Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Der gegenwärtige Prozess der Konstruktion mechanischer Systeme ist arbeitsintensiv und ineffizient, wobei manuelle Suche nach Komponenten und späte Leistungsanalysen zu suboptimalen Konstruktionen führen. Das Co-Des-Rahmenwerk adressiert diese Probleme, indem es den Designprozess automatisiert und eine bidirektionale Zusammenarbeit zwischen Erstausrüstern (OEM) und Systemintegratoren ermöglicht. Durch die Standardisierung der Bauteilbeschreibungen und die Nutzung des digitalen Twin Web ermöglicht das Rahmenwerk eine automatisierte Komponentenauswahl, Systemanalyse und -optimierung. Das Rahmenwerk wurde anhand einer Fallstudie zur Entwicklung eines Windkraftantriebs demonstriert, die sein Potenzial zur signifikanten Verbesserung der Konstruktionseffizienz und Systemleistung aufzeigte. Das Co-Des-Rahmenwerk hat das Potenzial, den Konstruktionsprozess mechanischer Systeme zu revolutionieren und zu effizienteren, langlebigeren und leistungsfähigeren Systemen zu führen.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
Analyzing multi-vendor mechanical system designs requires a significant amount of manual work, resulting in a design paradigm where analysis is conducted only after the design is locked and components are selected. This leads to a suboptimal design with compatibility issues, over-dimensioned components, inferior performance, poor energy efficiency, and a lack of collaboration between OEMs (original equipment manufacturers) and system integrators. To overcome these issues, this paper proposes Co-Des (collaborative design) framework for automated and collaborative multi-vendor system design. The framework relies on standardized digital twin documents (DTD) of system designs, components, and analyses. The discoverability and distribution of these DTDs are enabled with digital twin web (DTW). Co-Des framework allows for finding suitable components for the design task by automatically running selected analyses employing component digital twins. In addition, OEMs can provide customized components for system integrators using the initial system design defined in the system design DTD. The use of the Co-Des framework was demonstrated with a windmill powertrain design use case, and the applicability of the automated assembly analysis for component selection was verified with performance measurements. The adoption of the proposed framework will lead to a paradigm shift from manual and siloed work relying on the exchange of PDFs to a more automated and collaborative design of mechanical systems. The adoption rate is defined by the willingness of system integrators to publish their initial system designs and OEMs their components as public digital twins.
Communicated by T. Clark, S. Zschaler, V. Kulkarni, and D. E. Khelladi.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
Mechanical systems are typically built from components supplied by multiple vendors, which makes component integration a crucial phase of the design process. Currently, searching for suitable components and ensuring system components’ compatibility involves a substantial amount of manual work, such as asking for component details via emails or building component models, resulting in a time-consuming and inefficient design process. A large amount of manual work is also needed for system analysis. Therefore, it is usually conducted only after all components are selected, causing compatibility issues and leading to suboptimal performance. The compatibility issues are caused, for example, by interfering excitations and the natural frequencies of components, and the suboptimal performance is due to oversized components as large safety factors are needed to be used when accurate validation of the system is not possible. Furthermore, there is a lack of collaboration between OEMs (original equipment manufacturers) and system integrators: system integrators try to find suitable components from the product catalogs of OEMs instead of OEMs offering their already optimized components for initial system designs defined by system integrators.
Fig. 1
This paper proposes Co-Des framework for automated and collaborative system design. The framework relies on standard descriptions of designs (digital design template), analyses (analysis description) and components (component description), discoverability of the descriptions via digital twin web, and cloud analysis services
To elaborate on the research problem, we highlight the following issues in the current mechanical system design paradigm:
1.
There are several competing OEMs for any component of a complex mechanical system, making finding the most suitable components laborious. A good example of a complex mechanical system is an electric motor-driven powertrain.
2.
The components of the mechanical system need to be analyzed and simulated together to ensure compatibility and to evaluate the quasi-static and dynamic behavior of the system. This is required, for example, in the torsional vibration analysis of powertrains.
3.
OEMs do not provide component models or the models are incompatible with each other, as there are not widely used standards for model-share. In addition, components and their information cannot be automatically searched because semantic-based component descriptions are lacking. This leads to a situation in which the system-level analysis and simulation are not easily (or at all) conducted using the models from multiple OEMs.
4.
The inability to use existing models to conduct analyses leads to a vast amount of overlapping work, especially from the system integrator. The overlapping analysis and modeling work are done with incomplete information collected analogically piece by piece from digital models, emails, and data sheets, which leads to inaccuracies.
5.
Analysis and simulation are conducted only after components have been selected, which leads to suboptimal design and causes compatibility issues.
To overcome these issues, a more collaborative approach to system design needs to be adopted. The previous work on collaborative design has mainly focused on how to technically implement (co-)simulation [1], define system structure [2], exchange data and models, or analyze multi-component systems (maritime [3], automotive [4], and gears [5]). In addition, the adoption of digital twins to improve the product design process has been proposed in several papers [6, 7]. Digital twins have been used to create immersive visualization of workstation design [8], enable simulation [9], or provide data to the designers [6, 10]. They have also enabled information sharing during the design process [11], virtual verification [12], and collaborative design [13].
Anzeige
However, the previous research does not address the discoverability of component digital twins or automated system-level analysis based on standardized component descriptions and models in mechanical system design. Furthermore, current model databases usually rely on general versions of the components instead of providing a selection of different component versions from several OEMs. Therefore, the automated design process based on openly available component candidates is not currently possible. This can lead to suboptimal system designs and a lack of collaboration between system integrators and OEMs in the early design phase. On the other hand, the current lack of collaboration can also stem from the absence of real willingness to collaborate, for example, due to fear of leaking confidential product details or seeing no financial incentive.
The goal of this work is to enable more efficient design of multi-vendor mechanical systems. The work aims to enable bidirectional collaboration and automated analysis of system behavior in the domain of machine and ship design, especially focusing on large and complex powertrains with several components, such as marine vessel powertrains. Furthermore, the focus of the automated analyses is to examine how powertrains respond to various loads, such as determining the maximum torsional vibration of a cruise ship powertrain induced by ice loads.
To enable collaboration and automate finding suitable designs in complex mechanical system design, this paper proposes the Co-Des (Collaborative Design) framework. The framework relies on standardized systems’, components’, and analyses’ descriptions with digital twin documents (DTDs) [14] and their discoverability and distribution via digital twin web (DTW) [15, 16]. The overview of applying the Co-Des framework in system design is illustrated in Fig. 1. First, the system integrator defines the initial system design as a standardized digital design template (DDT), including requirements for the components and analyses that are conducted for the system. Thereafter, components that fulfill these requirements are searched from DTW. Component properties are described in standardized component description (CD) documents. When suitable components are found, all possible assemblies from these components are formed, and the assemblies are analyzed using selected analysis services. The properties of these analysis services, including inputs and outputs, are stored in analysis description documents. Finally, the system integrator selects the best design candidates based on the analysis results.
