HomeLong living human-machine systems in construction and production enabled by digital twins

ArticleOpen Access

Long living human-machine systems in construction and production enabled by digital twins

Exploring applications, challenges, and pathways to sustainability

Birgit Vogel-Heuser
Univ.-Prof. Dr.-Ing. Birgit Vogel-Heuser received a Diploma degree in Electrical Engineering and a Ph.D. in Mechanical Engineering from RWTH Aachen. Since 2009, she has been a full professor and director of the Insititute of Automation and Information Systems at the Technical University of Munich (TUM). Her current research focuses on systems and software engineering. She is a member of the acatech (German National Academy of Science and Engineering), editor of IEEE T-ASE, and IEEE Fellow and member of the science board of MIRMI at TUM.
,Fandi Hartl,Moritz Wittemer,Jingyun Zhao,Andreas Mayr,Martin Fleischer,Theresa Prinz,Anne Fischer,Jakob Trauer,Philipp Schroeder,Ann-Kathrin Goldbach,Florian Rothmeyer,Markus Zimmermann,Kai-Uwe Bletzinger,Johannes Fottner,Rüdiger Daub,Klaus Bengler,André Borrmann,Michael F. Zaeh andKatrin Wudy
Katrin Wudy, Prof. Dr. -Ing., holds the Professorship of Laser-based Additive Manufacturing at the Technical University of Munich. Her focus is on laser- and powder based additive manufacturing technologies. She bridges the disciplines of materials science, manufacturing and photonics. Prof. Wudy finished her PhD at the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) in the field of powder and beam-based additive manufacturing in 2017, after which she headed the “Additive Manufacturing” working group at the Institute of Polymer Technology until 2019 and was Managing Director of the CRC 814 “Additive Manufacturing” at the FAU. She was appointed tenure-track assistant professor in 2019, and promoted to associate professor in 2024.

Published/Copyright:September 10, 2024

Download Article (PDF)

Published by

Become an author with De Gruyter Brill

Submit Manuscript Author Information Explore this Subject

From the journalat - Automatisierungstechnik Volume 72 Issue 9

Abstract

In the industrial sector, products evolve significantly over their operational life. A key challenge has been maintaining precise, relevant engineering data. This paper explores the digital twin concept, merging engineering and operational data to enhance product information updates. It examines digital twin applications in construction, material flow, manufacturing and production, citing battery production and additive manufacturing. Digital twins aid in analyzing, experimenting with, and refining a system’s design and its operation, offering insights across product and system lifecycles. This includes tackling data management and model-data consistency challenges, as well as the recognition of synergies. This paper emphasizes sustainable, efficient management of engineering information, reflecting shifts in product longevity and documentation in industrial products and machinery.

Zusammenfassung

In der Industrie entwickeln sich Produkte während ihrer Lebensdauer erheblich weiter. Eine zentrale Herausforderung ist die Pflege präziser, relevanter technischer Daten. Dieser Beitrag beleuchtet das Konzept des digitalen Zwillings, welches Entwicklungs- und Laufzeitdaten integriert, um die Aktualität von Produktinformationen zu verbessern. Insbesondere werden Anwendungen im Bauwesen, dem Materialfluss, der Fertigung und Produktion, am Beispiel der Batterieproduktion und der additiven Fertigung, untersucht. Digitale Zwillinge unterstützen das Analysieren, Experimentieren mit und Verbessern von Systementwürfen und deren Nutzung und gewähren Einblicke in Produkt- und Systemlebenszyklen. Sie adressieren Herausforderungen im Datenmanagement und der Konsistenz zwischen Modellen und Daten sowie das Erkennen von Synergien. Dieser Beitrag stellt das nachhaltige, effiziente Management von Entwicklungsinformationen vor, das den Wandel in Langlebigkeit und Dokumentation von industriellen Produkten und Maschinen widerspiegelt.

Keywords:digital twin;production;construction;production and construction engineering;human machine interaction;Industry 4.0

Schlagwörter:Digitaler Zwilling;Produktion;Bauwesen;Produktions- und Bauingenieurwesen;Mensch-Maschine-Interaktion;Industrie 4.0

1 Introduction and important definitions

The longevity of products, particularly those in the industrial sector, often involves substantial evolution over their operational lifespan. A study by ZVEI indicates that machinery and plants can be in operation for up to 50 years, and construction equipment frequently has a similarly extended usage period [1], [2], [3]. Throughout this duration, these machines often undergo optimization, retrofitting, upgrading, and adaptation. Stationary machines may even be relocated based on product demands or market changes. In the context of mobile machinery, there is a tendency to sell these to developing countries [4]. Therefore, it is essential that these machines are designed for enduring performance, symbolizing a product that withstands the test of time. Historically, a major challenge in this field has been the preservation and relevance of engineering information. Often, as-built documentation became quickly outdated and synchronizing it with ongoing changes, especially during retrofitting, was almost unfeasible. This necessitated manual analysis of equipment before proposing any updates. A case in point is the lab demonstrator xPPU, where maintaining updated documentation post-retrofit (PPU to xPPU) proved challenging [5]. Furthermore, reliance on PDF files or paper-based documentation previously hindered the breadth and timeliness of updates. However, recent developments are altering this landscape, potentially enabling the maintenance of up-to-date documentation. These developments include: the increased availability of automation tools, facilitating easier data exchange between different engineering platforms, and enhancing the potential for later updates; advances in PDF file analysis, contributing to more effective document management [6] and lastly, the emergence of the Digital Twin (DT) paradigm, which encompasses not only the engineering information mentioned above but also the operational data. This integration significantly enhances the ability to update product information models. Therefore, this paper focuses on the evolving landscape of product longevity and documentation in the context of industrial machinery, underscoring the shift towards more sustainable and efficient management of engineering information.

This paper explores the role of DTs in engineering from a hierarchical perspective, spanning from the built environment and construction logistics, over intralogistics, to machine and plant engineering, technical product development, production, manufacturing processes, and the human-machine interface as shown inFigure 1. The aim is to provide a common definition and understanding of DT applications, the state of research and engineering, and challenges, particularly in systems integration.

1.1 What is a digital twin?

To establish a comprehensive understanding of the term “digital twin” in engineering, it was essential to derive a cross-domain applicable definition that encompasses the various perspectives and applications within the field. This process involved gathering and analyzing a collection of over 50 definitions^[1] from eight domains: built environment, construction logistics, material flow systems,^[2] machine and plant engineering, technical product development, production, manufacturing process and human in the workplace. The analysis focused on key questions such as the nature of a DT “What is a digital twin?”, its constituent components “What is it made of?”, and its overarching objectives “What is its goal?” (cp.Figure 2). While the wording in the analyzed definitions of DT varied, they all agreed upon the core elements of the definition given in the following.

Figure 1:

Digital twins in the domain of construction and production engineering with human in the loop.

This definition is used as a robust foundation for understanding and discussing DTs in the context of this paper:

Digital Twinsare defined as dynamic digital representations of specific real-world entities consisting of (interlinked) components and interfaces with application-specific attributes.Scales of DT-aspects to be considered are for example time, size, accuracy, hierarchy, life cycle phase. Digital twins have the goal of recurrent improvement in the real world.

The dominant question “What is a DT?” was answered by aggregating definitions to the resulting definition “Digital Twins in Engineering are defined as dynamic digital representations of specific real-world entities”. This answer was derived by analyzing cross-domain definitions (cp.Figure 2, left). The same approach was used to answer the other two questions. “What is a DT made of?” was answered by “A DT consists of (interlinked) components and interfaces with application-specific attributes and their scales like time, size, accuracy, hierarchy and life cycle phase”, where the scale is an implicit characteristic adjusted to applications (cp.Figure 2, center). The question “What is its goal?” was answered by “DTs have the goal of recurring improvement of the real world” (cp.Figure 2, right). In the standard ISO 23247-1 [7], [8], [9], [10], the comparison focuses only on manufacturing use cases and defines so-called Observable Manufacturing Elements with information attributes that are much more limited than the characteristics listed inTables 1 and2.

Table 1:

Detailed overview of DT types over the considered domains.

Engineering domain		Construction			Production				Humans
		Built environment [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38]	Construction industry [25], [39], [40], [41], [42], [43]	Material flow systems [44], [45], [46], [47], [48]	Machine and plant manufacturing [46], [47], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58]	Product development process (PDP) [11], [17], [22], [23], [24], [81], [82], [83]	Manufacturing process (add. manufacturing) [59], [60], [61], [62], [63], [64], [65], [66], [67], [68]	Production process (battery production) [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79], [80]	Humans [84], [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [95], [96], [97]
Real world entities		– existing buildings – construction sites – environmental impacts – sensors/monitoring devices	– products (geometry) – processes by different resources, like workers, vehicles, materials	– vehicles, like load carriers – stationary equipment, like conveyors, gates	– entire machines and plants – sensors and actuators – control units, like PLCs	– physical product – realized processes	– additive manufacturing process	– entire battery production plant – individual production equipment, like machines, belts	– human in the workplace
Interlinked components		– models, like multilayer CAD/BIM, FEM – data handling, like analysis, storage, transfer	– models, like geometry, environment, schedule, cost plans – data handling, like analysis, storage, transfer – decision units, like routing, scheduling, maintenance	– environmental model – decision units, like routing, scheduling, disposition	– data structure model – machine-readable data base like AAS or Ontology for an asset – visualization and Simulation (like Nvidia Omniverse)	– models – simulations – database – visualization, like interface to the user	– models, like physics-based process model, statistical AI-based model – decision units, like process parameter selection – visualization, like of sensor data	– models (physics-based models, AI-based models) – databases, like, data sheets, machine features, product data – visualizations	– dedicated human models, like cognitive models, physiological models
Scope of the DT and application		– from a predictive model for simulation of building’s structural behavior – to complete information model for the entity as the basis for decision support, learning to benefit from engineering data	– from a predictive model for simulation of the entities in construction industry – to a complete information model for the entity to get a true as-built model in every detail (building information, process information, equipment information	– from visualization of load units’ and vehicles’ position – over a complete information model for the entity as the basis for decision support, learning and benefit from engineering data – to control the material flow for efficiency optimization	– from a pure CAD or simulation model – to a complete information model for the entity as the basis for decision support, learning and benefit from engineering data	– from pure CAD – to a complete information model for the entity as the basis for decision support, learning and benefit from engineering data	– from visualization of sensor data for supporting machine operators – over a complete information model for the entity as the basis for decision support, learning and benefit from engineering data – to real-time process control for fist-time-right manufacturing	– from visualization of sensor data for supporting machine operators – to a complete information model for the entity for a highly interlinked process chain as the basis for decision support, learning and benefit from engineering data	– from the assessment of individual workplaces for ergonomic optimizations – to macro ergonomic information to maximize productivity
Input		– as designed data, like constructive design – as-built data, like resources providing, information about progress – monitoring data from operation of entity like energy consumption, damage, deformations – environmental influences – authoritative requirements	– as designed data, like geometry, environment, costs, schedule – as-built data, like resources providing, information about progress – monitoring data from operation of entity like energy consumption, damage, deformations – historical data, like probabilistic distribution	– as designed data, like material flow layout – monitoring data from operation of the entity like throughput – historical data, like maintenance requirements	– as designed data, like geometry, environmental constraints, costs, schedule – as-built data, like entities revised CAD models – monitoring data from operation of the entity, like energy consumption, damage, deformations,in-situ sensor data – environmental influences	– as designed data, like sustainability and cost factors – as-built data, like product’s shape, materials, functionality, utilization – monitoring data from operation of the entity, like real-time operational performance data	– as designed data, like physical/statistical model data – monitoring data from operation of the entity, like real-timein-situ sensor data (temperature, topography – environmental influences	– as designed data, like simulation models, structure of electrode coating – as-built data, like offline measurements – monitoring data from operation of the entity, like machine and raw material data – historical data, like previous process data, statistical models	– monitoring data from operation of the entity, like working conditions: physical loads, environment, organizational structures – as designed data, like human properties: physiological, demographics, knowledge
Output		– basis for decision-making, like load transfer in the structures – monitoring data, like progress for as-built models as basis for smart control loops	– basis for decision-making, like delivery schedule – monitoring data, like progress for as-built models	– basis for decision-making, like driving orders for specific vehicles	– basis for decision-making like maintenance, recovery, restart after fault – monitoring data, like OEE – system parameters, like line speed – predictive maintenance recommendations	– basis for decision-making, like manufacturability – product parameters, like design, control – product parameter, like system models	– basis for decision-making, like path plan – monitoring data, like temperature as basis for smart control loops – process parameters, like, input power, process speed	– basis for decision-making, like material flow – monitoring data, like product quality – predictive Maintenance recommendations	– monitoring data, like physical, cognitive emotional strain, health – requirements, like time, space, reliability, perception, cognition, decision making, action, training, organizational circumstances
Scales	Time	Seconds to decades	Seconds to weeks	Seconds to one work shift to one day	Micro-seconds (hard real-time) to decades	Micro-seconds (hard real-time to decades, like behavior to EOL	Micro-seconds (hard real-time) to days, like, build job supervision	Milli-seconds (real-time/near-real-time) to hours, days, like process or material flow simulation design	Seconds to months
	Size	cm, like small equipment to m, like entire buildings	cm, like small equipment, to m, like entire buildings	cm, like positioning unit, to km, like warehouse	cm, like small equipment to m, like entire factory	Single entities to entire systems	µm, like process zone monitoring, to cm, like part geometry	µm, like particle scale, to m, like factory	mm, like sensory perception, to m, like reaching height
	Accuracy	Sensors, actuators, models with their tolerance, behavioral uncertainty	Sensors, actuators, models with their tolerance, behavioral uncertainty	Vehicle positioning from several meters to approx. 10 mm	Sensors, actuators, models with their tolerance, behavioral uncertainty	Sensors, actuators, models with their tolerance, behavioral uncertainty	Sensors, actuators, models with their tolerance, behavioral uncertainty	Sensors, actuators, models with their tolerance, behavioral uncertainty	Sufficient accuracy diversity of humans
	Hierarchy	Objective dependent	Objective dependent	From ERP system to control system to vehicles	ISA88: all levels	From physical product to entire processes	From entire part over layers to individual tracks	From plant to process	From micro to macro ergonomic
	Life cycle phase	Design, planning, implementation, operation, disassembly	Design, planning, implementation, operation, disassembly	Planning, operation	Design, implementation, start-up, operation, redesign, disassembly	Design, redesign	Planning (of the process), operation (process monitoring/control)	Design, implementation, start-up, operation	Resign (for humans), operation, redesign

