Digital Twins in Engineering and Design PDF

! Hierarchy of Digital Twins Automation and Information Systems TU München Greenfield plants require an integrated construction and production engineering  Multifaceted utilization of digital twins across fields in engineering Construction Built environment Construction logistics Material flow syst...

! Hierarchy of Digital Twins Automation and Information Systems TU München Greenfield plants require an integrated construction and production engineering  Multifaceted utilization of digital twins across fields in engineering Construction Built environment Construction logistics Material flow systems Built Environment and construction Logistics Factory Digital Twin Construction Production Manufacturing Intralogistics (Material Flow System) Human factors Material Flow System ! Production Technical Product Development AIS Machine & Plant Engineering, Manufacturing & Production Process Human © AIS ! Consider digital twins in different domains  Improve operational efficiency, optimize processes, and establish seamless integration between the physical and virtual realms  Different engineering domains developed diverse aims and types of digital twins tailored to meet their requirements and needs Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 12 Industrial Revolution Automation and Information Systems TU München Industry 5.0 in the timeline of industrial development (expanded from [WahI17]) First industrial revolution Second industrial revolution Third industrial revolution Fourth industrial revolution Fifth industrial revolution Introduction of water- and steam-driven mechanical production equipment Introduction of electrically driven mass production based on division of labor Further automation of production through electronics and IT systems through the development of cyber-physical systems Focuses on humancentered production and resource efficiency First programmable logic controller (PLC) First production lines Modicon 084 1969 at the Cincinnati slaughterhouse in 1870 First mechanical loom 1784 | | 1800 | | | | | | | | | 1900 | | | | | | | | | 2000 | | | | | 2023 - Today Vogel-Heuser, Birgit, and Klaus Bengler. "Von Industrie 4.0 zu Industrie 5.0–Idee, Konzept und Wahrnehmung." HMD Praxis der Wirtschaftsinformatik (2023): 1-19. Wahlster, Wolfgang. "Industrie 4.0: Das Internet der Dinge kommt in die Fabriken." Zukunft Industrie., Deutscher Forschungszentrum für Künstliche Intelligenz (DFKI)., IHK Darmstadt 22 (2017). Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 | © AIS | Time and complexity 15 Automation and Information Systems TU München Definition of Digital Twin in the aPS domain From the industry's perspective “A digital twin refers to a virtual representation of a physical asset or system throughout its lifecycle” [Mik18] From a research perspective “A dynamic virtual representation of a physical object or system across its lifecycle, using real-time data to enable understanding, learning and reasoning” [Bol18] [NIC17] Common aspects Virtual representation of a physical asset Represents the complete life cycle Development of the definition From simulation to product lifecycle management to Industry 4.0 incl. operational data © AIS Heterogeneous definitions  Definitions depend on domain Mikell, Matthew, and Jen Clark. "Cheat sheet: what is Digital Twin." IBM Internet of things blog. See https://www. ibm. com/blogs/internet-of-things/iot-cheat-sheet-digital-twin (2018). Bolton, Ruth N., et al. "Customer experience challenges: bringing together digital, physical and social realms." Journal of service management 29.5 (2018): 776-808. National Infrastructure Commission. "Data for the public good NATIONAL INFRASTRUCTURE COMMISSION." (2017). Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 19 Automation and Information Systems TU München Digital model   a digital version of a pre-existing or planned physical object no automatic data exchange between the physical model and digital model Digital shadow  digital representation of an object that has a one-way flow between the physical and digital object Digital twin  data flows between an existing physical object and a digital object are fully integrated in both directions © AIS Digital Twin Misconceptions Fuller, Aidan, et al. "Digital twin: Enabling technologies, challenges and open research." IEEE access 8 (2020): 108952-108971. Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 20 Definition of Digital Twin in Engineering Automation and Information Systems TU München A cross-domain applicable definition: Digital Twins are defined as dynamic digital representations of specific real-world entities consisting of (interlinked) components and interfaces with application-specific attributes and scales (e.g., time, size, accuracy, hierarchy, life cycle phase). Digital twins have the goal of recurrent improvement in the real world. Key aspects  What is a digital twin?  Relation between real-world and digital world  What is a digital twin made of?  Interlinked components and interfaces  Application-specific attributes and scales  What is the goal of a digital twin?  recurrent improvement Material Flow System (Intralogistics) Technical Product Development Construction Machine & Plant Engineering, Manufacturing & Production Process Human Production © AIS Built Environment and Construction Logistics https://mediatum.ub.tum.de/doc/1716587/w5qit6s8dynv8sozeof2vzio3.pdf Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 21 Motivation: Digital Twins in automated Production Systems Automation and Information Systems TU München Motivation for the use of Digital Twins - Excerpt: Holistic view of the product life cycle Comprehensive monitoring and control for complex production systems Fast and flexible fault recovery and maintenance in production systems Predict and optimize the behavior of production systems Testbed for new processes and technologies to identify and resolve potential problems Software update (approx.1.5 years) Upgrade / Retrofit Operation / Maintenance Startup Acceptance test Software implementation Electrotechnical and software integration Commissioning Software design Cabinet construction Mechanical assembly System delivery Electrotechnical design Manufacturing Integrated test Mechanical design System integration Detailed design Component Test System design System specification Requirement specification Basic design Up to 50 years © AIS Electrotechnical retrofit (approx.7-20 years) Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 22 VDI/VDE - 2193 "I4.0 Language"- Components Design principles and rules for dialogs Dialog sequences Natural language Grammar of sentences I4.0 language Sentences formed from words Dictionary of words Single words Interaction protocols VDI 2193-2: Bidding process Structure of messages VDI 2193-1: Concept of an I4.0-Language and message structure Vocabulary of I4.0 language E.g., eCl@ss Machine-executable and semantically unambiguous abstraction of the real world Features Vocabulary Attribute ID Version Name Descri ption Symbol SI Unit Data type Value Value range Value 0173-1#02BAB576#005 V9.1 Nominal voltage - U V real 240 0..240 ID: 0173-1#02-BAB576#005 Value: 240 Sender Message: ID + Value Receiver Nennspannung 240V, 50Hz 電压 Tension nominale 240V 240V Nominal voltage 240 Betriebsspannung 240V LNI, Kompetenzzentrum 4.0 Hannover, Hannover Messe, ifak, VWS vernetzt 06.04.2020 Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 30 © AIS Automation and Information Systems TU München Automation and Information Systems TU München Digital Twin in Industry 4.0 – Asset Administration Shell Asset Administration Shell (AAS) Standardized digital representation of a plant and process  Forms I4.0 component together with a physical object (asset) and the administration shell Provides interface for I4.0 communication  Addresses issues such as access security, visibility, identity and permission management, confidentiality and integrity  Manufacturer-independent, industry-neutral standard I4.0 component Administration Shell Access to data and functionalities Asset Industrie 4.0 Plattform https://www.basys40.de/ Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 © AIS What types of data should be included in a digital twin model based on AAS? 35 UC: Database structure Automation and Information Systems TU München Engineeringphase Operational phase Data of the xPPU https://github.com/x-PPU Implementations based on state diagrams in PLC UML PLCOpenXML as exchange format between IEC development platforms PLC implementations in classic IEC 61131-3 languages AutomationML for PPR as Industrie 4.0 enabler for xPPU Technical documentation such SysML models (one model for each as drawings and circuit diagrams scenario) in behavior and structure Heterogeneous information is mapped in the digital twin Digital twin contains engineering data and operational data Documents: Technical documentation, IEC61131-3 code, SysML models, etc.  Difficult to handle in practice Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 © AIS Runtime signal data analysis for detection of anomalies in behavior 40 Digital Twin (Data/Information)  Ontologies  Semantic Digital Twin Automation and Information Systems TU München Digital Twins.XML virtual representation of a physical object focus on real-time monitoring, simulation and optimization Standardization Increased interoperability AAS.AASX standardized digital image of the plant focus on standardized information exchange and interoperability between objects Semantic Digital Twin Ontologies (knowledge) Enables automated usage of information e.g., for inconsistency management including compatibility checks.OWL Export AAS from DT in XML  Map XML to OWL Checks with SPARQL crane Properties Location Schedule.AASX.XML TBox Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 minRange maxRange ABox.OWL 47 © AIS AAS Relation between Digital Twins and Ontologies Automation and Information Systems TU München Ontology: Formal representation of concepts and relationships within a domain. Digital Twin: A digital image of a physical object throughout its life cycle Physical system Digital twin Semantic Digital Twin: Digital twin with a formal representation e.g., „semantic AAS“: Transform XML-representation of an AAS based Metamodel in RDF/OWL Benefits: Enables communication between machines and ensures compatibility with different digital frameworks and architectures...... Digital twin Service Descr.... OWL Queries Rules RIF RDFS SPARQL Semantic Web Technologies SQWRL Result Entailment RDF Update Constructors Math SWRL Builtins Filter Query... XML Unicode... Logical Conditionals Maths URI Ocker, Felix, et al. "Leveraging Digital Twins for Compatibility Checks in Production Systems Engineering." 2021 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM). IEEE, 2021. Beden, Sadeer, Qiushi Cao, and Arnold Beckmann. "Semantic asset administration shells in industry 4.0: A survey." 2021 4th IEEE International Conference on Industrial Cyber-Physical Systems (ICPS). IEEE, 2021. Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 Date/ Time String Comparison © AIS Physical system 48 Automation and Information Systems TU München ! Summary and Outlook Definition of digital twin Digital Twins are defined as dynamic digital representations of specific real-world entities consisting of (interlinked) components and interfaces with application-specific attributes and scales (e.g., time, size, accuracy, hierarchy, life cycle phase). Digital twins have the goal of recurrent improvement in the real world. Digital twin aggregate Engineering data: e.g. SysML and CAD-models Operational data: e.g. sensor values Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 © AIS Currently various, partly parallel, standardization efforts with industry support (e.g. AAS) but no complete consensus yet 58 Automation and Information Systems TU München Challenges and Outlook of Digital Twin Objective Provide accurate asset visualization Enhance asset management and control Support predictive maintenance Enable real-time process adjustments Challenges heterogeneous data exchange formats, inconsistent/incomplete information sources Evolution of DT to meet changing needs Lack of machine and plant DTs, requiring component suppliers to integrate them into various tools © AIS Outlook Emphasize real-time aspects to ensure production safety and stability Enhance the bidirectional flow of information between virtual and real-world assets Prof. Dr.-Ing. Birgit Vogel-Heuser | Introduction Digital Twin in Engineering and Design | 18.10.2023 59 Different Manufacturing Processes after DIN EN ISO/ASTM 52900 Subtractive shaping: The desired shape is acquired by selective removal of material. Examples: Milling, turning, drilling, etc. Formative shaping: The desired shape is acquired by application of pressure to a body of raw material. Examples: forging, bending, casting, injection moulding,etc. Additive shaping: The desired shape is acquired by successive addition of material. Quelle: www.techpilot.de Quelle: www.ingenieurkurse.de Quelle: https://additive.industrie.de/ In Additive Manufacturing, components are built up layer by layer and are not created by removing or reforming material as in conventional processes Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 6 When to use AM? Customization Complexity Additive Manufacturing Component size Costs per part Costs per part Costs per part Complexity Number of parts Component size Conventional manufacturing AM's cost advantages lie in the production of complex, individualized and small components as well as in small series production. Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 10 Influencing factors of the process strategy on the end product Material Type (Polymer, Metal, Ceramic, …) 𝜎 𝜀 Material State (liquid, powder, suspension, filament…) Product Machine architecture Energy input strategy Source: Siemens h w Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 12 Challenges in Additive Manufacturing Reproducibility Inconsistent process results due to unstable process behavior Quality assurance Insufficient process understanding requires 100% inspection rates Monitoring Very little available monitoring solutions inhibit a process and part evaluation during the build. DT can facilitate control Unavailability of control strategies to decrease strategies DT can provide in-line Complex quality non-destructive testing and testing assurance. DT can facilitate Long build times (days) to process monitoring support timely operator interference. Inefficient quality assurance High temporal and financial invest prior to known part properties waste. High reject and waste rates Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 25 State of the art overview No fully fledged digital twin for additive manufacturing is available in the industry There is no published example of a complete digital twin for additive manufacturing in research Individual components and interfaces are available or researched: Condition monitoring Prior experience from other manufacturing systems and predictive maintenance approaches (e.g. milling, casting) Monitoring of machine status Monitoring of process atmosphere Process monitoring Existing monitoring approaches for various individual process properties (specific to machine and process) Simulations Existing simulations for individual process aspects (specific to machine and process) Example 1: Melt pool monitoring Example 2: Powder bed monitoring Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing Example 1: PBF-LB/M multi-physics model Example 2: DED temperature gradients 27 Overview Multitude of processes Lack of Standardization Complexity of the process [Khairallah 2021] [3DCAD World] Flow3D AM] Various processes with different conditions and requirements Various machine manufacturer with different interfaces and possibility of control Complex process trough various influencing factors and their interaction Complex DT development for a complex process, which must be tailored to process and machine Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 38 Overview Multitude of processes Lack of Standardization Complexity of the process [Khairallah 2021] [3DCAD World] Flow3D AM] Various processes with different conditions and requirements Various machine manufacturer with different interfaces and possibility of control Complex process trough various influencing factors and their interaction Complex DT development for a complex process, which must be tailored to process and machine Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 42 Real World Entity Inputs Machine data Inherently existing in machine and necessary for machine workings Actor control data e.g. position data, laser power, movement speed, wire transport rate Specialized Sensor data Additional sensory specialized to monitor the process and its quality, e.g. thermal imagery, powder bed monitoring, geometric measurement Condition data Data regarding the process condition and context Process condition, e.g. humidity, material data Internal sensor data, e.g. oxygen content, temperature of heating elements or build space Meta data for data structure and assignment, e.g. product serial number, date & time, manufacturing system Digital Twin Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 45 Components & Interfaces Data processing Processing and fusion of inputs to evaluate process for quality assessment Simulations Input-based simulations for process behavior prediction and support decision making Digital Twin Controller Control loops for real-time control/ layer-wise control Big data archive Historical and statistical data archive to support data evaluation Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing Visualizations Visualization of processed data to allow operator supervision [Sigma additive solutions] Many more … Applications specific number and nature of DT components for most efficient task fulfillment … 46 Digital Twin Outputs Control values Control loop outputs as actor target values, e.g. laser power, motor current, welding amp Process/ Part Quality Process/part evaluation result quality assurance and iterative parameter adjustment Part good Short-term process stability improvement trough process control Process documentation DT recordings as process documentation for archiving purposes and statistical and scientific evaluation Part bad Medium–term process improvement trough detailed process quality assurance Long-term process improvement trough profound process understanding Real World Entity Prof. Dr.-Ing. Katrin Wudy (TUM) | Professorship of Laser-based Additive Manufacturing 47 By combining the real and the digital worlds, Siemens empowers its customers to master their digital transformation and sustainability challenges Automotive Pharma Campus Page 4 Food and Beverage Machine Building Chemicals Additive Manufacturing Digital worlds Hospitals Energy Distribution Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Cranes Real worlds Rail Airports Process Plants Logistics Buildings Oil and Gas Thecombining By comprehensive the real Digital and the Twin digital is at the worlds, heartSiemens of empowers its customers to combining themaster real and their thedigital digitaltransformation worlds and sustainability challenges Requirements Automotive Machine Building Performance Data Food and Beverage Chemicals Sensor Data Additive Manufacturing Cranes Pharma Campus Buildings Multiphysics Simulation Control ModelsRail Ergonomic Simulation Hospitals Energy Distribution Airports Process Plants The comprehensive Digital Twin Page 5 Logistics Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Oil and Gas Opportunities for the Digital Twin Digital worlds Boost innovation via Digital Engineering Page 6 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Real worlds Master complexity Via novel Digital Services Why now? Enablers & Drivers. "Moore’s Law" – More than Moore - Cloud: Exploding computing capacity beyond scaling of chip performance and cloud power, e.g. GPUs, Reconfigurable Computing, … Algorithmic improvements: Creating breakthroughs will contribute significantly to efficiency of engineering process as well as open new ways of working and business propositions Integrating Heterogeneous Models: different physics, different formulations, different scales: Multiphysics simulation – Co-simulation – FMI/FMU - Model Order Reduction - … Internet of Things: performance data everywhere and readily accessible Data analytics – