommunication. To assign physical aspects to the model tain a sort of reference point. There are actuators on rigid bodies and collision bodies are defined and con- every axis for every moving machine in the factory. strained by joints created by the user. Creating internal
ribbon cable to a programmable logic controller (PLC) that handles the logic operating the factory. With the use of Siemens’s totally integrated automation (TIA) or other Structure Language Text program, the logic is compiled and downloaded to the PLC. In this case, to compile and run codes, TIA Portal is used. However, the logic and interfaces used to program the PLC may vary since many companies use their proprietary software.
4.1. Digital Twin Modelling
trated in figure 3. As mentioned in section Experimental Setup the factory consists of four stations, and like so, the digital visualization is also put together. The model receives signals from a simulated PLC which runs the same code as the physical twin.
to simulate the digital twin with respect to the physical factory. This was achieved through the external signal configuration feature in NX MCD. There are multiple technologies like OPC UA, OPC Data Access (OPC DA), Shared Memory network (SHM), Matlab, PLC Simulation (PLCSIM) Advanced, Transmission Control Protocol (TCP), UDP, Profinet, Functional mock-up Unit (FMU), and Virtual Machine (CMVM) that can be used to map signals to the DT. In our experiment, we used protocols like OPC UA, TCP, and PLCSIM Advance to simulate the DT as shown in figure 6.
plications, and it is, therefore, crucial to making the simulation as fast as possible to reduce the amount of delay. 4.2.1. OPC UA As a result, the conveyor belt is heavily simplified. Figure 4 illustrates accepted simplifications for the application.
nication between diferent devices in the industrial sector 4.1.2. Information-centric View and IoT applications [30]. The high-bay warehouse part of the factory was connected to a Siemens Simatic ET Eclipse Ditto is used to create the factory’s information- 200SP PLC with an OPC UA server in it. The sensor and centric digital twin. The only part of the assembly that actuator signals from the PLC were mapped to individual has been worked on so far is the high-bay warehouse. nodes of the OPC UA server. Once the OPC UA server is The steps for making the digital twin are shown in the added to the signal configuration and the input-output architectural diagram in figure 5. The data models for the nodes are mapped to the sensors and actuators in the 3D factory parts were written in javascript object notation Transmission Control Protocol, or TCP, is a communications standard that enables computer hardware and software to exchange messages over a network. It is meant to transfer packets across the internet and make sure that data and messages are successfully sent through networks. The digital model of the factory is able to communicate through TCP. Once a TCP server is created the model is able to act as a TCP client and connect to the data. The structure of the data sent through the TCP protocol is handled as a byte array where each byte is related to a specific I/O. By mimicking the logic created in structured language the digital model operates in a similar fashion.
The virtual simulation acts identically to a real PLC and can connect to a digital model just as a physical machine can connect to a PLC. Taking advantage of PLCSIM Advanced it is possible to make sure that the two twins operate identically. 4.3. Incorporating Open source IoT
Platform To achieve interoperability among twins (both physical and digital twins), our solution is to extract microservices out of the digital twins into a common IoT platform. We aim to use the Eclipse Arrowhead Framework as the IoT platform that enables interoperability between systems operating on diferent technologies and protocols. In our use case, each twin will be an Arrowhead complaint application system, that ofers micro-services to read the status of all sensors and actuators. Now, with the Arrowhead service exchange architecture, the Arrowhead complaint application systems of the twins can exchange information irrespective of the protocols they are using.