Table 2:

Overview of general digital twin characteristics over the considered domains.

Engineering	Construction			Production				Humans
domain
	Built environment	Construction industry	Material flow systems	Machine and plant manufacturing	Product development process (PDP)	Manufacturing process (add. manufacturing)	Production process (battery production)	Humans
Aim	– improve sustainability, like resource-efficient conversion – increase of structure lifetime, like predictive design and analysis – cost optimization – reduction of safety concerns trough structural health monitoring	– design control – cost optimization by efficient production supply and smart traffic control	– improve sustainability, like transport efficiency – cost optimization by efficient production supply and smart traffic control	– provide accurate entity visualization – entity control for smart process adaptions during runtime – increase plant lifetime like predictive maintenance – cost optimization by increase of OEE	– improve sustainability – cost optimization by efficient design and manufacturability	– provide accurate entity visualization – improve and accelerate process design by overcoming trial-and-error approach – in-line control for quality improvement, like part and waste reduction – increase profound entity understanding, like laser-material- interaction – cost optimization by first-time-right manufacturing	– provide accurate entity visualization – improve and accelerate process design by overcoming trial-and-error approach – in-line control for quality improvement, like electrode, cell and waste reduction – increase profound entity understanding, like material-process-quality-property relationships – guide decision-making – cost optimization by more efficient production	– reduction of health and safety concerns trough human appropriate design of workplaces – cost optimization by assessment and optimization of work processes
Challenges	– interface design (efficient and flexible communication between low-level and high-level systems) – uncertainties from construction process and environmental impacts Domain specific challenges: – construction specific requirements, like uniqueness, relatively low standardized processes – DT is more than BIM (geometry, costs, schedule), it is also about the workflow and the entities fulfilling the tasks	– interface design (efficient and flexible communication between low-level and high-level systems) Domain specific challenges: – still relatively low degree of digitization – construction specific requirements, like uniqueness, relatively low standardized processes – DT is more than BIM (geometry, costs, schedule), it is also about the workflow and the entities fulfilling the tasks	– interface Design (efficient and flexible communication between low-level and high-level systems) Domain specific challenges: – incorporation of vehicles with different capabilities regarding transport capacity, load handling, sensing, or communicating	– interface design (efficient and flexible communication low-level high-level systems) – heterogeneous data (exchange) formats, inconsistent/incomplete information sources, lack of meta data, different tools for data exchange (formats, protocols, automation) – uncertainties from design/manufacturing process and environmental impacts Domain specific challenges: – leveraging small data and rare events to improve OEE – evolution of DT to meet changing entity needs – lack of machine and plant DTs, requiring component suppliers to integrate them into various tools.	– heterogeneous data (exchange) formats, inconsistent/incomplete information sources, lack of meta data, different tools for data exchange (formats, protocols, automation) Domain specific challenges: – Non-technical issues, e.g., shortage of specialists	– interface design (efficient and flexible communication between low-level and high-level systems) – heterogeneous data (exchange) formats, inconsistent/incomplete information sources, lack of meta data, different tools for data exchange (formats, protocols, automation) – uncertainties from insufficient process understanding Domain specific challenges: – high computation times for particle scale simulations – application specific temporal and spatial data resolution requirements – diversity of AD processes	– interface design (efficient and flexible communication, like merging the individual processes and a consistent virtual electrode as a product – heterogeneous data (exchange) formats, inconsistent/incomplete information sources, lack of meta data, different tools for data exchange (formats, protocols, automation) Domain specific challenges: – processing of a variety of raw materials with varying property profiles along one process chain – merging the individual processes and a consistent virtual electrode as a product – High computation times for particle scale simulations	– interface design (efficient and flexible communication models of different human aspects) – heterogeneous data (exchange) formats, inconsistent/incomplete information sources, lack of meta data, different tools for data exchange, like gaps between hierarchical levels may cause incompatibility among Human Digitals Twins Domain specific challenges: – data privacy of the worker
Outlook	– use of data fusion: from isolated technical solutions to a “system of systems” – potential to achieve new levels of resource efficiency – focus on improved tool integration (monitoring and simulation)	– use of data fusion: from isolated technical solutions to a “system of systems” – increased interfaces standardization of technical solutions for better translation of human experiences into machine readable tasks, like by using standardized modularized process descriptions via ontologies	– increased interfaces standardization of control systems and vehicles for better interoperability and expandability – recording of long-term data from the transport system offers possibility of process improvements via big data analysis – strictly centralized control structures can be replaced by variable hierarchies for resilience when errors occur – vehicle functions that need to be incorporated in the control structure	– focus on real-time aspects to ensure production safety and stability – improve backflow of information (virtual to real-world asset)	– prediction models for improved product performance	– increased interfaces standardization of machines for increase accessibility of machine and sensor data – profound process understanding for implementation of real time control	– increased interfaces standardization of simulation models for process twins’ linkage – continuous development of DT modeling – enabling a tracking and tracing system – enabling a virtual instance of the electrode along the process chain – integration of different inline sensors for individual process steps for comprehensive quality monitoring	– human DTs enable individually adaptable work environments – human centered workplaces enhance productivity and satisfaction – risk for injuries and human errors are reduced

1.2 Digital twin in engineering

The engineering domain encompasses a vast range of industries, including construction, material flow systems, manufacturing, humans, and more. The advent of DTs has had a transformative impact on how engineers approach system design, operation, and optimization [11], [12], [13]. In the manufacturing sector, DTs are used to replicate and optimize production processes, machinery, products, and resources. By creating virtual representations of production lines and equipment, engineers can obtain more data to simulate and test various scenarios, leading to improved productivity, reduced downtime, and enhanced product quality. In the construction industry, virtual replicas of building designs can assist engineers to visualize and analyze the entire construction process, to improve safety measures, and to identify potential clashes, as well as wear and tear issues before the actual construction begins.

This paper categorizes DTs into the well-established hierarchy levels used in Cyber-Physical Systems (CPS) (cp.Figure 3, left) and Industry 4.0. The Reference Architecture Model Industry 4.0 (RAMI 4.0) [14] serves as a standardized conceptual model for Industry 4.0 systems (cp.Figure 3, right). According to RAMI 4.0, the asset layer represents the physical entities (mapped to real-world entity), including embedded systems, while the integration layer facilitates data exchange and contextualization (mapped to interactions among real-world entities and the transformation of real-world raw data to digital data). The communication layer ensures standardized communication with systems (mapped to the transformation of real-world raw data to digital data). The information layer focuses on intelligent data processing and analysis (mapped to DT), and the functional layer drives strategic decision-making and optimization (mapped to the goal of DT). DTs utilize CPS data to simulate and model the behavior and performance of the physical system, which is further implemented in the functional, information, and communication layers. The integration layer handles real-time data, reflecting the current state of the physical entity, as well as the digitization and integration of analog-format engineering data into the DT.

Figure 2:

Analysis of definitions to derive a cross-domain valid definition of digital twin in engineering (words used in 50 definitions to answer the questions above).

This concept of the relationship between the physical entity (Figure 4, bottom left) and the DT (Figure 4, middle left) illustrates the dynamic relationship between the physical entity and the DT within the context of the physical reality-virtuality boundary. The mapping to the Industry 4.0 levels is included (Figure 4, right). The figure showcases the diverse nature of in- and output interactions, which can range from fully automated real-time interfaces to more operator-dependent data transfers. This reciprocal interaction facilitates the desired improvement by leveraging application-specific data processing capabilities inherent in the DT. This processing is accomplished through a network of interconnected components, each varying in complexity. These components encompass a range of functionalities, from visualizing raw data, over a physics-based model, to the predictive simulation-based control loops. The complexity of the DT’s attributes is tailored to meet the specific objectives of the application, without extending beyond what is necessary to achieve the desired outcomes [17].

Figure 3:

Mapping of digital twins in CPS and Industry 4.0 architecture based on [14], [15], [16].

When considering DTs in engineering, it is important to note that they extend beyond pure simulation, which typically occurs in the design phase. While simulation is a key aspect, DTs also integrate real-time processes and alarm data from the operation of the physical entity or system throughout its entire lifecycle [18], [19]. This data-driven approach allows engineers to monitor and optimize performance, detect anomalies, and make data-informed decisions in real time [20]. To achieve industrial interoperability, information integration can be addressed via the RAMI 4.0 framework (Figure 3, right).

Figure 4:

Concept of relationship between the physical entity and the digital twin (left) and its mapping to established concepts like CPS and Industry 4.0 based on [16] (right).