The proposed Co-Des framework aims to transform the current mechanical system design paradigm. It contributes toward a more automated and collaborative design process, which has the potential to improve mechanical systems by reducing manufacturing and operational emissions, enhancing performance, and improving durability. The Co-Des framework is designed to be flexible and able to support the design of mechanical systems ranging from simple to complex in various domains. The main contributions of the paper are as follows:
1.
Introduce Co-Des framework for automated and collaborative design of mechanical systems.
2.
Publish an open-source implementation of the framework in GitHub [17].
3.
Apply the developed framework to a windmill powertrain design case considering torsional vibration.
4.
Measure the time to find suitable windmill powertrains to demonstrate framework applicability to a real-world use case.
Anzeige
2 Background and related work
This chapter presents the background for this work in Sect. 2.1, which illustrates the role of digital twins, and Sect. 2.2, describing the need for ontologies in this work. Thereafter, Sect. 2.3 reviews the current literature on collaborative design enabled by digital twins and highlights the research gaps.
Fig. 2
Relationships between formats and ontologies that are relevant in describing digital twins and enable collaborative design. Bolded ones are used in this work, and the others are provided for reference
Digital twins (DTs) are virtual representations of real-world entities. These virtual representations can have various sets of features based on the use case, such as simulation capabilities or analysis services [18]. The definition by the Industrial Internet Consortium [19] expresses this variety of digital twins well: “A digital twin is a formal digital representation of some asset, process or system that captures attributes and behaviors of that entity suitable for communication, storage, interpretation or processing within a certain context.” In addition, views on digital twins can also be divided into modeling-oriented and information-management-oriented [20], the latter of which is adopted by this paper. Nevertheless, information management also includes the management of product models, and these two views are not mutually exclusive.
Digital twins play a key role in the digitalization of industry and Industry 4.0, and they were considered the key drivers toward the smart manufacturing paradigm in the review of digital twins in industrial applications by Tao et al. [21]. The use of digital twins has also been proposed to improve the product design process [7, 21]. They can be employed as a part of data-driven design, enabling feedback from the field and customers to the designers of new-generation products [6]. In addition, digital twins allow virtual prototyping [7], simulation [9], and immersive visualizations [8]. However, these previous studies do not adopt the full potential of networked digital twins enabling the discoverability of suitable components, automated system analysis, and collaboration between system integrators and OEMs.
Creating a network of digital twins requires the standardization of digital twin descriptions. There are already several attempts to standardize descriptions of digital twins, and four major candidates for that can be identified: Asset Administration Shell (AAS) by Plattform Industrie 4.0 [22], Web of Things (WoT) Things Description (TD) [23], Next Generation Service Interfaces-Linked Data (NGSI-LD) by ETSI [24], and Microsoft Digital Twin Definition Language (DTDL) [25]. Standardization of digital twins enables their interoperability, discoverability, and accessibility, as stated in the review of current standards by Jacoby and Usländer [26].
The previous research of the authors has also emphasized the need for standardized descriptions of digital twins [14], and we have demonstrated the use of the description documents in factory automation [27] and extended reality application development [28] use cases. In this paper, we chose to use digital twin documents (DTDs), which present a generic example of digital twin descriptions. This choice was made to demonstrate which information is strictly necessary to enable an automated design process, reduce extraneous information required by current digital twin standards, and avoid making an implementation-specific solution. Furthermore, the information presented in DTD is straightforward to include as part of current standards, such as WoT TD or AAS.
This paper relies on digital twin web to enable the discoverability of the DTDs [15] and make these documents available. A web for digital twins has also been proposed in [29] to create a digital twin ecosystem, make digital twins interoperable, and enable knowledge representation. The implementation of the framework in the case study adopts Twinbase, which is open-source server software for the digital twin web [16], analogous to how the Apache HTTP Server is an open-source server software for the World Wide Web. Twinbase was selected as a development platform as it is an easy-to-set-up solution for hosting DTDs, and it enables traceability of changes in DTDs. However, the Co-Des framework is not dependent on the features of Twinbase, and any web service capable of hosting documents could be used.
2.2 Ontologies for collaborative design
Ontologies are key to making digital twin descriptions machine-readable and interoperable. They describe knowledge in selected areas. An ontology may define classes of knowledge, include instances of real-world entities described according to the classes, and describe relations between those. Ontologies may refer to each other to form a large network of knowledge, enabling the discovery of new knowledge from the Semantic Web. Semantic Web is an extension of the World Wide Web with the goal of making Internet data machine-readable [30]. There are several standards for implementing the Semantic Web, such as RDF (Resource Description Framework) [31], OWL (Web Ontology Language) [32], and JSON-LD (JavaScript Object Notation for Linked Data) [33].
The use of ontologies has remained foremost an academic activity, although for example internet search providers have started enriching search results with the knowledge found from JSON-LD documents embedded on web pages. The search providers follow the Schema.org ontology [34], which is available publicly on a human-readable website, enabling developers to adopt it effortlessly in their software. This kind of shared information model can also enable new methods for collaborative design.
Currently, many of the ontologies required for industrial applications are missing, but some have started to emerge. The Schema.org ontology provides classes for a wide array of basic knowledge, for example for products, and the GS1 Web Vocabulary [35] provides more detailed classes to describe consumer goods. Engineering assets have been described in the SAREF ontology [36] and its extensions. There is also an ECLASS standard [37] which does not follow Semantic Web standards and its use requires a license, which may limit the rate of adoption. Also OPC UA leverages shared information models but does not follow Semantic Web standards.
As a conclusion, automated collaborative design requires shared information models for sharing system designs, component descriptions, and analysis descriptions. Using Semantic Web-compatible ontologies formatted in JSON-LD documents seems to be the best method to implement this. In this paper, we used our own draft ontologies as we could not find existing ones that would fit the needs of the proposed framework and its application. These ontologies, twinschema, ddt, and tors, are published in GitHub [17] and may potentially be utilized in the standardization process of ontologies for collaborative and automated design in the future. Figure 2 illustrates the relationships between the generic formats of describing information, different digital twin description formats, and ontologies for presenting knowledge and shows which of them were used in this paper.
2.3 Collaborative design with digital twins
This section presents a literature review on the use of digital twins in collaborative design. Yue et al. [13] presented a human-centric approach for the collaborative design of centrifugal pumps using digital twins. They identified information siloing, relying on the knowledge of a single person, and a lack of parametrized design as major obstacles in the design process. The paper presented methods to receive feedback from end-users and apply them to the design of the pump. The presented approach aims for a more automated design and uses digital twins’ multi-physics simulation capabilities and ability to describe component relationships to improve the design process. However, the paper does not consider collaboration between the OEMs and system integrators in multi-component system design.
Arista et al. [38] introduced a tradespace framework to support aircraft manufacturing system design relying on ontology-based engineering, model-based system engineering, and semantics. The framework is capable of suggesting possible solutions for the manufacturing system design based on the given requirements, and the use of the framework was demonstrated in designing the manufacturing process for the fuselage orbital joint assembly of an aircraft. Cognitive digital twins were formed during the design process, which were then used to validate the system and could be used in the later lifecycle phases. The integration of ontologies was considered to improve collaboration in the design process. However, the complexity of modeling necessary knowledge in ontology-based engineering was seen as a substantial challenge in the paper.