An inter-cloud communication between the Arrowhead systems is demonstrated in figure 7. In both local clouds, the mandatory core systems such as the ServiceRegistry, the Authorization System, and the Orchestrator are running along with the supporting core systems Gatekeeper and Gateway. DT Cloud 1 has an application provider system for the document-based DT as a proxy from Eclipse Ditto, that keeps track of the historical data for the factory’s sensors and actuators. DT Cloud 2 has 3 application systems, including one for the physical twin, one for the graphical DT, and a consumer SCADA system to monitor the factory. Both the physical twin system and the graphical DT system can provide microservices to read the status of sensors and actuators and write values to the actuators. The SCADA system can communicate with all 3 provider application systems and keep track of all twins and make decisions based on the status of the factory and DT. 4.4. Full Assembly of the Factory
The full assembly can be divided into four parts: conveyer belt, multi-processing station, vacuum gripper robot, and high bay warehouse. Each piece has its replica, as shown in figure 8. Each part can be communicated with its physical twin using the earlier protocols. The further target is to incorporate diferent communication protocols for each piece and achieve interoperability among the digital twins. The document-based twin (ditto model) of each component will also be in-loop to demonstrate interoperability between diferent types of DTs.
monitor diferent processes, especially the production process in the event of connectivity issues i.e. with minimal connectivity it can propagate the message across and the system will have an updated information-centric view of the actual physical system at all times.
building a digital twin of a physical training factory. A Acknowledgments six-step approach is adapted to accomplish the use-case scenario, as shown in figure 1. The digital replica of The research work is carried out under the project ’Operathe physical factory is created using two open-source tional eXcellence by Integrating Learned information into tools, one with three-dimensional CAD (Siemens NX- the AcTionable Expertise (OXILATE) sponsored by ITEA MCD), which is of graphical type. We also created an 3.0. This is a collaborative work between the Univerinformation-centric view or document-based digital twin sity of Oulu, Finland, and LuleåUniversity of Technology of the same factory using Eclipse Ditto. The communica- (LUT), Sweden. The authors would like to express their tion in the digital twin framework has been established by gratitude to Prof. Jerker Delsing for using the Laboratory incorporating open-source frameworks, system binding and resources at LUT. The authors would like to thank at run-time, and adaptability of multiple protocol commu- Fischetechnik Gmbh for providing the training factory. nication. Digital twins and predictive maintenance are the perfect pair as DT technology ushers the exact replica in the digital format of a process, product, or service. As a References further extension of this work, we propose two use-case scenarios. With the emergence of IoT and DTs, there is a need for a predictive maintenance model that optimizes the maintenance cycle and balances between a corrective and preventive maintenance approach. Another use-case scenario of the information-centric DT is to constantly [3] G. Erboz, How to define industry 4.0: main pillars of cial Power Systems Europe (EEEIC/I&CPS Europe), industry 4.0, Managerial trends in the development IEEE, 2022, pp. 1–6.
of enterprises in globalization era 761 (2017) 767. [18] J. Dangermond, M. F. Goodchild, Building geospa[4] F. Tao, H. Zhang, A. Liu, A. Y. Nee, Digital twin tial infrastructure, Geo-spatial Information Science in industry: State-of-the-art, IEEE Transactions on 23 (2020) 1–9.
industrial informatics 15 (2018) 2405–2415. [19] M. Liu, S. Fang, H. Dong, C. Xu, Review of digital [5] A. Parrott, L. Warshaw, Industry 4.0 and the digital twin about concepts, technologies, and industrial twin, in: Deloitte University Press, 2017, pp. 1–17. applications, Journal of Manufacturing Systems 58 [6] O. Wintercorn, Interfacing digital and physical (2021) 346–361.
twins with a microservice architecture, 2022. [20] N. Wittig, Digitalization: laser metal deposi[7] F. Pires, A. Cachada, J. Barbosa, A. P. Moreira, tion—the future of spare parts and repairs for inP. Leitão, Digital twin in industry 4.0: Technologies, dustrial steam turbines, in: Turbo Expo: Power for applications and challenges, in: 2019 IEEE 17th In- Land, Sea, and Air, volume 51173, American Society ternational Conference on Industrial Informatics of Mechanical Engineers, 2018, p. V008T29A002. (INDIN), volume 1, IEEE, 2019, pp. 721–726. [21] Z. Su, S. Guo, M. Dai, T. H. Luan, Y. Liu, et al., A [8] A. Geraci, IEEE standard computer dictionary: survey on digital twins: Architecture, enabling techCompilation of IEEE standard computer glossaries, nologies, security and privacy, and future prospects IEEE Press, 1991. (2023). [9] A. Zeid, S. Sundaram, M. Moghaddam, S. Kamarthi, [22] R. Ala-Laurinaho, et al., Api-based digital twinsT. Marion, Interoperability in smart manufacturing: architecture for building modular digital twins folResearch challenges, Machines 7 (2019) 21. lowing microservices architectural style (2021). [10] I. I. Consortium, Industry internet of [23] J. Autiosalo, R. Ala-Laurinaho, J. Mattila, M. Valtothings vocabulary, 2022. URL: https: nen, V. Peltoranta, K. Tammi, Towards integrated //www.iiconsortium.org/wp-content/uploads/ digital twins for industrial products: Case study on sites/2/2022/04/Industry-IoT-Vocabulary.pdf . an overhead crane, Applied Sciences 11 (2021) 683. [11] D. T. Consortium, Digital twin consortium is the [24] J. Mattila, R. Ala-Laurinaho, J. Autiosalo, P. Salmiauthority in digital twin, 2020. URL: https://www. nen, K. Tammi, Using digital twin documents to digitaltwinconsortium.org/about-us/. control a smart factory: Simulation approach with [12] B. Boss, S. Malakuti, S.-W. Lin, T. Usländer, E. Clauer, ros, gazebo, and twinbase, Machines 10 (2022) 225.
M. Hofmeister, L. Stojanovic, B. Flubacher, Digital [25] M. Kherbache, M. Maimour, E. Rondeau, Digital twin and asset administration shell concepts and twin network for the iiot using eclipse ditto and application in the industrial internet and industrie hono, IFAC-PapersOnLine 55 (2022) 37–42. 4.0, Plattform Industrie 4 (2020) 13–14. [26] M. Platenius-Mohr, S. Malakuti, S. Grüner, [13] O. I. . Alliance, Open industrial partners and J. Schmitt, T. Goldschmidt, File-and api-based associated organizations, 2022. URL: https:// interoperability of digital twins by model openindustry4.com/. transformation: An iiot case study using asset [14] E. Barnstedt, B. Boss, E. Clauer, D. Isaacs, S. Lin, administration shell, Future generation computer S. Malakuti, P. Van Schalkwyk, T. W. Martins, Open systems 113 (2020) 94–105. source drives digital twin adoption, IIC J. Innov [27] S. Acharya, M. Hanif, T. Päivärinta, Development (2021). of pragmatic research artifacts on digital twin tech[15] V. Damjanovic-Behrendt, W. Behrendt, An open nology using dsr theories, 2022. source approach to the design and implementation [28] J. Delsing, Iot automation: Arrowhead framework, of digital twins for smart manufacturing, Interna- Crc Press, 2017. tional Journal of Computer Integrated Manufactur- [29] Fischertechnik, Factory simulation, proding 32 (2019) 366–384. uct code: 536634, 2022. URL: https: [16] J. Moyne, Y. Qamsane, E. C. Balta, I. Kovalenko, //www.fischertechnikwebshop.com/en-gb/ J. Faris, K. Barton, D. M. Tilbury, A requirements factory-simulation-24v-536634-en-gb. driven digital twin framework: Specification and [30] S. Grüner, J. Pfrommer, F. Palm, Restful industrial opportunities, IEEE Access 8 (2020) 107781–107801. communication with opc ua, IEEE Transactions on [17] L. Pugi, L. Berzi, L. Franchi, A. Casazza, G. Mattei, Industrial Informatics 12 (2016) 1832–1841.
R. Fiorenzan, I. Domina, Preliminary design and simulation of an hybrid-parallel, fixed-wing uav with eight-rotors vtol system, in: 2022 IEEE International Conference on Environment and Electrical Engineering and 2022 IEEE Industrial and Commer