The main contribution of this paper is on the one hand a comprehensive exploration of DTs in various domains like construction, building and production engineering, including the human in the loop, and on the other hand a revelation of similarities in these domains to lay the base for cross domain approaches to acquire synergies in future work. The domains in focus were chosen because manufacturing, material flow systems, machine and plant manufacturing industry and the buildings they are situated in need to be considered as a whole system of systems with a carbon and ecological footprint. The construction process to build such entities is consequently included as well.

The paper has the following structure:Section 2 introduces the aim of DTs for the different engineering domains in focus (cp.Figure 1) and presents the key characteristics used to compare the different domains in detail inSection 3.Section 4 summarizes the results and provides an outlook for future work.

2 Aim and categories of digital twins in engineering

We investigate the multifaceted utilization of DTs in different engineering domains, including construction and production. Firstly, we delve into the applications in the construction domain, which includes built environment, construction logistics, and material flow systems. Furthermore, we examine the DTs’ relevance in the production domain, including material flow systems, technical product development, production, i.e., machine and plant engineering, manufacturing process, production process, and human factors in the real world and virtual world (cp.Figure 1). Based on the definition of DT (cp.Figure 4), we classify the DTs using the following criteria: real-world entities, interlinked components, scope of the DT and application, input and output, scales, aim, challenges and outlook (cp.Tables 1 and2). The adoption of DTs in these disciplines aims to improve operational efficiency, optimize processes, and establish seamless integration between the physical and virtual realms. Different engineering domains developed diverse aims and types of DTs tailored to meet their requirements and needs. DTs in the engineering domain address a variety of use cases and focus levels and relate to products, processes, and resources. The term machine is used as a synonym for the resource of the Product Process Resource (PPR) paradigm [21]. DTs can be implemented for each hierarchical focus level and can be interdependent as modules in the form of data exchange. In the overall context of DTs in engineering, these relationships depend on the domain/discipline and the specific requirements (cp.Figure 1). In engineering design, the DT can be classified as an engineering twin, a production twin, an operation twin or a service twin, depending on the lifecycle phase the DT is contributing to [22]. In this paper, we neglect the service phase due to comprehensibility and paper length, despite the obvious benefit of a DT to offer appropriate, timely services on the one hand and to use digital information gathered during a service activity to evolve other DTs of a component on the other hand. Further, the scale of the DT is relevant: A Digital Twin Instance (DTI) represents a single asset [23]. Multiple DTIs can be linked in a Digital Twin Aggregate (DTA), enabling efficient data allocation and decision-making [24]. Two or more DTAs can communicate via a DT System (DTS), exchanging information between multiple DTAs. Data handling encompasses all the activities and technologies required to manage data flows and ensure the coherence and synchronization of information across the various components of the DT.

A DT has the purpose of providing a comprehensive information model for an asset throughout its entire life cycle. By integrating engineering information and enriching it with operational data, DTs serve as a valuable resource for learning and leveraging existing knowledge. While real-time control based on the DT is desirable, the associated effort and cost may not always be justified by the benefits gained. Depending on the application, the scope of the DT can be adjusted accordingly. It can range from visualizing sensor data for operator adjustment to implementing automatic control. The attributes of DTs can vary significantly across different subtypes, encompassing modelling approaches, data-based methods, computer-aided design (CAD) models, real-time sensor data, digital simulations, semantic technologies like ontologies to support inconsistency checking, and technologies like virtual reality (VR) and augmented reality (AR) to ease operation by humans. Typically, the DT is adapted or composed according to the specific task’s requirements.

3 Digital twins in the construction, building and production engineering domain

In the following, the aspects of DTs in the engineering domains, more specifically built environment, construction logistics, intralogistics, production engineering including additive manufacturing and battery production, and human in the workplace are discussed. First, the aims for the specific application domain, and the characteristics of DTs derived to introduce their components, interfaces, input, and output mechanisms are introduced. Because the domains differ according to different scales, like size and number of components, the contribution tries to elaborate the concept of scaling, further assuming that concepts of DT will need to be scalable if applicable in general or at least to the different domains included in this study (cp.Table 1).Tables 1 and2 summarize key insights and findings based on a literature review, as well as first and second discussion focus groups with domain experts from the different research groups included. Each subsection corresponds to specific table columns. By understanding these key characteristics, we can gain deeper insights into the unique features and functionalities of DTs in the engineering domains with focus on identifying similarities and differences respectively, and elaborating synergies in the future, while being aware of potential scalability issues. DTs have emerged as transformative, enabling the seamless integration of physical systems and their virtual counterparts. DTs seamlessly integrate physical systems with virtual counterparts, enabling data exchange, simulation, and analysis through interconnected components and interfaces. Inputs to DTs in engineering encompass a wide range of data sources, including data from the engineering and operation phases. The outputs of DTs vary depending on the application and domain. DTs exhibit scalability in multiple dimensions. They can operate in real-time or cover long-time scales, providing accurate and detailed information on projects, including geometry, physical properties, and engineering parameters [25]. They can operate at different levels of hierarchy, ranging from individual components to entire buildings, systems, factories, or supply chains. Various tools and formats, such as CAD models, simulation software, 3D models, and data analytics platforms, support the implementation and utilization of DTs. DTs operate at varying scales depending on the specific application, purpose, and lifecycle phase. DTs in engineering face significant challenges that hinder their widespread adoption and effectiveness. Future research in the field of DTs will be driven by the need to address general cross-domain challenges and explore new possibilities for a synergy-integrated DT. In the context of automation in different domains, ongoing research aims to enhance the real-time aspect of DTs, ensuring production safety and stability. This involves exploring and leveraging hardware advancements to enable faster system updates and improving the bidirectional flow of information between the DT and real-world assets.

3.1 Built environment

3.1.1 Aims

DTs in the built environment exist in all life cycle stages. Compared to other industries – with the exception of e.g. large scale process plants – buildings are typically unique and only built once. Predictive models have to be exact, since there is no real-scale physical testing and validation phase on a regular basis (as e.g. in the car industry). Monitoring data is collected from the construction phase onwards and can lead to an accurate estimation of a building’s condition if linked to a DT. Additionally, Environmental impact should also be considered, as it encompasses measurable effects that must be modeled to account for their interaction with the built environment. Examples include wind and mud floods. Of particular interest is the monitoring of the energy consumption for operating a building, especially concerning the thermal comfort of its users and the employment of smart control loops. During building construction, the monitoring of energy consumption is also important as underscored by the telemetry guideline AEMP that references the ISO 15143-3 where several elements such as “fuel used in past 24 h” are defined [26]. Apart from these observations, structural DTs in the built environment can be used to investigate and steer or optimize future structural behavior and usage, thus ensuring that maintenance and renovations are performed efficiently and sustainably, and ultimately enabling a significantly extended lifetime of the structures [27], [28]. This is of particular relevance for maintaining the large stock of existing and aging built infrastructure in Europe and beyond [29]. A major challenge lies in the fact that for most of the existing built facilities, digital models do not yet exist. Hence, they must be created by automated methods by processing a multitude of sensor data as well as legacy 2D drawings, eventually resulting in high-quality semantic-geometric replicas on the required level of detail [30], [31], [32], [33].

3.1.2 Components and interfaces

The components and interfaces of DTs in the built environment encompass data acquisition, transmission, modeling, integration, and services. These elements incorporate technologies such as sensors, data models, Building Information Modeling (BIM), Discrete Event Simulation (DES), and data transmission protocols to enable seamless connectivity between physical and virtual counterparts, facilitating data exchange and analysis to enhance construction processes and project outcomes. In the context of the built environment, buildings, and infrastructure, there are interlinked components and interfaces that go beyond traditional CAD/BIM models. These components include a CAD/BIM model with multiple “layers” that captures the intricate details of the structure, a Finite Element Method (FEM) model for analyzing structural behavior, and interfaces for monitoring data and environmental impact data [27], [34].

3.1.3 Inputs

In domains, such as built environment, input data can include information related to material properties and distribution, constructive design and detailing, monitoring data for damage detection, and environmental influences like wind, water, snow, earthquakes, heat and fire, including any changes in these factors. Additionally, authoritative requirements play a significant role in defining the input data for DTs, ensuring that the twin aligns with specific regulations or standards. By collecting and incorporating data from these diverse sources, DTs can capture a holistic view of the physical system and its surrounding environment [23].

3.1.4 Outputs

The output of DTs in the built environment encompasses essential aspects such as load transfer in structures and parameters for intelligent control loops, particularly in the context of heating systems.

3.1.5 Scale

The scale and tools of DTs in the built environment vary depending on their specific purpose. For instance, resources such as humans, equipment, and materials on-site can be monitored using sensors or cameras [35], [36], and their behavior can be simulated using DES techniques. In the case of BIM, the scale is determined by the level of detail (LoD), encompassing various aspects such as geometry, quality, time, and costs [37]. In the domain of construction, a consensus has been reached regarding the dispensability of real-time updates, favoring instead “right time” updates informed by sensor data as generally satisfactory. This is exemplified by the customary practice of inspecting bridges on an annual basis, which is deemed satisfactory in most cases. Depending on the varying requirements, the update cycles differ accordingly.

3.1.6 Challenges

In the built environment, adopting DTs is particularly complex due to construction-specific requirements. Factors such as the unique nature of each construction site, transient processes, dependence on location and weather conditions, the use of diverse technologies, industry fragmentation, and segmentation along the product life cycle or process chain pose challenges to implementing DTs. Managing the exchange of heterogeneous data throughout a construction project’s life cycle is crucial before DTs can be effectively adopted. Addressing these challenges requires the development of interfaces that enable seamless data exchange between DTs and BIM systems, ensuring the redistribution and communication of relevant results [38].

3.1.7 Outlook

In the built environment, DTs have the potential to improve resource efficiency by enhancing built structures, predicting changes in usage, and assessing environmental impacts. Future research in this area will focus on refining communication and interfaces between the diverse tools used for data gathering, monitoring, and simulation in complex boundary conditions.

3.2 Construction logistics

3.2.1 Aims

DTs in the construction logistics can achieve aims such as improved project visualization and planning, enhanced construction monitoring and control, better collaboration and communication, enhanced safety and risk management, efficient facility management and maintenance, and data-driven performance analysis and optimization [25]. There exists a differentiation between the terms “construction digital twin (CDT)” and the “digital twin in construction (DTC)” [39]. The CDT focuses primarily on the technology and digital representation of the construction site [40], whereas the DTC expands the concept to encompass a comprehensive workflow and methodology for utilizing DTs to monitor and optimize construction activities [40], [41], [42]. The DT concept assumes a pivotal role in additive manufacturing within the construction industry, facilitating the monitoring and detection of discrepancies that may arise between the intended design and the as-built geometries [43].

3.2.2 Components and interfaces

In the construction logistics, components and interfaces of DTs play a vital role in unlocking the full potential of data-driven engineering and optimization. Through machine-readable databases, physical simulation software, and integration with diverse sensors and models, DTs enable real-time monitoring, visualization, and predictive analysis in engineering.

3.2.3 Inputs

The input for construction logistics in DTs includes historical models which provide insights from past projects, and real-time models that capture the status of ongoing construction activities.

3.2.4 Outputs

The construction logistics DT generates an information model that captures the current state of a project, encompassing both the as-planned and as-performed aspects. This model provides valuable insights into the project’s progress, allowing stakeholders to compare the planned outcomes with the actual results and to make informed decisions.

3.2.5 Scale

DTs in construction logistics operate across a range of scales, bridging different timeframes, sizes, accuracy levels, hierarchies, and life cycle phases. In terms of time, DTs can operate at a granular level, such as seconds for tasks related to maintenance and material allocation or they can take up to days and weeks for scheduling purposes. The size of the DT can vary significantly, ranging from small devices like equipment sensors to entire buildings. The accuracy of the DT depends on the sensors used to capture and interpret data. The hierarchy of the DT is determined by the specific objectives, with the potential to cover different levels, from equipment maintenance and operation to a comprehensive building model. Finally, the DT can span various life cycle phases, including design, planning, execution, facility management, and even demolition.

3.2.6 Challenges

The construction industry poses a challenge in the adoption of DTs from Industry 4.0, as it requires careful integration of these technologies while considering the industry’s unique requirements and complexities.

3.2.7 Outlook

The outlook for DTs in the construction logistics involves addressing limitations related to object-like characteristics and expanding the application of BIM to encompass other construction types beyond its current scope.

3.3 Material flow systems

3.3.1 Aims

The efficient management of material flow systems is essential for ensuring flexibility in production supply chains, especially in times of supply chain difficulties. In the domain of intralogistics, there is to the best of our knowledge no unified definition of Digital Twin, but the concepts defined in this paper are incorporated in fleet management systems. DTs in intralogistics optimize performance and reduce logistics costs by analyzing factors such as distance, traffic conditions, and load capacity to determine the most efficient transportation routes. By integrating real-time inventory data, demand forecasting algorithms, and order profiles, DTs enhance order fulfilment efficiency, improve inventory control, and minimize stockouts within material flow systems.

3.3.2 Components and interfaces

For DTs of material flow systems, the real-world entities include vehicles, doors, gates, and stationary conveyor technology, while the virtual entities consist of an environment model and a decision unit, which is a control system that decides on scheduling, routing/traffic control, and dispatching.

3.3.3 Inputs

In the context of material flow systems, input data encompasses vehicle positions and transport orders.

3.3.4 Outputs

The output of DTs in material flow systems can include various elements such as driving orders and providing essential information for optimizing and managing the flow of materials within the system.

3.3.5 Scale

The DT for material flow systems is designed to provide a high-resolution perspective with a timeframe of seconds and an 8-h horizon. They correspond to the size of production or storage systems, as the intralogistics system extends across these systems. Therefore, the system can encompass a wide range of sizes, from 5 × 5 to 500 × 500 m, ensuring accuracy and precision at a resolution of 20 cm. It operates within a hierarchical structure that integrates Enterprise Resource Planning (ERP), control systems, and vehicles. This DT is applicable during the operational phase and, if required, can also be utilized in the planning phase, to enable efficient management and optimization of intralogistics and transport operations.

3.3.6 Challenges

One major challenge in material flow systems is the design of an efficient and standardized interface that can communicate with both higher-level and lower-level systems. Other challenges include ensuring the flexibility of control software, pre-controlling heterogeneous transport systems with multiple master controllers, and increasing expandability through interface standardization. For all three Industry 4.0 features, the key challenge is the efficient creation and evolution of the knowledge base [44]. This challenge has been initially addressed by generating executable IEC 61131-3 code from SysML-AT engineering models [45] or generating knowledge from ontologies [44] [46], and recently derived knowledge from operation data [47]. Consequently, the DT includes all these models and their runtime versions of the knowledge base as part of the automation architecture. Because operators need white box or grey box models to trust the system, appropriate visualization is another prerequisite [48].

3.3.7 Outlook

In the domain of material flow systems, future research will center on big data evaluations. This includes predictive maintenance, layout optimization, manpower deployment, and vehicle deployment. Additionally, exploring variable control hierarchies, such as centralized and decentralized approaches, will increase system resilience. The inclusion of more vehicle functions will enhance effectiveness, and standardizing interfaces will facilitate system expandability.

3.4 Machine and plant manufacturing industries

3.4.1 Aims

In the Automated Production Systems domain, several definitions of a DT to digital shadow exist [49], [50]. In the frame of the Bavarian KI.Fabrik [51] the DT includes, depending on the use case, all digitally available and useful models containing information about the product, the production process, and the production resource. Referring to adaptive and flexible production plants including both manufacturing and intralogistics, the digital information created in the design phase can be used as input for the runtime knowledge base and can enable the adaptation in case of a faulty device, the automatic adaptation to changed product requirements, and the integration of new physical components without downtime [47]. In engineering of machines and plants, DTs are virtual instances of a physical component that aim to include all as-built information necessary for start-up, operation, maintenance and service. DTs are desired to enable bi-directional information flow between the physical and virtual worlds to be realized automatically and optimally following real-time requirements.

3.4.2 Components and interfaces

In machine and plant engineering, the data twin, which encompasses a machine-readable database using technologies such as AAS or ontologies, represents the digital counterpart of physical entities. Visualization of the DT is achieved through physical simulation software like Nvidia Omniverse.

3.4.3 Inputs

The input of DTs in machine and plant engineering encompasses crucial engineering data, such as asset’s CAD models with computer-aided technologies (CAX) obtained during the design phase, as well as operational data collected from Internet of Things (IoT) devices throughout the production process. For adaptive production plants, inputs from the design phase serve as valuable input for the runtime knowledge base, enabling automatic adaptation, fault detection, and integration of new components without downtime.

3.4.4 Outputs

DTs in machine and plant engineering offer valuable output, including real-time information on asset conditions, predictive maintenance hints to enhance reliability, and reconfiguration/optimization parameters to improve operational efficiency and performance, which can be measured by overall equipment effectiveness (OEE).

3.4.5 Scale

In the context of machine and plant engineering, the requirements regarding time scale are, on the one hand, microseconds in the case of synchronized drives and, on the other hand, the longevity of machine and plant operation of more than 30 years. The size scale is equally broad, from atomic components like connectors (electrical, mechanical) to the entire plant in km. As a supporting structure, standards like the levels of ISA 88 are well established to classify the hierarchy of the plan components, and the life cycle models like VDI 2206 [52] describe the different phases. The AAS represents the German concept to enable the integration of new components. OPC-UA is proposed for vertical information integration [53]. AutomationML is the standard for exchange of data between different tools of the different disciplines and phases of the life cycle [18]. To achieve industrial interoperability, information integration can be addressed via the RAMI 4.0 framework (Figure 3). This involves using AutomationML during asset development and planning phases, OPC UA for asset data communication during productive use and maintenance, and AAS for communication within the manufacturer's value chain network across all lifecycle phases (cpFigure 5) [18].

Figure 5:

Information integration through AutomationML, OPC UA, and AAS via the RAMI 4.0 (adapted from [18]).

3.4.6 Challenges

In machine and plant engineering, one major obstacle is the systematic management of information and models, considering different component manufacturers, data exchange formats, inconsistent or incomplete data sources, and the lack of labeled and contextualized data. To overcome these challenges, several considerations must be addressed. This includes enriching and combining existing engineering models, ensuring data and model consistency, leveraging small data and rare events, enabling the evolution of DTs over time, and evaluating real-world applications. The integration of real models in the DT encompasses the incorporation of component suppliers at the lowest level, extending to the entire plant. This integration goes beyond structural information and encompasses behavioral aspects, as well as the complex challenges of fault handling and fault recovery. DTs for selected components, such as valves and motors, are commonly found in heterogeneous data formats. These formats range from unstructured ones like PDF to structured formats like STEP, AML, URDF, XML, and even formalized data like data in ontologies. These DTs primarily provide device specification data, often in formats like ECLASS, but simulation data is rarely available. DTs of machines and plants are missing besides AR models of buildings and plants to discuss with customers during the engineering process or for the operator training [54]. Consequently, tools like Unity [55] and Omniverse [56] are pushing into the market from the engineering perspective, and furthermore, niche tools like machineering [57] provide simulations close to the machine level [46], [58].

3.4.7 Outlook

In machine and plant engineering, DTs also serve as the foundation for other digital technologies to build upon. For example, integrating AI models into DTs enables future predictions, such as cost and emission estimations at the early stages of product development. DTs can also track a product’s carbon footprint throughout its lifecycle, facilitating the creation of prediction models that inform design improvements. Additionally, exploring generative and parametric design techniques within the DT framework can enhance product performance and efficiency.

3.5 Manufacturing process by using the example of additive manufacturing

3.5.1 Aims

In addition to the machine twin, the process twin also considers resulting process conditions, like temperature distributions, to represent the manufacturing process. The process twin is connected to the process result. Among the different production processes in a factory, this paper focuses in detail on additive manufacturing technologies, due to this process category’s complexity, lengthy duration, and already high degree of digitalization. The aim of DTs in additive manufacturing is centered on process optimization and real-time monitoring, and control of the additive manufacturing process. DTs aim to optimize the additive manufacturing process by predictively simulating and analyzing various parameters such as material properties, machine settings, and process conditions to optimize the planned process and to monitor and control the process to avoid deviations from the planned process due to process instabilities and material irregularities [59]. The application of DTs in the field of additive manufacturing is still in its early stages [60]. However, a first generation of DTs is beginning to emerge [61]. While the potential benefits of DTs in additive manufacturing are understood, current implementations often lack crucial aspects required to fully unlock their capabilities [62].

3.5.2 Components and interfaces

To accurately mirror the production process in additive manufacturing, one approach relies on a grey box model consisting of two main components. The physics-based models capture essential aspects of the process physics, such as energy absorption, thermal dynamics, and in powder bed fusion, which is a category of additive manufacturing processes, while the AI-based model compensates for uncertainties arising from the simplifications in the physics model. Creating DTs in additive manufacturing requires multiscale and Multiphysics models, but linking submodels and scarcity of experimental data pose difficulties [63].

3.5.3 Inputs

In the context of additive manufacturing, a DT requires multiple input streams. Real-timein-situ sensor data, such as temperature and topography, provide immediate feedback for timely decision-making [59], while a continuously growing data pool, consisting of statistical data and physics-based process models, facilitates interpretation and analysis of sensor streams [64], [65]. Depending on the type of DT, process twin or a DT for process-structure-property relationships in manufacturing processes, such as additive manufacturing, additional information of the part produced can be added. Some researchers go even further and recommend an additive manufacturing data model containing product lifecycle data [66].

3.5.4 Outputs

In terms of additive manufacturing, the output of the DT must be sufficient to enable timely and effective control of the process for the process twin. This involves providing relevant information about process parameters that have a direct and immediate impact on the desired outcomes, such as input power or process speed. By monitoring and adjusting these parameters in real time, the DT facilitates efficient process control and optimization [66]. Another output could be the link between the process behavior or rather the process signature with part properties.

3.5.5 Scale

Additive manufacturing DTs have dual foci, with one on the small-scale process zone where material is currently being added, and the other on the state of the already manufactured part represented by the deposited material and associated data in the DT [67]. Compared to the aforementioned DT of the machine and plant manufacturing industry the additive manufacturing DT is always rather on the part scale of several mm to cm.

3.5.6 Challenges

For DTs in additive manufacturing, challenges arise from the speed and the complexity of the process. The high control frequency required for fast processes and the complexity of modeling and capturing sensor data pose difficulties. Additionally, the multitude of additive manufacturing processes, each with its dynamics and influencing mechanisms makes it challenging to acquire sufficient data to model and optimize the process, especially due to the lack of standardization between original equipment manufacturers (OEMs) in machine and software frameworks [60].

3.5.7 Outlook

Further research is required to fully exploit the potential of DTs in additive manufacturing. This includes addressing challenges related to the speed of the process, capturing data from different additive manufacturing techniques, and standardizing interfaces for seamless data integration from various machines and software frameworks. Furthermore, the efficient creation of multi-scale Multiphysics models is a barrier for the DT in additive manufacturing [68]. Overcoming these challenges will unlock the ability of DTs to optimize additive manufacturing processes and improve product performance.

3.6 Production process by using the example of battery production

3.6.1 Aims

In contrast to the previously discussed manufacturing process in the factory, battery production represents a complex system of highly linked and cost-intensive process steps in a dynamically evolving environment [69]. At the core of lithium-ion battery cell production are electrode manufacturing, cell assembly and cell finalization with up to 15 separate process steps. The electrodes undergo multi-stage manufacturing processes that affect their product properties and are further processed in the cell assembly and cell finalization stages to build a lithium-ion battery cell. These stages exhibit a significant number of cause-and-effect relationships and time-consuming testing of intermediate product features. This is also associated with the challenging transition from batch production of electrodes to unit production of cells. To cope with the complexity of the process chain, it is necessary to build up a profound understanding of material-process-quality-property relationships [70]. DTs are an emerging trend in battery production and aim to overcome the costly trial-and-error method and enable real-time process control and quality monitoring in a digitized production environment [71]. This is accompanied by the overall goal of reducing scrap rates in lithium-ion battery production and improving in-depth process knowledge to enable quality predictions of product features along the process chain. These goals are motivated by high material costs, the aforementioned complex cause-and-effect relationships, and the associated sensitivity to manufacturing defects. The combined use of DTs and data-based analytics enables, e.g., an improvement in the product quality, safety and lifetime aspects of the lithium-ion battery cells in operation through the connectivity with the broader battery production ecosystem [72]. The use of DTs further facilitates the introduction of a digital battery passport for future batteries by providing aggregated manufacturing data.

3.6.2 Components and interfaces

In battery production, a DT needs to encompass multiple dimensions, which according to Krauß et al. [73] includes the Digital Product Twin, Digital Machine Twin, and Digital Building Twin. Such a DT for battery production requires high integration, adaptability to changing production conditions, and real-time incorporation of new data and insights along the interconnected process chain [74]. A comprehensive tracking and tracing system is essential to continuously incorporate data into the DTs along the process chain [75].

3.6.3 Inputs

Depending on the scope of DT in battery production, the input data stream may include various aspects, including raw material properties, electrode mesostructures from simulation models, inline sensors for individual processes, and machine features [72], [74], [76].

3.6.4 Outputs

In battery production, the outputs can vary depending on the specific use cases and the focus level of the DTs [77]. This can include information about transportation orders to optimize material flow, product quality parameters, and potential alterations in the production process [78]. Such outputs form the basis for multi-objective structural optimization models, guiding decision-making and identifying opportunities for improvements regarding efficiency and performance [72]. The coupling of input and output along the process chain to link models and parameters is addressed, for example, by a digitization platform to analyze interdependencies at the product, process and manufacturing levels [79].

3.6.5 Scale

In battery production, DTs need to be able to perform multi-scale mapping of various product features [80]. This includes the molecular and particle scale (micro level) considering material properties, the component scale (millimeter to centimeter) analyzing interactions between battery components, battery packs in the vehicle application and plant scale (meter level) encompassing production facilities and assets, and the factory scale considering the entire factory and its optimization across production equipment, logistics systems, and business processes [77], [78], [80].

3.6.6 Challenges

The challenge of implementing DTs in battery production beyond existing concepts is particularly associated with the aforementioned high complexity and cause-effect relationships along the process chain. Challenges for the successful implementation of DTs in battery production also include heterogeneous and voluminous data sources, long computation times of microstructural simulation models, and low compatibilities between simulation environments [71], [72], [74]. This applies both to individual process steps along the process chain and in the overall context of a battery cell production factory.

3.6.7 Outlook

The use of DTs in battery production throughout the life cycle enables product quality improvement. Future research is needed to further advance the implementation of DTs. This includes the integration of different inline sensors for individual process steps for quality monitoring, the implementation of a complete tracking and tracing system to enable continuous data mapping along the process chain, the application of AI-based methods in DTs, and the further development of existing simulation models in terms of computation time and compatibility.

3.7 Technical product development

3.7.1 Aims

Tools and procedures become more and more digitized as technical product development methods. A crucial aspect of this digital transformation is the integration of DTs during the design phase [11]. In technical product development, a DT serves as a virtual dynamic representation of a physical system, enabling bidirectional data exchange over its entire lifecycle [22].

3.7.2 Components and interfaces

The DT in technical product development captures and converts data from the initial design phase to manufacturing, operation, maintenance, and disposal, facilitating analysis, experimentation, and refinement of the product’s performance before physical construction [17], [81]. Similarly, an operation twin oversees real-time performance, enabling predictive maintenance and data-driven troubleshooting to enhance dependability and availability. Additional subtypes of DTs include production twins, which focus on the production process; end-of-life (EoL) twins for sustainability considerations [82], and DTI, which represents the final physics. These DT subtypes offer opportunities to optimize cost structures, simulate and predict costs over the lifecycle, and drive improvements specific to their respective domains.

3.7.3 Inputs

DTs of technical product development can receive inputs from various sources, such as requirements, 3D models, operational stages, processes, and metadata, enabling comprehensive product or process-related data integration. In general, data is sourced from various entities, including sensors, machines, and humans.

3.7.4 Outputs

In technical product development, DTs provide valuable outputs such as recommendations, design parameters, control parameters, requirements, and system models. These outputs guide decision-making, specify design elements, optimize product performance, capture essential criteria, and enhance understanding of the product system.

3.7.5 Scale

DTs for technical product development can have varying scales depending on the lifecycle phase and use case. For instance, the digital twin prototype (DTP) captures significant information during the early stages of product development. As the prototype evolves, it transforms into a DTI, which represents the final physical product [23]. Multiple DTIs can be linked together in a DTA, enabling high-level data allocation and decision-making [24].

3.7.6 Challenges

For technical product development, non-technical issues, such as a shortage of specialists and expertise, can impede the implementation of DTs. However, technical challenges, including the absence of standardized data and models, also need to be addressed. These challenges highlight the need for comprehensive solutions and interdisciplinary collaboration [83].

3.7.7 Outlook

The outlook for technical product development using DTs is promising, as they serve as the foundation for integrating digital technologies, enabling the development of prediction models, and ultimately leading to improved product performance and optimization.

3.8 Humans in the loop

3.8.1 Aims

In all of the aspects regarding the factory and the construction, the humans or rather keeping the humans in the loop is a key aspect. Human DT models enable the development of machines and systems that prioritize human needs. Furthermore, human models can be utilized to ensure and further optimize the suitability of the system during its operational phase [84], [85]. Human DTs are digital representations of human beings that can capture their physical [86], [87], behavioral [88], [89], social [90], physiological [91], psychological [92], cognitive [93], and biological [94] aspects. They can be used to monitor, simulate, and optimize human performance and well-being in the context of production processes. Human DTs share the basic concept of DTs with other physical systems and processes. However, human DTs are more complex and challenging to develop and maintain due to the dynamic nature of humans, the challenges towards sensors and models caused by it, and data protection/privacy rights.

3.8.2 Components and interfaces

Human DTs involve multiple components that reflect different levels of abstraction and representation. Integrating models of cognitive and physical aspects poses a significant research challenge to establish a coherent and dynamic system. The linking of these individual components requires bridging gaps between hierarchical levels of representation and standardization efforts.

3.8.3 Inputs

Human DTs require inputs related to working conditions and properties of the workforce, which include demographics, physical conditions, mental conditions, and environmental factors.

3.8.4 Outputs

Human DTs provide outputs on micro and macro ergonomic levels, including physical, cognitive, and emotional strain, as well as information on skill, training, and organizational circumstances. On a macro ergonomic scale, the DT provides valuable information about skill levels, training needs, and organizational circumstances. This information is crucial for optimizing the deployment, training, and interaction of the available workforce, ensuring that the right skills are utilized for specific tasks and improving overall productivity. By combining micro and macro ergonomic considerations, DTs enable the adaptation of workplaces to individual workers, enhancing their comfort, safety, and efficiency.

3.8.5 Scale

In the realm of human DTs, the scale ranges from micro to macro ergonomic aspects. The micro ergonomic analysis involves assessing individual worker performance concerning environmental factors, which requires data at the individual level on a scale of seconds. Conversely, macro ergonomic assessments, such as skill and training management, focus on organizational aspects and can span larger time scales of weeks or months. The data for macro ergonomic analysis also needs to be individualized to the human, but the slower pace of change allows for longer processing times. Consequently, the multidimensional scale needs to be as accurate as necessary to enable the intended use of human DTs, while considering the relevant physical, behavioral, social, physiological, psychological, cognitive, and biological diversity.

3.8.6 Challenges

In the context of human DTs, one of the biggest challenges is the availability and acquisition of data [95]. The amount and quality of data required for a human DT are substantial, often requiring specialized experimental settings, questionnaires, or professional evaluation [96]. Real-time data acquisition is hindered by the lack of automated processes and the need for expensive, body-worn sensors [97] that disrupt the natural workflow and raise privacy concerns. Additionally, certain data, such as training and skill levels of individuals, is often not recorded. Balancing the individualization of data with data privacy is an ethical and legal dilemma that needs careful consideration.

While DTs in engineering offer immense potential, overcoming these challenges will be essential to fully leverage their benefits and drive advancements in various domains. Addressing these challenges requires a holistic approach that considers the technical aspects, the system architecture, and an industry-wide collaboration [95].

3.8.7 Outlook

In the realm of human DTs, future research will focus on the development of individually adaptable work environments. By creating human-centered workplaces, productivity and worker satisfaction can be increased while minimizing the risk of injuries and human errors. This necessitates the acquisition and processing of individualized data, overcoming challenges related to data availability, privacy protection, and automation of data acquisition.

4 Conclusion and future work

In light of the changing dynamics of product longevity and documentation, this paper emphasizes the advantages of more sustainable and efficient management of engineering data in industrial products and machinery. The shift towards the DT technology not only can improve the management of engineering information but also enhances the sustainability of machinery operations, evidenced by longer operational lifespans and reduced resource consumption in retrofitting processes.

The analysis highlights the potential for cross-domain synergies and the continuous evolution of DTs across product and system lifecycles, addressing key challenges like data management and model-data consistency that is key to sustainable and long living products. DTs should continuously evolve to meet the dynamic needs of modern complex product and system management. First, DTs should adapt and refine models based on (real-time) data inputs from various lifecycle stages, ensuring accuracy and relevance. Second, integrating advanced technologies like data models and simulation software enhances DTs’ predictive capabilities and decision-making processes. Third, cross-domain scalability allows DTs to evolve with expanding and complex systems, while maintaining their effectiveness as application scope grows.

In conclusion, this paper has explored the concept of DTs and their applications in selected domains: construction, including the built environment, construction industry and material flow systems; production engineering, including machine and plant manufacturing, production development process, manufacturing process with the example of additive manufacturing, production process with the example of battery production; and human in the loop (cp.Tables 1 and2). The objective is to provide a comprehensive understanding of DTs, their role in systems integration, and the challenges associated with their implementation. DTs have emerged as powerful tools for optimizing and improving processes, enabling adaptive and flexible production plants, enhancing human work environments, and facilitating data-driven decision-making. They offer a holistic and integrated approach by bridging the physical and digital worlds, capturing real-time data, and providing valuable insights for performance optimization and predictive maintenance. The development of DTs requires careful consideration of input sources, which can vary depending on the scale and application. These sources include data from sensors, machines, humans, and authoritative requirements. By harnessing these inputs, DTs can create a complete information model that spans the entire life cycle of an asset, incorporating engineering information and operational data for continuous learning and improvement. The outputs of DTs play a crucial role in process control, workforce optimization, structural assessment, and project management. Macro ergonomic outputs inform decision-making regarding skill levels, training, and organizational circumstances, optimizing workforce deployment and interaction. On the other hand, timely process control is facilitated by outputs that focus on relevant process parameters, enabling efficient and effective control of operations.

The success of DTs relies on addressing challenges such as data integration, standardization, scalability, and the utilization of advanced modeling techniques. Overcoming these challenges requires interdisciplinary collaboration, technological advancements, and a focus on developing robust methodologies and tools. By understanding the key characteristics of the engineering domains, the paper provides insights into the unique features and functionalities of DTs in different engineering domains to identify similarities and differences respectively and elaborate synergies in the future. As DT technology continues to evolve and mature, it holds great promise for driving innovation, productivity, and sustainability across a wide range of industries. Future research should focus on further refining and expanding the capabilities of DTs, exploring new applications, and addressing emerging challenges to unlock their full potential. In summary, DTs have the potential to revolutionize engineering and other domains, offering new opportunities for optimization, efficiency, and resilience. By embracing the power of DTs, organizations can gain a competitive edge in the digital era and pave the way for a smarter and more interconnected future.

Corresponding author: Birgit Vogel-Heuser,Technical University of Munich, Institute of Automation and Information Systems,Boltzmannstr. 15, 85748Garching bei München,Germany, E-mail: vogel-heuser@tum.de

About the authors

Birgit Vogel-Heuser

Univ.-Prof. Dr.-Ing. Birgit Vogel-Heuser received a Diploma degree in Electrical Engineering and a Ph.D. in Mechanical Engineering from RWTH Aachen. Since 2009, she has been a full professor and director of the Insititute of Automation and Information Systems at the Technical University of Munich (TUM). Her current research focuses on systems and software engineering. She is a member of the acatech (German National Academy of Science and Engineering), editor of IEEE T-ASE, and IEEE Fellow and member of the science board of MIRMI at TUM.

Katrin Wudy

Katrin Wudy, Prof. Dr. -Ing., holds the Professorship of Laser-based Additive Manufacturing at the Technical University of Munich. Her focus is on laser- and powder based additive manufacturing technologies. She bridges the disciplines of materials science, manufacturing and photonics. Prof. Wudy finished her PhD at the Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) in the field of powder and beam-based additive manufacturing in 2017, after which she headed the “Additive Manufacturing” working group at the Institute of Polymer Technology until 2019 and was Managing Director of the CRC 814 “Additive Manufacturing” at the FAU. She was appointed tenure-track assistant professor in 2019, and promoted to associate professor in 2024.

Acknowledgment

The authors wish to thank the Bavarian State Ministry for Economic Affairs, Regional Development and Energy (StMWi) Lighthouse Initiative KI.FABRIK, (Phase 1: Infrastructure and R&D program, grant no. DIK0249), for the funding of the project.

Research ethics: Not applicable.
Informed consent: Informed consent was obtained from all individuals included in this study, or their legal guardians or wards.
Author contributions: The authors have accepted responsibility for the entire content of this manuscript and approved its submission.
Competing interests: The authors state no conflict of interest.
Research funding: None declared.
Data availability: The raw data can be obtained on request from the corresponding author.

References

[1]R. Weiß, A. Hilmer, and J. Friedrich,Baubetriebslehre – Kosten- und Leistungsrechnung – Bauverfahren, Wiesbaden, Vieweg+Teubner Verlag, 1998.Search in Google Scholar

[2]B. Vogel-Heuser, J. Fischer, S. Feldmann, S. Ulewicz, and S. Rösch, “Modularity and architecture of PLC-based software for automated production Systems: an analysis in industrial companies,”J. Syst. Softw., vol. 131, pp. 35–62, 2017.https://doi.org/10.1016/j.jss.2017.05.051.Search in Google Scholar

[3]B. Vogel-Heuser, A. Fay, I. Schaefer, and M. Tichy, “Evolution of software in automated production systems: challenges and research directions,”J. Syst. Softw., vol. 110, pp. 54–84, 2015.https://doi.org/10.1016/j.jss.2015.08.026.Search in Google Scholar

[4]Q. H. Dong and B. Vogel-Heuser, “Modelling technical compromises in electronics manufacturing with BPMN+TD – an industrial use case,”IFAC-PapersOnLine, vol. 54, pp. 912–917, 2021.https://doi.org/10.1016/j.ifacol.2021.08.108.Search in Google Scholar

[5]B. Vogel-Heuser, C. Legat, J. Folmer, and S. Feldmann, Researching Evolution in Industrial Plant Automation: Scenarios and Documentation of the Pick and Place Unit, 2014,” Tech. Rep. TUM-AIS-TR-01-14-02 [Online]. Available athttps://mediatum.ub.tum.de/node?id=1208973.Search in Google Scholar

[6]E. Arroyo, M. Hoernicke, P. Rodríguez, and A. Fay, “Automatic derivation of qualitative plant simulation models from legacy piping and instrumentation diagrams,”Comput. Chem. Eng., vol. 92, pp. 112–132, 2016.https://doi.org/10.1016/j.compchemeng.2016.04.040.Search in Google Scholar

[7]ISO/DIS 23247-1,Automation Systems and Integration — Digital Twin Framework for Manufacturing — Part 1: Overview and General Principles, Switzerland, International Organization for Standardization Geneva, 2020.Search in Google Scholar

[8]ISO/DIS 23247-1,Automation Systems and Integration — Digital Twin Framework for Manufacturing — Part 2: Reference Architecture, Switzerland, International Organization for Standardization Geneva, 2020.Search in Google Scholar

[9]ISO/DIS 23247-1,Automation Systems and Integration — Digital Twin Framework for Manufacturing — Part 3: Digital Representation of Manufacturing Elements, Switzerland, International Organization for Standardization Geneva, 2020.Search in Google Scholar

[10]ISO/DIS 23247-1,Automation Systems and Integration — Digital Twin Framework for Manufacturing — Part 4: Information Exchange, Switzerland, International Organization for Standardization Geneva, 2020.Search in Google Scholar

[11]O. Isaksson and C. Eckert,Product Development 2040, the Design Society, 2020.10.35199/report.pd2040Search in Google Scholar

[12]M. M. Ergün, A. Kocabay, Y. M. Yesilcimen, and M. Turanli Parlaktuna, “Digital twin and its applications,” inIndustry 4.0, A. Azizi, and R. V. Barenji, Eds., Singapore, Springer Nature Singapore, 2023, pp. 151–170.10.1007/978-981-19-2012-7_7Search in Google Scholar

[13]J. Stjepandić, M. Sommer, and S. Stobrawa, “Digital twin: conclusion and future perspectives,” inDigiTwin: An Approach for Production Process Optimization in a Built Environment, J. Stjepandić, M. Sommer, and B. Denkena, Eds., Cham, Springer International Publishing, 2022, pp. 235–259.10.1007/978-3-030-77539-1_11Search in Google Scholar

[14]Deutsches Institut für Normung: DIN,DIN SPEC 91345 Standard: Reference Architecture Model Industrie 4.0 (RAMI4.0), Breuth, 2016. Available at:https://www.beuth.de/en/technical-rule/din-spec-91345/250940128.Search in Google Scholar

[15]M. Broy, Ed.,Cyber-Physical Systems: Innovation durch softwareintensive eingebettete Systeme, Berlin, Springer, 2010.10.1007/978-3-642-14901-6Search in Google Scholar

[16]V. Melo, F. de La Prieta, and P. Leitão, “Alignment of digital twin systems with the RAMI 4.0 model using multi-agent systems,” inService Oriented, Holonic and Multi-Agent Manufacturing Systems for Industry of the Future, T. Borangiu, D. Trentesaux, and P. Leitão, Eds., Cham, Springer International Publishing, 2023, pp. 23–35.10.1007/978-3-031-24291-5_2Search in Google Scholar

[17]S. Schweigert-Recksiek, J. Trauer, C. Engel, K. Spreitzer, and M. Zimmermann, “Conception of a digital twin in mechanical engineering – a case study in technical product development,”Proc. Des. Soc.: Des. Conf., vol. 1, pp. 383–392, 2020.https://doi.org/10.1017/dsd.2020.23.Search in Google Scholar

[18]R. Drath, et al.., “Diskussionspapier--Interoperabilität mit der Verwaltungsschale, OPC UA und AutomationML,” 2023. Available at:https://industrialdigitaltwin.org/wp-content/uploads/2023/04/Diskussionspapier-Zielbild-und-Handlungsempfehlungen-fuer-industrielle-Interoperabilitaet-5.3.pdf Accessed: May. 28, 2024.Search in Google Scholar

[19]B. Vogel-Heuser and K. Bengler, “Von Industrie 4.0 zu Industrie 5.0 – Idee, Konzept und Wahrnehmung,”HMD Praxis der Wirtschaftsinformatik, vol. 60, pp. 1124–1142, 2023.https://doi.org/10.1365/s40702-023-01002-x.Search in Google Scholar

[20]J. Trauer, S. Schweigert-Recksiek, L. Onuma Okamoto, K. Spreitzer, M. Mörtl, and M. Zimmermann, “Data-driven engineering – definitions and insights from an industrial case study for a new approach in technical product development,” inBalancing Innovation and Operation, The Design Society, 2020.10.35199/NORDDESIGN2020.46Search in Google Scholar

[21]M. Seitz, F. Gehlhoff, L. A. Cruz Salazar, A. Fay, and B. Vogel-Heuser, “Automation platform independent multi-agent system for robust networks of production resources in industry 4.0,”J. Intell. Manuf., vol. 32, no. 7, pp. 2023–2041, 2021.https://doi.org/10.1007/s10845-021-01759-2.Search in Google Scholar

[22]J. Trauer, S. Schweigert-Recksiek, C. Engel, K. Spreitzer, and M. Zimmermann, “What is a digital twin? – Definitions and insights from an industrial case study in technical product development,”Proc. Des. Soc.: Des. Conf., vol. 1, pp. 757–766, 2020.https://doi.org/10.1017/dsd.2020.15.Search in Google Scholar

[23]M. Grieves and J. Vickers, “Digital twin: mitigating unpredictable, undesirable emergent behavior in complex systems,” inTransdisciplinary Perspectives on Complex Systems, F.-J. Kahlen, S. Flumerfelt, and A. Alves, Eds., Cham, Springer International Publishing, 2017, pp. 85–113.10.1007/978-3-319-38756-7_4Search in Google Scholar

[24]R. Soni and M. Bhatia, “others, Digital twin: intersection of mind and machine,”Int. J. Comput. Intell. IoT, vol. 2, no. 3, 2019.Search in Google Scholar

[25]A. Borrmann, J. Schlenger, N. Bus, and R. Sacks, “AEC digital twin data‐why structure matters,” in19th ICCCBE, University of Cape Town, 2022, pp. 26–28.Search in Google Scholar

[26]ISO International Organization for Standardization,ISO/TS 15143-3:2020 Earth-Moving Machinery and Mobile Road Construction Machinery — Worksite Data Exchange: Part 3: Telematics Data, ISO International Organization for Standardization, 2020.Search in Google Scholar

[27]F. Wedel, D. Opitz, C. Tiedemann, and M. Meyer‐Westphal, “Das 3‐D‐Modell als Grundlage des digitalen Zwillings,”Bautechnik, vol. 99, pp. 104–113, 2022.https://doi.org/10.1002/bate.202100092.Search in Google Scholar

[28]M. Botz, A. Emiroglu, K. Osterminski, M. Raith, R. Wüchner, and C. Große, “Überwachung und Modellierung der Tragstruktur von Windenergieanlagen,”Beton‐ und Stahlbetonbau, vol. 115, pp. 342–354, 2020.https://doi.org/10.1002/best.202000001.Search in Google Scholar

[29]S. Vilgertshofer, et al.., “TwinGen: advanced technologies to automatically generate digital twins for operation and maintenance of existing bridges,” inECPPM 2022 – eWork and eBusiness in Architecture, Engineering and Construction 2022, R. J. Scherer, S. F. Sujan, and E. Hjelseth, Eds., London, CRC Press, 2023, pp. 213–220.10.1201/9781003354222-27Search in Google Scholar

[30]M. Mehranfar, A. Braun, and A. Borrmann, “A hybrid top-down, bottom-up approach for 3D space parsing using dense RGB point clouds,” inECPPM 2022 – eWork and eBusiness in Architecture, Engineering and Construction 2022, R. J. Scherer, S. F. Sujan, and E. Hjelseth, Eds., London, CRC Press, 2023, pp. 551–558.10.1201/9781003354222-70Search in Google Scholar

[31]M. S. Mafipour, S. Vilgertshofer, and A. Borrmann, “Digital twinning of bridges from point cloud data by deep learning and parametric models,” inECPPM 2022 – eWork and eBusiness in Architecture, Engineering and Construction 2022, R. J. Scherer, S. F. Sujan, and E. Hjelseth, Eds., London, CRC Press, 2023, pp. 543–550.10.1201/9781003354222-69Search in Google Scholar

[32]Y. Pan, A. Braun, A. Borrmann, and I. Brilakis, “3D deep-learning-enhanced void-growing approach in creating geometric digital twins of buildings,”Proc. Inst. Civ. Eng.: Smart Infrastruct. Const., vol. 176, no. 1, pp. 24–40, 2023.https://doi.org/10.1680/jsmic.21.00035.Search in Google Scholar

[33]Y. Pan, A. Braun, I. Brilakis, and A. Borrmann, “Enriching geometric digital twins of buildings with small objects by fusing laser scanning and AI-based image recognition,”Autom. Constr., vol. 140, 2022, Art. no. 104375.https://doi.org/10.1016/j.autcon.2022.104375.Search in Google Scholar

[34]T. Oberbichler, A. M. Bauer, A.-K. Goldbach, R. Wüchner, and K.-U. Bletzinger, “CAD‐integrierte Analyse im Entwurfsprozess,”Bautechnik, vol. 96, pp. 400–408, 2019.https://doi.org/10.1002/bate.201800105.Search in Google Scholar

[35]F. Pfitzner, A. Braun, and A. Borrmann, “Object detection based knowledge graph creation: enabling insight into construction processes,” inProc. of the International Conference on Computing in Civil Engineering 2023, 2023.10.1061/9780784485224.023Search in Google Scholar

[36]F. Pfitzner, A. Braun, and A. Borrmann, “Towards data mining on construction sites: heterogeneous data acquisition and fusion,” inProc. of ECPPM 2022, 2022.Search in Google Scholar

[37]A. Borrmann, M. König, C. Koch, and J. Beetz,Building Information Modeling, Wiesbaden, Springer Fachmedien Wiesbaden, 2021.10.1007/978-3-658-33361-4Search in Google Scholar

[38]Bundesministerium für Digitales und Verkehr,Digitaler Zwilling von Brücken, Beitrag zum Masterplan Digitaler Zwilling Bundesfernstraßen, 2023.Search in Google Scholar

[39]C. Boje, A. Guerriero, S. Kubicki, and Y. Rezgui, “Towards a semantic construction digital twin: directions for future research,”Autom. Constr., vol. 114, 2020, Art. no. 103179.https://doi.org/10.1016/j.autcon.2020.103179.Search in Google Scholar

[40]R. Sacks, I. Brilakis, E. Pikas, H. S. Xie, and M. Girolami, “Construction with digital twin information systems,”DCE 1, vol. 1, 2020, Art. no. e14.https://doi.org/10.1017/dce.2020.16.Search in Google Scholar

[41]J. Schlenger, F. Pfitzner, A. Braun, S. Vilgertshofer, and A. Borrmann, “Digital twin construction site – construction site monitoring for automated time and cost control (orig. Digitaler Zwilling Baustelle – Baustellenüberwachung zur automatisierten Zeit‐ und Kostenkontrolle),”Bautechnik, vol. 100, no. 4, pp. 190–197, 2023.https://doi.org/10.1002/bate.202300005.Search in Google Scholar

[42]A. Braun, S. Tuttas, A. Borrmann, and U. Stilla, “Improving progress monitoring by fusing point clouds, semantic data and computer vision,”Autom. Constr., vol. 116, 2020, Art. no. 103210.https://doi.org/10.1016/j.autcon.2020.103210.Search in Google Scholar

[43]M. Slepicka, S. Vilgertshofer, and A. Borrmann, “Fabrication information modeling: interfacing building information modeling with digital fabrication,”Const. Robot., vol. 6, pp. 87–99, 2022.https://doi.org/10.1007/s41693-022-00075-2.Search in Google Scholar

[44]B. Vogel-Heuser, F. Ocker, I. Weiß, R. Mieth, and F. Mann, “Potential for combining semantics and data analysis in the context of digital twins,”Philos. Trans. R. Soc., A, vol. 379, 2021, Art. no. 20200368.https://doi.org/10.1098/rsta.2020.0368.Search in Google Scholar PubMed PubMed Central

[45]B. Vogel-Heuser, F. Ocker, and T. Scheuer, “An approach for leveraging Digital Twins in agent-based production systems,”at – Automatisierungstechnik, vol. 69, no. 12, pp. 1026–1039, 2021.https://doi.org/10.1515/auto-2021-0081.Search in Google Scholar

[46]J. Zhao, et al.., “A semi-automatic approach for asset administration Shell creation from heterogeneous data,” in22nd IFAC World Congress, Yokohama, Japan, 2023, p. 7.Search in Google Scholar

[47]F. Ocker, B. Vogel-Heuser, H. Schon, and R. Mieth, “Leveraging digital twins for compatibility checks in production systems engineering,” in2021 IEEM, Singapore, IEEE, 2021, pp. 103–107.10.1109/IEEM50564.2021.9672892Search in Google Scholar

[48]B. Vogel-Heuser and D. Hess, “Guest editorial industry 4.0–prerequisites and visions,”IEEE Trans. Autom. Sci. Eng., vol. 13, pp. 411–413, 2016.https://doi.org/10.1109/TASE.2016.2523639.Search in Google Scholar

[49]R. N. Bolton, et al.., “Customer experience challenges: bringing together digital, physical and social realms,”JOSM, vol. 29, no. 5, pp. 776–808, 2018.https://doi.org/10.1108/JOSM-04-2018-0113.Search in Google Scholar

[50]E. Glaessgen and D. Stargel, “The digital twin paradigm for future NASA and U.S. Air force vehicles,” in53rd Structures, Structural Dynamics and Materials Conference<BR>20th Adaptive Structures Conference, Honolulu, Hawaii, Reston, Virginia, American Institute of Aeronautics and Astronautics, 2012.10.2514/6.2012-1818Search in Google Scholar

[51]TUM – MIRMI,KI.FABRIK, 2023. Available at:https://kifabrik.mirmi.tum.de/team/ Accessed: Jul. 24, 2023.Search in Google Scholar

[52]J. Gausemeier and S. Moehringer, “VDI 2206- A new guideline for the design of mechatronic systems,”IFAC Proc. Vol., vol. 35, no. 2, pp. 785–790, 2002.https://doi.org/10.1016/S1474-6670(17)34035-1.Search in Google Scholar

[53]Plattform Industrie 4.0, “Was ist die Plattform Industrie 4.0,” 2023. Available at:https://www.plattform-i40.de/IP/Navigation/DE/Home/home.html.Search in Google Scholar

[54]L. Rudolph, et al.., “Maintenance in process industries with digital twins and mixed reality: potentials, scenarios and requirements,” in2022 IEEM, Kuala Lumpur, Malaysia, IEEE, 2022, pp. 474–481.10.1109/IEEM55944.2022.9989826Search in Google Scholar

[55]Unity, “Unity real-time development platform,” 2023. Available at:https://unity.com/ Accessed: Jul. 12, 2023.Search in Google Scholar

[56]Omniverse,NVIDIA Omniverse, 2023. Available at:https://www.nvidia.com/en-us/omniverse/ Accessed: Jul. 12, 2023.Search in Google Scholar

[57]machineering, “Virtual commissioning VIBN with iPhysics (orig. Virtuelle Inbetriebnahme VIBN mit iPhysics),” 2023. Available at:https://www.machineering.com/en/ Accessed: Jul. 12, 2023.Search in Google Scholar

[58]J. Höfgen, et al.., “Architecture of a versatile digital twin with socket-based communication and azure DT,” in19th IEEE CASE, Cordis, New Zealand, 2023, p. 8.10.1109/CASE56687.2023.10260340Search in Google Scholar

[59]A. Kantaros, D. Piromalis, G. Tsaramirsis, P. Papageorgas, and H. Tamimi, “3D printing and implementation of digital twins: current trends and limitations,”ASI, vol. 5, no. 1, p. 7, 2022.https://doi.org/10.3390/asi5010007.Search in Google Scholar

[60]L. Zhang, et al.., “Digital twins for additive manufacturing: a state-of-the-art review,”Appl. Sci., vol. 10, no. 23, p. 8350, 2020.https://doi.org/10.3390/app10238350.Search in Google Scholar

[61]T. DebRoy, W. Zhang, J. Turner, and S. S. Babu, “Building digital twins of 3D printing machines,”Scr. Mater., vol. 135, pp. 119–124, 2017.https://doi.org/10.1016/j.scriptamat.2016.12.005.Search in Google Scholar

[62]K. Bartsch, A. Pettke, A. Hübert, J. Lakämper, and F. Lange, “On the digital twin application and the role of artificial intelligence in additive manufacturing: a systematic review,”J. Phys.: Mater., vol. 4, no. 3, 2021, Art. no. 32005.https://doi.org/10.1088/2515-7639/abf3cf.Search in Google Scholar

[63]D. R. Gunasegaram, et al.., “Towards developing multiscale-multiphysics models and their surrogates for digital twins of metal additive manufacturing,”Addit. Manuf., vol. 46, 2021, Art. no. 102089.https://doi.org/10.1016/j.addma.2021.102089.Search in Google Scholar

[64]Z. Yang, “Model-based Predictive Analytics for Additive and Smart Manufacturing,” Ph.D. dissertation, ScholarWorks@UMass Amherst, University of Massachusetts Amherst, 2018.Search in Google Scholar

[65]A. Majeed, et al.., “A big data-driven framework for sustainable and smart additive manufacturing,”Robot. Comput.-Integr. Manuf., vol. 67, 2021, Art. no. 102026.https://doi.org/10.1016/j.rcim.2020.102026.Search in Google Scholar

[66]C. Liu, L. Le Roux, C. Körner, O. Tabaste, F. Lacan, and S. Bigot, “Digital twin-enabled collaborative data management for metal additive manufacturing systems,”J. Manuf. Syst., vol. 62, pp. 857–874, 2022.https://doi.org/10.1016/j.jmsy.2020.05.010.Search in Google Scholar

[67]T. W. Heo, et al.., “A mesoscopic digital twin that bridges length and time scales for control of additively manufactured metal microstructures,”J. Phys.: Mater., vol. 4, no. 3, 2021, Art. no. 34012.https://doi.org/10.1088/2515-7639/abeef8.Search in Google Scholar

[68]D. R. Gunasegaram, A. B. Murphy, M. J. Matthews, and T. DebRoy, “The case for digital twins in metal additive manufacturing,”J. Phys.: Mater., vol. 4, 2021, Art. no. 40401,https://doi.org/10.1088/2515-7639/ac09fb.Search in Google Scholar

[69]A. Kwade, W. Haselrieder, R. Leithoff, A. Modlinger, F. Dietrich, and K. Droeder, “Current status and challenges for automotive battery production technologies,”Nat. Energy, vol. 3, pp. 290–300, 2018.https://doi.org/10.1038/s41560-018-0130-3.Search in Google Scholar

[70]T. Günther, N. Billot, J. Schuster, J. Schnell, F. B. Spingler, and H. A. Gasteiger, “The manufacturing of electrodes: key process for the future success of lithium-ion batteries,”AMR, vol. 1140, pp. 304–311, 2016.https://doi.org/10.4028/www.scientific.net/AMR.1140.304.Search in Google Scholar

[71]E. Ayerbe, M. Berecibar, S. Clark, A. A. Franco, and J. Ruhland, “Digitalization of battery manufacturing: current status, challenges, and opportunities,”Adv. Energy Mater., vol. 12, 2022, Art. no. 2102696.https://doi.org/10.1002/aenm.202102696.Search in Google Scholar

[72]F. M. Zanotto, et al.., “Data specifications for battery manufacturing digitalization: current status, challenges, and opportunities,”Batter. Supercaps, vol. 5, no. 9, 2022.https://doi.org/10.1002/batt.202200224.Search in Google Scholar

[73]J. Krauß, et al.., “Digital twins in battery cell production,” inProduction at the Leading Edge of Technology, M. Liewald, A. Verl, T. Bauernhansl, and H.-C. Möhring, Eds., Cham, Springer International Publishing, 2023, pp. 823–832.Search in Google Scholar

[74]A. Kampker, et al.., “Concept for digital product twins in battery cell production,”WEVJ, vol. 14, no. 4, p. 108, 2023.https://doi.org/10.3390/wevj14040108.Search in Google Scholar

[75]A. Sommer, M. Leeb, L. Weishaeupl, and R. Daub, “Integration of electrode markings into the manufacturing process of lithium-ion battery cells for tracking and tracing applications,”Batteries, vol. 9, no. 2, p. 89, 2023.https://doi.org/10.3390/batteries9020089.Search in Google Scholar

[76]A. Mayr, D. Schreiner, B. Stumper, and R. Daub, “In-line sensor-based process control of the calendering process for lithium-ion batteries,”Procedia CIRP, vol. 107, pp. 295–301, 2022.https://doi.org/10.1016/j.procir.2022.04.048.Search in Google Scholar

[77]S. Henschel, S. Otte, D. Mayer, and J. Fleischer, “Use cases for digital twins in battery cell manufacturing,” inProduction at the Leading Edge of Technology, M. Liewald, A. Verl, T. Bauernhansl, and H.-C. Möhring, Eds., Cham, Springer International Publishing, 2023, pp. 833–842.10.1007/978-3-031-18318-8_82Search in Google Scholar

[78]A. D. Kies, J. Krauß, A. Schmetz, R. H. Schmitt, and C. Brecher, “Interaction of digital twins in a sustainable battery cell production,”Procedia CIRP, vol. 107, pp. 1216–1220, 2022.https://doi.org/10.1016/j.procir.2022.05.134.Search in Google Scholar

[79]G. Ventura Silva, et al.., “Others, digitalization platform for sustainable battery cell production: coupling of process, production, and product models,”Energy Technol., vol. 11, no. 5, 2022, Art. no. 2200801.https://doi.org/10.1002/ente.202200801.Search in Google Scholar

[80]S. Yang, R. He, Z. Zhang, Y. Cao, X. Gao, and X. Liu, “CHAIN: cyber hierarchy and interactional network enabling digital solution for battery full-lifespan management,”Matter, vol. 3, pp. 27–41, 2020.https://doi.org/10.1016/j.matt.2020.04.015.Search in Google Scholar

[81]D. Jones, C. Snider, A. Nassehi, J. Yon, and B. Hicks, “Characterising the Digital Twin: a systematic literature review,”CIRP J. Manuf. Sci. Technol., vol. 29, pp. 36–52, 2020.https://doi.org/10.1016/j.cirpj.2020.02.002.Search in Google Scholar

[82]J. Trauer, S. Pfingstl, M. Finsterer, and M. Zimmermann, “Improving production efficiency with a digital twin based on anomaly detection,”Sustainability, vol. 13, no. 18, 2021, Art. no. 10155.https://doi.org/10.3390/su131810155.Search in Google Scholar

[83]J. Trauer, M. Mutschler, M. Mörtl, and M. Zimmermann, “Challenges in implementing digital twins – a survey,” inProceedings of 2022 IDETC-CIE, St. Louis, Missouri, USA, New York, N.Y., the American Society of Mechanical Engineers, 2022.10.1115/DETC2022-88786Search in Google Scholar

[84]G. E. Paul, “Modeling and simulation of human systems,” inHandbook of Human Factors and Ergonomics, 2021.10.1002/9781119636113.ch27Search in Google Scholar

[85]M. Spitzhirn, S. Ullman, S. Bauer, and L. Fritzsche, “Digital production planning and human simulation of manual and hybrid work processes using the ema Software Suite,” inProceedings of the 7th International Digital Human Modeling Symposium, 2022.10.17077/dhm.31740Search in Google Scholar

[86]M. Uhl, et al.., “Research approach for predicting body postures and musculoskeletal stress due to disruptive design changes on power tools,” inHuman Interaction, Emerging Technologies and Future Systems V, Cham, Springer International Publishing, 2022, pp. 462–467.10.1007/978-3-030-85540-6_59Search in Google Scholar

[87]C. M. Harbauer, M. Fleischer, T. Nguyen, F. Bos, and K. Bengler, “Too close to comfort? A new approach of designing a soft cable-driven exoskeleton for lifting tasks under ergonomic aspects,” in2020 3rd International Conference on Intelligent Robotic and Control Engineering (IRCE), 2020, pp. 105–109.10.1109/IRCE50905.2020.9199238Search in Google Scholar

[88]D. D. Salvucci, “Rapid prototyping and evaluation of in-vehicle interfaces,”ACM Trans. Comput. Hum. Interact., vol. 16, pp. 1–33, 2009,https://doi.org/10.1145/1534903.1534906.Search in Google Scholar

[89]M. S. L. Scharfe-Scherf, S. Wiese, and N. Russwinkel, “A cognitive model to anticipate variations of situation awareness and attention for the takeover in highly automated driving,”Information, vol. 13, no. 9, p. 418, 2022.https://doi.org/10.3390/info13090418.Search in Google Scholar

[90]C. Fan, C. Zhang, A. Yahja, and M. Ali, “Disaster City Digital Twin: a vision for integrating artificial and human intelligence for disaster management,”Int. J. Inf. Manag., vol. 56, 2021, Art. no. 102049.https://doi.org/10.1016/j.ijinfomgt.2019.102049.Search in Google Scholar

[91]M. Spitzhirn, S. Ullmann, and L. Fritzsche, “Considering individual abilities and age-related changes in digital production planning—human-centered design of industrial work tasks with ema softwareIntegration individueller und altersspezifischer Eigenschaften des Menschen in die digitale Produktionsplanung – Gestaltung fähigkeitsgerechter Arbeitsprozesse in der Industrie mittels ema Work Designer,”Z. Arb. Wiss., vol. 76, pp. 459–477, 2022.https://doi.org/10.1007/s41449-022-00343-5.Search in Google Scholar

[92]T. I. Buldakova and S. I. Suyatinov, “Hierarchy of human operator models for digital twin,” in2019 International Russian Automation Conference (RusAutoCon), 2019, pp. 1–5.10.1109/RUSAUTOCON.2019.8867602Search in Google Scholar

[93]S. Prezenski, A. Brechmann, S. Wolff, and N. Russwinkel, “A cognitive modeling approach to strategy formation in dynamic decision making,”Front. Psychol., vol. 8, 2017, Art. no. 266988,https://doi.org/10.3389/fpsyg.2017.01335.Search in Google Scholar PubMed PubMed Central

[94]B. R. Barricelli, E. Casiraghi, J. Gliozzo, A. Petrini, and S. Valtolina, “Human digital twin for fitness management,”IEEE Access, vol. 8, pp. 26637–26664, 2020.https://doi.org/10.1109/ACCESS.2020.2971576.Search in Google Scholar

[95]M. E. Miller and E. Spatz, “A unified view of a human digital twin,” inHuman-Intelligent Systems Integration, vol. 4, 2022, pp. 23–33.10.1007/s42454-022-00041-xSearch in Google Scholar

[96]Z. Yao, et al.., “A novel approach to simulating realistic exoskeleton behavior in response to human motion,”Robotics, vol. 13, no. 2, p. 27, 2024.https://doi.org/10.3390/robotics13020027.Search in Google Scholar

[97]C. Scheuermann, T. Binderberger, N. von Frankenberg, and A. Werner,Digital Twin -A Machine Learning Approach to Predict Individual Stress Levels in Extreme Environments, New York, Association for Computing Machinery (ACM), 2020.10.1145/3410530.3414316Search in Google Scholar

Received:2023-12-05

Accepted:2024-06-18

Published Online:2024-09-10

Published in Print:2024-09-25

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

DOI: doi.org/10.1515/auto-2023-0227

Keywords for this article

digital twin;production;construction;production and construction engineering;human machine interaction;Industry 4.0

Creative Commons

BY 4.0

Movatterモバイル変換

Long living human-machine systems in construction and production enabled by digital twins

Article

Abstract

Zusammenfassung

1 Introduction and important definitions

1.1 What is a digital twin?

1.2 Digital twin in engineering

2 Aim and categories of digital twins in engineering

3 Digital twins in the construction, building and production engineering domain

3.1 Built environment

3.1.1 Aims

3.1.2 Components and interfaces

3.1.3 Inputs

3.1.4 Outputs

3.1.5 Scale

3.1.6 Challenges

3.1.7 Outlook

3.2 Construction logistics

3.2.1 Aims

3.2.2 Components and interfaces

3.2.3 Inputs

3.2.4 Outputs

3.2.5 Scale

3.2.6 Challenges

3.2.7 Outlook

3.3 Material flow systems

3.3.1 Aims

3.3.2 Components and interfaces

3.3.3 Inputs

3.3.4 Outputs

3.3.5 Scale

3.3.6 Challenges

3.3.7 Outlook

3.4 Machine and plant manufacturing industries

3.4.1 Aims

3.4.2 Components and interfaces

3.4.3 Inputs

3.4.4 Outputs

3.4.5 Scale

3.4.6 Challenges

3.4.7 Outlook

3.5 Manufacturing process by using the example of additive manufacturing

3.5.1 Aims

3.5.2 Components and interfaces

3.5.3 Inputs

3.5.4 Outputs

3.5.5 Scale

3.5.6 Challenges

3.5.7 Outlook

3.6 Production process by using the example of battery production

3.6.1 Aims

3.6.2 Components and interfaces

3.6.3 Inputs

3.6.4 Outputs

3.6.5 Scale

3.6.6 Challenges

3.6.7 Outlook

3.7 Technical product development

3.7.1 Aims

3.7.2 Components and interfaces

3.7.3 Inputs

3.7.4 Outputs

3.7.5 Scale

3.7.6 Challenges

3.7.7 Outlook

3.8 Humans in the loop

3.8.1 Aims

3.8.2 Components and interfaces

3.8.3 Inputs

3.8.4 Outputs

3.8.5 Scale

3.8.6 Challenges

3.8.7 Outlook

4 Conclusion and future work

About the authors

Acknowledgment

References

Articles in the same Issue

Articles in the same Issue