Guo et al. [11] applied digital twins in the design process of a nuclear plant. Digital twins enabled more collaborative design, as they provided an up-to-date information source for all parties. Furthermore, they could provide a link between different domains, such as engineering design and construction. Digital twins were also utilized in the computer-aided design and virtual testing and validation of the plant. The paper considered the complexity of developing comprehensive digital twins as an issue in their adoption.
Wu et al. [12] presented a framework for collaborative multi-disciplinary design. The framework utilized digital twins for sharing information, multi-disciplinary system modeling, and virtual verification. The framework also focused on preventing the detachment between the design and manufacturing of the product. Digital twins were used to gather feedback throughout the design process. The proposed framework was applied to an automated cutting machine design process. However, the framework does not address how the feedback from the system integrator could be adopted by OEMs.
Zhang et al. [39] integrated design-manufacturing -operations and maintenance utilizing digital twins. They proposed the use of digital twins throughout the design process, from collecting the requirements to the commissioning of the product. Digital twins provide a multi-disciplinary simulation model that can be used to verify design. Furthermore, digital twins were used to distribute information. The paper also introduced a collaborative design platform that offers different simulation and simulation software as well as other tools to support the product design process. The methods presented in the paper were demonstrated in the fault diagnosis and prediction of an Electric Multiple Unit (EMU) bogie. However, the paper does not clearly specify how the collaborative design approach is applied in practice among multiple stakeholders.
In summary, there are only a few papers that consider the use of digital twins in collaborative design, especially in the context of mechanical systems. The previous literature also highlighted several research gaps, such as the need to simulate the system in an early phase before manufacturing, the lack of papers focusing on the product design process, and the verification of the system virtually [12]. The previous literature is also scarce regarding the collaboration between different stakeholders, and to the authors’ best knowledge, there are no papers that consider truly bidirectional collaboration between system integrators and OEMs.
3 Co-Des framework
Co-Des framework allows a collaborative and automated multi-vendor system design process. The targeted systems are complex mechanical systems that consist of several components, as the potential of saving manual work is largest in these systems. The general workflow of the design process enabled by Co-Des framework consists of the following steps: 1) defining initial system designs by a system integrators, 2) finding the suitable components for these designs, 3) creating all possible assemblies from these components, 4) analyzing the formed assemblies, and 5) selecting the best assembly candidates based on analysis results, as presented in Fig. 1.
A prerequisite for the framework and the automated design process is that system designs, components, and analyses are described in a standardized format. For that, this work uses digital twin documents that rely on standardized ontologies for presenting information. Ontologies enable defining the terms explicitly and accurately across all the parties. Furthermore, ontologies can provide a link between different digital twin description formats and allow their interoperability. Digital twin documents must also be made available and discoverable, and, for that, Co-Des framework relies on digital twin web [15]. Digital twin document types and digital twin web are described more extensively in the following subsection.
3.1 Main components
This section presents the fundamental building blocks of the Co-Des framework in more detail. These blocks are three different types of digital twin documents and digital twin web for the distribution of these documents.
3.1.1 Digital design template
is a standard format digital twin document that describes the system design using standardized ontologies, initial versions of which were developed in this paper. This document is defined by a system integrator, and it contains positions for components, requirements for these components, such as dimensions or required power, and analyses that are used to evaluate the system. Creating an optimal DDT that includes all necessary requirements for the system, and its components is a crucial step for the overall success of the design process. Therefore, system integrators should exert substantial effort and utilize expert knowledge in its creation. An example of a windmill DDT is shown in Fig. 3. DDTs are published on digital twin web, which allows their discoverability. Therefore, OEMs can find initial designs to which they can provide their components.
is also a standardized text-formatted DTD that is used to describe a single component and its digital twin (Fig. 4). For example, the description of a component includes the mechanical properties of the component, such as dimensions and maximum allowed torsional vibration amplitude. All relevant information is not necessarily included in the document itself, but the component description may also describe how to retrieve necessary information from other services, such as external databases or PLM (product lifecycle management) systems. Furthermore, the document can also include links to external resources, such as component simulation models that can be used as part of the co-simulation of the system. In the Co-Des framework, component properties saved in component description documents are checked against the requirements given in the DDT.
In order to run analyses automatically, the Co-Des framework requires a standardized description for each analysis service. This analysis description DTD includes the overall description of the analysis, the standards which the analysis follows, such as ISO 22266-1:2022 for torsional vibration of rotating machinery [40], instructions on how to interpret analysis results, and the interface definition of the analysis service. The interface of the analysis service can be for example REST API (Representational State Transfer Application Programming Interface), and the interface definition can follow, for example, OpenAPI standard [41]. An example of analysis description document is shown in Fig. 5.
stores digital twin documents, allows their management, and enables the discovery of them. These documents can contain links to other DTDs forming a network of interconnected digital twins. For example, a large component could consist of multiple subcomponents or there could be a link between component and the digital twin of the factory in which it was produced. Each DTD in DTW is identified by a unique identifier called DTID (Digital Twin IDentifier) that is an IRI (Internationalized Resource Identifier), for example, https://dtid.org/2ef85647-aee2-40c5-bb5a-380c9563ed16.
3.2 System integrator view
Fig. 6
System integrator uses Co-Des framework to find suitable components for the mechanical system, verify their compatibility, and examine the system performance
The Co-Des framework enables bidirectional collaboration between system integrators and OEMs, and both can use it as a part of their design processes. The design process using the Co-Des framework from the system integrator perspective is presented in Fig. 6. With Co-Des, the system integrator is automatically able to analyze a large number of possible components and their joint performance for the initial system design. After the analysis, the system integrator can either fine-tune their design or select the best components and contact the OEMs to make a tender or request modifications to the components. Next, each step of using the Co-Des framework from the system integrator point of view is described in more detail:
1.
Define initial design For using the Co-Des framework, an initial system design has to be defined by a system integrator. The system integrator decides the design based on their expertise and using the necessary software and tools. The design includes component positions, their types, and more specific requirements for the individual components. These requirements can be, for example, dimensions, performance values, or durability metrics. The initial design phase also includes selecting the analyses that are used to evaluate the system performance and properties and parameters for these analyses. Finally, the system design including component requirements and selected analyses is translated into DDT format, which is then given as input to the Assembly finder algorithm.
2.
Find suitable components Using DDT, which includes the needed components, suitable components are searched from digital twin web. DTW contains DTDs published by component manufacturers and works effectively as a product catalog for components. Component properties are compared to the requirements specified in DDT, and components that fulfill these requirements are selected as component candidates for further analysis.
3.
Form assemblies After components candidates are found by crawling through all component descriptions in DTW, all possible assemblies are formed from these components. An assembly is a specific set of components that construct a mechanical system defined in DDT.
4.
Fetch analyses’ details Analyses that are used to evaluate the formed assemblies are defined in DDT. However, details of the analyses are specified in analysis description documents, which are fetched from DTW. Analysis descriptions can contain links to other necessary documents, such as API descriptions, and these documents are also fetched in this step. Analyses’ details and API descriptions are necessary for sending assemblies to be analyzed with the needed parameters in a correct format.
5.
Run analyses The formed assemblies are sent to be analyzed by the selected analysis services with the parameters specified in DDT. Analysis services can be provided by third parties, and these services could be certified by a standardization organization or classification society in the maritime industry, such as DNV. Analysis services may use DTDs of the components to find the necessary information to conduct analysis, such as component physical properties and models. After the analysis is conducted, the analysis service returns results, such as maximum torsional vibrations within the operating speed range given in the parameters.
6.
Return assembly candidates After all possible assemblies are analyzed using selected analysis services, the assembly candidates are returned to the system integrator. The system integrator can then decide how individual analysis results are weighted and select the best assembly for the intended use of the system. Because the selected components have already been analyzed together in the same assembly, the likelihood of incompatible components is diminished and the system-level performance is verified.
The comprehensive presentation of the design workflow with the Co-Des framework, including design process steps and actors, is presented in Fig. 7. This figure summarizes the use of the framework when the design process is initiated by the system integrator.
Fig. 7
Roles of different actors during the automated design process of a mechanical system. AD = Analysis Description
Co-Des framework allows an OEM to find possible customers, verify the performance of their components in a certain mechanical system, and optimize their component performance for the system based on analysis results
From the OEM point of view, the Co-Des framework allows for finding potential customers, verifying the component performance in a certain system, and optimizing the component for the specific use case. For that, system integrators need to publish their initial system designs in DTW. The workflow of using Co-Des framework from the OEM perspective is presented in Fig. 8. The major difference between the system integrator and OEM perspective is that the OEM gives a component description(s) as input to the collaborative design finder algorithm instead of DDT. The algorithm then crawls through DTW to find DDTs, where the component could be fitted, and OEM can then offer their product to the publishers of the DDTs. Each step presented in Fig. 8 is explained further below:
1.
Design component OEM designs a component and creates a Component Description DTD for the component. This CD is given as an input for the Collaborative design finder algorithm. It is also possible to give multiple components, such as a generator and motor, as input at the same time.
2.
Find suitable initial designs The collaborative design finder crawls through DDTs published in DTW to find system designs where the given component(s) could be fitted.
3.
Find other suitable components When an initial system design where the component fits is found, the remaining required components are searched from DTW, similar
4.
Form assemblies All possible assemblies from the given component and other suitable components are formed for analysis.
5.
Fetch analyses’ details This step is similar to step 4 in the system integrator view.
6.
Run analyses This step is similar to step 5 in the system integrator view.
7.
Return analyses results Analyses results are returned to the OEM, who then evaluates if the component performance and properties are at a sufficient level or if modifications are necessary. If component performance is not satisfactory, jump to phase 8, otherwise, jump to phase 9.
8.
Modify component Based on the analysis results, OEM can decide to modify the component, such as its dimensions, materials, or structure. After these modifications, OEM can either jump back to phase 1 and find possible initial designs and other components again or jump back to phase 4 and run the analysis with the same system design and other components. In the first option, OEM wants to optimize the general performance of the component and find new initial designs for the component. (The component can fit new ones after modifications.) In the latter, OEM optimizes the component behavior for the specific design.
9.
Offer component to the system integrator If the component performance is suitable and the requirements for the system behavior given in the DDT are fulfilled, the component is offered to the system integrator. The contact information of system integrator would also be possible to define in DDT, which would allow also automating sending tenders.
3.4 Analysis service provider view
The analysis service provider is an entity that conducts well-defined analyses for the given sets of components forming an assembly. These service providers are responsible for ensuring that the analyses are performed correctly, following the relevant standards, and yield reliable results. Furthermore, these entities provide the necessary infrastructure for running the analyses, such as web interfaces and cloud services. The analysis service providers are expected to be trusted third parties that request a specific price per simulation. A system integrator can also act as an analysis service provider if the tests are highly confidential or if they already have the necessary knowledge to conduct the analyses.
To conduct the analyses, the analysis service provider needs component models, a description of how these components are connected, and the parameters used to perform the analysis. The component models are included in component description documents, either as a text-based description, which can then be used to create the models in the selected simulation software, as in the windmill powertrain design use case presented in Chapter 4, or they can contain a link to the component model located in a separate web storage, from which the actual simulation model can be fetched. For describing the connections between the components, the digital design template of the analyzed system is used. The parameters for the analysis can be unique for the specific analysis, or they can be directly adopted from standards if they specify the exact parameters used, such as maximum applied force to the system. The analysis service providers can also offer various analysis suites that contain multiple pre-defined parameter sets to provide comprehensive information on the system behavior in various cases.
When a specific analysis has been implemented, it is published in DTW using the analysis description document, an excerpt of which is shown in Fig. 5. The document outlines the conducted analysis, the service interface, and possible standards the analysis follows. An analysis service provider can offer several different analyses, all of which are described using a specific analysis description document.
4 Case study of a windmill powertrain design
The use of the Co-Des framework was demonstrated with a real-world windmill design use case, in which the most suitable components for the powertrain considering torsional vibration were selected based on automated assessment of component candidates. This demonstration of automated assessment highlights the benefit of the Co-Des framework compared to the current design approach, where the system integrator would manually collect the required information and models of each component for the torsional vibration analysis. This manual effort of the system integrator is completely mitigated by the proposed Co-Des framework. In Fig. 7, the workflow of the design process and the roles of each actor involved are presented. Next, applying Co-Des framework for the windmill design use case is considered in more detail, starting with the description of actors and thereafter presenting the design process steps in more detail.
4.1 Actors in the automated design process
There are five main actors in Co-Des framework when a system integrator uses it to find suitable components for a system design. These actors are as follows:
System integrator defines the windmill powertrain design and requirements for its properties. The system integrator delivers the windmill to the customer and completes the final assembly using components provided by the OEMs.
Assembly finder algorithm follows the basic workflow presented in Fig. 6: DDT is given to the algorithm that finds suitable components, forms assemblies, sends assemblies to the analysis server, and finally presents the analysis results to the system integrator. The algorithm was implemented with Python and the complete source code can be found on GitHub [17]. Digital twin documents were fetched from Twinbase [16] employing dtweb library [42]. For reading multiple DTDs and sending multiple assemblies for analysis simultaneously, Python built-in threading module was used.
Digital Twin Web was adapted to this paper using an open-source implementation called Twinbase relying on GitHub [16]. Digital twin web was necessary to store and distribute DTDs, which were used to describe system design (DDTs), components (CDs), and analyses (analysis descriptions). Example DTDs stored in Twinbase are listed in Table 1.
Table 1
Examples of digital twin documents employed in the windmill use case
OEMs are manufacturers of components, such as rotors, turbines, and shafts, in this case. OEMs used digital twin web as a platform for distributing descriptions of their components, analogously to a virtual product catalog. Therefore, system integrators do not need to manually request information from OEMs about their products, which streamlines the design process.
Analysis service(s) are provided by third parties to evaluate and verify the system behavior in selected conditions. In this case, only the torsional vibration of the system was considered, and therefore only one analysis service was implemented as a simple web server with the Python Flask framework [43]. The server takes assembly and simulation parameters as input and responds with the maximum torsional vibration for the system. The torsional vibration analysis was implemented with openTorsion, which is an open-source library written in Python for conducting torsional vibration analyses [44, 45]. openTorsion is independent of the implementation of the analysis server and only acts as a calculation back-end for the torsional vibration analysis.
The Flask server functionality is illustrated in Fig. 9. To run the analysis, the input needs to be converted to openTorsion format. For that, the component descriptions need to be fetched from DTW, as the assembly is provided to the analysis service as a list of DTIDs. Thereafter, openTorsion_converter.py script is used to translate these CDs into openTorsion elements. Then, an assembly is created from these components, and an excitation matrix is formed based on the rotor Component Description (Fig. 10). Finally, openTorsion library’s forced_response_analysis() function is used to calculate the maximum torsional vibration amplitude for the windmill powertrain.
Fig. 9
Simple Python Flask server is used to analyze the torsional vibration of assemblies
Torsional excitations caused by the wind turbine are defined in the rotor Component Description document as a list of value pairs containing the frequencies and amplitudes of the torque harmonics
Next, each step in the windmill design process presented in Fig. 7 is introduced in more detail.
4.2.1 Define DTDs
The system integrator selected the initial system design and defined requirements for both the system and component behavior. The design and requirements were then described in DDT. An excerpt of DDT for a windmill is shown in Fig. 3. The DDT of a windmill includes three components, which are a turbine, rotor, and shaft connecting the previous two. The DDT document developed for the use case relies on indexes starting from 0 to define component positions in relation to each other. Indexing-based description of system structure was chosen, as it is a straightforward way to describe simple systems that do not branch, such as windmill powertrains. In addition to the system structure, it is necessary to define the component type needed for a specific position, for which tors ontology was used. The more specific requirements for the component were given as a list that relied on DDT ontology.
The DDT sets the following requirements for the windmill components: (1) Turbine power has to be equal to 500 kW, (2) Shaft length has to be equal to 5 m, and (3) Rotor power has to be larger than or equal to 500 kW and an excitation matrix for it has to be defined. It is also defined in DDT that torsional vibration analysis has to be run for the powertrain using 5000 linearly increasing rotation speeds from 0.1 rpm to 25 rpm and the vibratory torque amplitude must be lower than 50 kNm. In this phase, DTDs for the components and analyses were also defined by OEMs and the analysis service provider, respectively, and published in DTW. Three main ontologies were used for the DTDs: twinschema for digital twins related information, ddt for Digital Design Template specific information, and tors for torsional vibration. In addition, general rdfs ontology by World Wide Web Consortium was used for allowing comments.
4.2.2 Find suitable components
DDT was given as input for the assembly finder algorithm. For each component defined in DDT, the algorithm searched component candidates by crawling through DTW. For that, DTIDs of all DTDs stored in DTW were given as input to the algorithm. When determining if a component was suitable for the design defined in DDT, a component description was fetched from DTW, and then the component type and other properties were compared to the ones defined in DDT (Fig. 11). As can be seen from Fig. 11, ddt and tors ontologies are used to define component properties, ensuring the explicit definition of the terms used. In addition, the component description document contained metadata, such as component name and description, as shown in Fig. 4.
Fig. 11
Assembly finder algorithm selects component candidates by fetching component descriptions from digital twin web and then comparing the component type (1) and other properties (2) specified in CD to the ones defined in DDT. In this case, the component is not accepted as the specified power does not equal the required value in DDT
The assemblies were sent to be analyzed by the torsional vibration analysis service. For that, first, the analysis description document of the service needed to be fetched from DTW. This document contains a description of the torsional vibration analysis and a link to the analysis server interface description that was written using OpenAPI 3.0.3 standard. This standard description allows using Swagger to inspect the API, as can be seen from here1. The interface description defines input for the analysis, that is, a list of DTIDs forming assembly and the rpm values that are used in the torsional vibration analysis. In the case study, a DDT document is not required as input to specify the system structure because it is assumed that position of each component in the list of DTIDs also determines its position in the powertrain. The interface description also defines that analysis returns the maximum amplitude of torsional vibration.
The analysis results are shown in Fig. 12, in which assemblies are sorted from the best to worst. It can be seen that there is a significant difference between the lowest and highest maximum torsional vibration amplitude and that many assemblies exceed the specified design limit. The selected vibration amplitude limit of 50 kNm is only exemplary, and in reality, the metrics and corresponding limits would be selected by domain experts.
Fig. 12
Torsional vibration amplitudes for 1000 different assemblies sorted from lowest to highest. Each index corresponds to an assembly with specific components. Red dashed line shows the limit of torsional vibration amplitude defined in DDT (color figure online)
After the torsional vibration analysis results were received for all assemblies, the results were returned to the system integrator. In this phase, the results could also be compared to the requirements for the system behavior given in the DDT, and only return the assemblies that fulfill requirements. However, this was not implemented in the assembly finder algorithm, as it only returned the three best solutions (solutions with the lowest torsional vibration amplitude) for the system integrator. Based on the results and other factors that are not necessarily documented in the DDT, such as price or preferred OEMs, the system integrator sends requests for quotations for the component manufacturers. Based on these quotations, the system integrator makes the final decision on which components are selected.
5 Performance evaluation
This section presents the performance evaluation of the Co-Des framework. Performance evaluation is essential to assess the proposed Co-Des framework’s initial scalability and validity. Performance was measured by recording the time to find suitable components for a windmill powertrain. The details of this use case were presented in the previous section.
5.1 Measurement setup
The Co-Des framework performance was evaluated by measuring time to find component candidates from DTW and analyze all possible powertrain assemblies from these components for the windmill powertrain presented in Chapter 4 use case. The details of the conducted analysis are presented in Appendix A. The performance tests were conducted using a desktop PC with AMD Ryzen 3900X and 16 GB RAM. The analysis service server and the Assembly finder algorithm were run on the same PC. The measurement scripts are available online in GitHub [17].
The analyzed powertrain consisted of three components: a wind turbine, a grid-connected generator, and a shaft connecting the previous two. For each component, 20 CDs were defined, 10 of which fulfilled the requirements given in the DDT document. Therefore, 1000 \((10\times 10\times 10)\) unique combinations, that is, assemblies, could be formed from the components. One measurement consisted of running the Assembly finder algorithm that selected suitable components, formed assemblies, and analyzed the torsional vibration of the assemblies using an analysis server. The time to complete each of these phases was recorded, and the measurement was conducted 100 times.
5.2 Results
Performance measurements showed that even proof-of-concept implementation of the Co-Des framework provided reasonable execution times (Table 2) compared to the total time of designing a mechanical system. The mean reading time for a DDT was approximately 1 s, and the mean for finding suitable components from 60 components was 3.5 s. Analyzing 1000 assemblies with suitable components took 3730 s on average with a mean absolute deviation of 7.4 s with 100 measurements. The measurements are visualized with violin plots as they did not follow a Gaussian distribution (Fig. 13).
Table 2
Performance measurements of finding suitable components from 60 options by analyzing 1000 assemblies
Value
Read DDT
Find comps.
Analyze assy.
Total
Min
0.906
3.390
3711.3
3715.8
Max
1.203
8.235
3763.1
3767.7
Mean
0.980
3.679
3730.3
3735.1
Median
0.969
3.500
3730.4
3735.0
MAD*
0.030
0.308
7.4
7.4
*MAD = Mean absolute deviation
Fig. 13
Execution times of different phases of the assembly finder algorithm. Violin plots are used to present the non-Gaussian distribution of values, and the blue horizontal lines present the 5th, 50th, and 95th percentiles of the execution times (color figure online)
The foundations of the Co-Des framework lie in the standardized descriptions of system designs, components, and analyses, and their discoverability and open distribution over the Internet. Digital twin documents were found to be a suitable solution for standard descriptions and digital twin web for sharing these descriptions. The use of standardized descriptions requires defining ontologies, and three mock-up ontologies were used in this paper: twinschema for digital twins, ddt for system design, and tors for torsional vibration. These ontologies were also sufficient to define all necessary information for the windmill use case. The proposed ontologies could be utilized in the standardization of automated design ontologies in the future and adopted as a part of existing digital twin descriptions, such as WoT TD, AAS, and DTDL. Furthermore, these ontologies could enable interoperability between the digital twin description formats, allowing seamless translation from one format to another. Thus, the manufacturers could use several formats in parallel to describe their twins.
The Co-Des framework is intended to support the design process of various systems in several domains. Therefore, the building blocks of the framework, DTDs, ontologies, and DTW are all highly flexible and can be applied to various use cases. The windmill use case demonstrated that these general building blocks are easily applicable to a simple real-world design scenario. However, extending the current ontologies, introducing new terms in DTDs, and further standardization work are required for different use cases. This standardization work, nevertheless, does not require any modifications to the presented workflows of the Co-Des framework.
The core feature of DTDs is the ability to provide hyperlinks to other services and resources, allowing the accessibility and discoverability of information from multiple storage systems. This information could include measurements from the products or component models, such as FMUs (functional mock-up units) [1]. Thus, the role of digital twins is to act as an information interface to access necessary information from systems or components.
From a system integrator point of view, Co-Des framework reduces manual work drastically. Instead of searching for suitable components from OEM catalogs, asking for details via emails, building component models based on incomplete information, and finally analyzing component behavior, assemblies that fulfill the given requirements can be found automatically with the assembly finder algorithm. The proof-of-concept implementation of the algorithm showed that the Co-Des framework can easily be adopted in real-world design scenarios.
The goal of the Co-Des framework is not just to allow system integrators to find suitable components but to enable truly bidirectional collaboration with OEMs. OEMs use the framework to find possible customers and optimize their component design for a certain use case. This optimization process could also be automated, so that component parameters are varied with given limits based on the analysis results. This would ensure there are several suitable component candidates available for system analysis. Moreover, if analysis of early phase system design becomes a common practice, as expected by the authors, OEMs have a strong incentive to publish openly available versions of their components. If component models are not published, they have a high risk of exclusion from tender competitions. Implementing proof-of-concept software from the OEM side of the Co-Des framework was not deemed necessary as the implementation would not differ significantly from the assembly finder algorithm.
6.2 Windmill design case study
The proof-of-concept implementation of the Assembly finder algorithm [17] was applied for analyzing a windmill powertrain torsional vibration. This demonstrator highlighted the usability of the Co-Des framework in automating the component selection process for mechanical systems. It showed that the building blocks of the Co-Des framework—DTDs, ontologies, and DTW—could readily be adopted to implement automated assembly evaluation, finding a suitable set of components for a windmill system.
The torsional vibration analysis of the windmill powertrain showed that the maximum torsional vibration depends significantly on the used components and how their mechanical properties fit together, which was an expected result. This highlights the need for conducting an analysis of the whole mechanical system already in the early design phase. Assessing torsional vibrations is only one aspect of designing a powertrain, and to find the best components several analyses would need to be run. Selecting the best powertrain would then require more advanced multivariate optimization algorithms.
The case example also has many other limitations, such as a low number of components, relevant components excluded for simplicity, a very simple mechanical design of components, and unproven criteria for component evaluation. In addition, both the case example and the hardware used for the evaluation of the computing times represent isolated examples. However, the pinnacle of the demonstration was not to design a perfect powertrain for windmills but to analyze the feasibility of the developed design paradigm. For that context, the windmill use case demonstrated that the Co-Des framework is applicable to real-world design problems. The use case also showed that the developed digital twin documents and analysis service were sufficient to enable an automated component selection process with a simple system considering one factor. For more complex examples, the demonstrator versions of ontologies should be extended. Furthermore, the current way of describing system structure in DDT with indexing could require modifications to allow more complex structures to be presented.
A limitation of the case study is that it intentionally excluded the mechanical compatibility of the components, such as matching dimensions, to focus on the dynamic behavior and compatibility of components in complex mechanical systems. Nevertheless, mechanical compatibility can easily be addressed by extending component description and digital design template documents by suitable ontologies. These ontologies should describe unambiguously the mechanical interfaces between components.
6.3 Experiments
To verify the usability of the proposed framework, performance measurements were conducted to find suitable assemblies for the windmill powertrain. Reading a DDT that described the design of the powertrain took approximately one second, which was in line with measurements presented in [16]. The mean of searching suitable components from 60 options was approximately 3.7 seconds, which shows the power of concurrency: multiple requests could be sent simultaneously, and reading 180 (three times 60 component descriptions) CD documents was only a few times slower than reading a single DDT document. However, more optimized algorithms than looping through all components component options are necessary when the number of CDs in digital twin web increases. Developing such algorithms are part of future work, and are not considered relevant in this paper. Analyzing 1000 assemblies took approximately 1 h and 2 min on average, which can be considered reasonable, taking into account that the analysis was run on the Flask development server without any focus on improving the performance of computing.
6.4 Scientific and practical impact
There are also other initiatives that aim to facilitate openness in design processes with selected tools and technical means on how to run co-simulations. One method to build multi-component simulations is to rely on Modelica Libraries [46] that include both free and commercial packages of simulation models. However, these libraries do not seem to include several versions of similar components from several vendors. Another approach for multi-component simulation from the maritime industry is Open Simulation Platform [3], in which the selected design process is the design of a ship and the selected tool for simulation is FMI [1]. For connecting FMI models, an initiative called System Structure and Parameterization (SSP) was launched. SSP allows describing complete systems and connections between components in standardized format [2]. However, the Co-Des framework brings an additional layer on top of these initiatives with openly available descriptions of designs, components, and analysis services in the format of digital twin documents, and making these documents publicly available with digital twin web. Co-Des connects the system integrators and OEMs by enabling an automated discovery of components and automated analysis of assemblies, benefitting both.
The main contributions of this paper are enabling the truly bidirectional collaboration between OEMs and system integrators and automated component analysis, which is an integral part of the collaborative design process. As all parties can change their designs according to the results of automated analysis, each component of the system is optimized instead of fine-tuning only subsections of the system. The development of the Co-Des framework is supported by significant actors in the shipbuilding industry, and the practical impact of the work is realized by the industry-university collaboration project. Furthermore, the proof-of-concept open-source implementation Co-Des framework [17] serves as an initial step toward the wide adoption of the principles of the framework.
The adoption of the Co-Des framework has the potential to transform the current design processes, leading to a paradigm shift in the mechanical system design. The current practice of manually transferring models and component information is minimized when adopting the open sharing of models. Furthermore, the framework allows both system integrators and OEMs to propose design changes, fostering a collaborative design process and bidirectional communication. Finally, the automated assessment of system designs can enable finding suitable components faster as well as quicker iteration cycles, as modifications to the designs can be promptly tested. These changes in the design process provide a foundation for more refined designs with reduced operational costs and emissions, fewer compatibility issues, and improved reliability. Furthermore, the carbon footprint and production expenses could be diminished as intentional oversizing of components is no longer required due to more accurate system analysis in the earlier design phases.
6.5 Limitations and future work
Writing digital twin documents manually is a laborious task that must be done meticulously and without errors. Consequently, developing automated tools to help in the creation of DTDs would be beneficial and is considered a part of future work. These tools could provide a graphical user interface that employs a drag-and-drop approach to define the system structure. This paper did not cover the development of automated tools, as they are not a core part of the Co-Des framework, and they are seen as straightforward to implement with limited scientific novelty.
Standardization of the digital twin documents format is also a crucial part of future work. Without proper standardization, the automated search of components and assembly analysis is not possible. There are several existing digital twins standards, such as AAS, WoT TD, DTDL, and NGSI-LD, whose suitability for presenting the component, system, and analysis information should be assessed. Furthermore, developing suitable ontologies for describing the necessary details for automated component selection and assembly analysis, including mechanical properties, component dependencies, system and component-level requirements, and mechanical connections, is part of the future work.
In addition to the digital twin documents, the standardization of component model formats is necessary. One option for standardization is FMI, which is an established standard in co-simulation. Another aspect related to the models is to ensure that there are enough different models available for various systems. The authors acknowledge the issue with model availability at the beginning of adopting the proposed approach. Nevertheless, it is a common issue that enough users need to be acquired in order to reach the full potential of the platform-based approaches and achieve the network effects. Therefore, when adopting the Co-Des framework, it is crucial to attract the key companies of the domain in the early phase. To facilitate this, easy tools to publish component and system models and protect their immaterial property rights (IPR) are needed. These tools are considered an integral part of future work.
For high technology companies, IPR are of utter importance, and in most cases, the open share of component or system models with their parameters and details is not possible. In this case, the use of the Co-Des framework is possible with a few different approaches: (1) IPR-protected component details and models are only shared with analysis services but not with system integrators or other OEMs. This requires that analysis services are only provided by trusted third parties. (2) The whole framework, including DTDs and component models, is hosted by a trusted third party that offers to find suitable components or new customers as a service. (3) Component models are translated into black box models: these models can be shared freely as the model preserves functionality without revealing IPR. Tevajärvi [47] presented a comprehensive analysis of the methods to protect the IPR of component models in the context of the collaborative simulation. However, the use of black box models with tightly coupled systems requires further examination. Applying IPR protection with black box models to the Co-Des framework is part of future research.
In the Co-Des framework, analysis services have an important role in ensuring analyses are run according to the standards and using the best available methods. In addition to using analysis services in the early design phase, these services could also be used to verify design virtually as a part of the acceptance tests. For that, analysis service providers could be accredited by a classification society in the future. This would revolutionize the ship design verification process and reduce the costs of the verification. Furthermore, there is potential to reduce the need for sea tests in the future if the most accurate models of ship powertrains can be used in the verification process.
The Co-Des framework was developed for the automated design of complex mechanical systems that require simulating the system as a whole to verify its performance, such as ship powertrains. Nevertheless, the use of the framework is not limited to any specific domain, and it can be applied to various mechanical systems. Demonstrating and validating the use of the Co-Des framework in various domains and analyses is part of future work.
The performance measurements showed that 1000 assembly candidates could be analyzed in approximately one hour using the proof-of-concept implementation of the Co-Des framework. This time can be considered reasonable compared to the total time taken by the design process or the time to conduct traditional analyses, such as finite element method analysis. Furthermore, the time to search for suitable components is in order of magnitude that it can be performed overnight. However, a more complete analysis of the scalability of the system is a part of the future research. If the number of components in the powertrain is increased, the number of possible assemblies increases rapidly. Furthermore, if there is a large number of options for each component, the simulation time quickly grows. To limit the number of simulations conducted, curated lists of the best available components from reputable OEMs can be provided. These lists could be created by trusted third parties that utilize the domain knowledge in the selection process. In addition, better filtering methods for component candidates, which take into account the preferences of the system integrator, the reputation of the OEMs, and previous simulation results, should be developed to reduce the number of simulations. Future work to reduce computation time also includes developing efficient search algorithms and indexing methods for component digital twins, as well as optimizing the simulation algorithms and methods. Furthermore, ways of incorporating expert knowledge in the component search process should be addressed in future research.
7 Conclusion
This paper presented Co-Des framework to address issues in the current mechanical system design process: (1) Selecting components is laborious and includes a significant amount of manual work, (2) Component descriptions are not in a standardized format, which makes it difficult to find relevant information, and (3) System performance is not analyzed until the component selection is finalized, which leads to suboptimal performance. Co-Des framework automates the early phase of complex mechanical system design by finding suitable assembly candidates for the selected mechanical system design. Various optimization methods can then be used to select the best assembly from these candidates. Co-Des framework also enables bidirectional collaboration between OEMs and system integrators, which allows further optimization of the components for the selected use case. All this leads to a paradigm shift in mechanical system design toward automated and collaborative design, and as an end result, mechanical systems that perform better and are more durable.
Future work includes research on sharing components and designs in a format that preserves critical IPR but still allows system analysis. Another approach would be to rely on trusted third parties to find and analyze suitable assemblies. The other part of the future work side is the standardization of digital twin document format and ontologies that are required for describing components and system designs.
Acknowledgements
This research was funded by Business Finland under Grant 3508/31/2019 (ITEA 3 Call 5 MACHINAIDE), 7464/31/2022 “Twinbase,” 3014/31/2021 “GOOD-Future electrified mobile machinery for harsh conditions,” and 243/31/2022 “Co-Des-Digital transformation of collaborative powertrain design.” The work reported in this paper has been funded in part by European Union’s Horizon 2020 under IoT-NGIN project Grant No 957246, and in part by the Academy of Finland (Centre of Excellence in High-Speed Electromechanical Energy Conversion Systems) under grant number 346443. The authors would like to thank Joel Mattila and Albert Dömötör for their help in testing the software.
Declarations
Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Riku Ala-Laurinaho
is a postdoctoral researcher at Aalto University, earning his BSc in 2018, MSc in 2019, and DSc in 2021—all from Aalto University. With a passion for driving digital transformation in industrial settings, he currently leads the Aalto DigiTwin Lab. The laboratory focuses on networked digital twins, industrial information integration, and metaverse through industrial collaboration. He also serves as the project manager of the Co-Des (digital transformation of collaborative powertrain design) project, which aims to enable automated ship powertrain design using digital twins.
Juuso Autiosalo
received the BSc, MSc, and DSc degrees from Aalto University, Espoo, Finland, in 2015, 2017, and 2021, respectively. He was a Visiting Research Scholar at Michigan State University in 2020. His research aims to enable the creation of a global network of digital twins. Currently, he is preparing to commercialize the results of his doctoral dissertation as the project manager of a research-to-business project. From 2018 to 2019, he was the project manager for the DigiTwin project, which laid the foundation for Aalto’s digital twin research in collaboration with industry partners.
Sampo Laine
received his MSc degree in mechanical engineering from Aalto University in 2021. He is currently working as a Doctoral Researcher at Aalto University. His research interests include rotor dynamics, vibration control, torsional vibrations, electric powertrains and simulation software.
Urho Hakonen
received his BSc and MSc degrees in mechanical engineering from Aalto University in 2021 and 2023, where he is currently pursuing a Doctoral degree. His research interests include structural dynamics and virtual sensors.
Raine Viitala
was born in 1992. He received his MSc and DSc degrees in mechanical engineering from Aalto University in 2017 and 2018, and was appointed Assistant Professor in 2020. His doctoral dissertation was acknowledged with the Aalto University School of Engineering dissertation award. He has a solid background in experimental large rotor research, including vibration analysis, bearing excitations, roundness measurements and manufacturing for operating conditions.
Torsional vibration analysis of a windmill powertrain
The collaborative design use case focuses on analyzing the torsional vibrations of a wind turbine generator system presented in [48]. Torsional vibration is an important factor to be considered when designing mechanical systems because the stress caused by the vibrations poses design limitations and affects the lifecycle of the system. The distinct feature of torsional vibration analysis is that it cannot be conducted separately for individual components, but a model of the entire torque transmitting system is required.
Torsional vibration is vibrating motion due to the twisting of a system about its axis of rotation. Torsional vibration causes stresses, which has to be considered when designing rotating machinery. The shaft-line FEM (finite element method) is commonly used in the torsional analysis of rotating mechanical systems [49]. The mechanical system is modeled as lumped elements with equivalent stiffness, damping, and inertia. The windmill powertrain used in the experiment is a three-degree of freedom system; thus, the equations are presented accordingly. The equations of motion of the system are formed as
where \(\varvec{\theta }\) contains the rotations of the ends of each shaft element and the vector \({\textbf {T}}\) is the harmonic torque applied to the system. The mass matrix \({\textbf {M}}\) containing the inertia values and the stiffness matrix \({\textbf {K}}\) containing the stiffness values are
where \(\omega _i\) and \(\varphi _i\) are the eigenfrequency and eigenmode of the ith mode, respectively. The eigenmodes form the modal matrix \(\varvec{\Phi } = [\varphi _1 \varphi _2 \varphi _3]\). The damping matrix \({\textbf {C}}\) used in torsional forced response analysis is calculated as follows:
In the present case experiment, the excitation frequency depends on rotational speed. The response of shaft element j can be evaluated by summing up the vibratory torque calculated at one rotational speed. Repeating this for different speeds yields the vibratory torque of element j over a range of operating speeds.
Havard, V., Jeanne, B., Lacomblez, M., Baudry, D.: Digital twin and virtual reality: a co-simulation environment for design and assessment of industrial workstations. Product. Manuf. Res. 7(1), 472–489 (2019). https://doi.org/10.1080/21693277.2019.1660283CrossRef
9.
Schluse, M., Rossmann, J.: From simulation to experimentable digital twins: Simulation-based development and operation of complex technical systems. In: From simulation to experimentable digital twins: Simulation-based development and operation of complex technical systems, 2016 IEEE International Symposium on Systems Engineering (ISSE), 1–6 IEEE, (2016). https://doi.org/10.1109/SysEng.2016.7753162
10.
Autiosalo, J., et al.: Towards integrated digital twins for industrial products: case study on an overhead crane. Appl. Sci. 11(2), 683 (2021). https://doi.org/10.3390/app11020683CrossRef
11.
Guo, C., Xu, X., Hu, J., Zhang, J., Zhou, Z.: (eds.) Intelligent design and construction coordination of nuclear power based on digital twin. In: Proceedings of the 3rd International Conference on Green Energy, Environment and Sustainable Development (GEESD2022), aaa, 144–153 IOS Press, (2022). https://doi.org/10.3233/ATDE220276
12.
Wu, Y., Zhou, L., Zheng, P., Sun, Y., Zhang, K.: A digital twin-based multidisciplinary collaborative design approach for complex engineering product development. Adv. Eng. Inform. 52, 101635 (2022). https://doi.org/10.1016/j.aei.2022.101635CrossRef
Ala-Laurinaho, R., Autiosalo, J., Nikander, A., Mattila, J., Tammi, K.: Data link for the creation of digital twins. IEEE Access 8, 228675–228684 (2020). https://doi.org/10.1109/ACCESS.2020.3045856CrossRef
15.
Autiosalo, J.: Discovering the Digital Twin Web–From singular applications to a scalable network. Dissertation, Aalto University (2021). http://urn.fi/URN:ISBN:978-952-64-0621-3
Liu, M., Fang, S., Dong, H., Xu, C.: Review of digital twin about concepts, technologies, and industrial applications. J. Manuf. Syst. 58, 346–361 (2021). https://doi.org/10.1016/j.jmsy.2020.06.017CrossRef
Mattila, J., Ala-Laurinaho, R., Autiosalo, J., Salminen, P., Tammi, K.: Using digital twin documents to control a smart factory: simulation approach with ROS, Gazebo, and Twinbase. Machines 10(4), 225 (2022). https://doi.org/10.3390/machines10040225CrossRef
Ricci, A., Croatti, A., Mariani, S., Montagna, S., Picone, M.: Web of digital twins. ACM Trans. Internet Technol. (2022). https://doi.org/10.1145/3507909CrossRef
Arista, R., Zheng, X., Lu, J., Mas, F.: An ontology-based engineering system to support aircraft manufacturing system design. J. Manuf. Syst. 68, 270–288 (2023). https://doi.org/10.1016/j.jmsy.2023.02.012CrossRef
39.
Zhang, C., Dong, L., Wang, Y.: Design-manufacturing-operation & maintenance (o &m) integration of complex product based on digital twin. Appl. Sci. (2023). https://doi.org/10.3390/app13021052CrossRef
40.
International Organization for Standardization. ISO 22266-1:2022 Mechanical vibration—torsional vibration of rotating machinery—part 1: evaluation of steam and gas turbine generator sets due to electrical excitation. Standard, International Organization for Standardization (2022). https://www.iso.org/standard/78144.html
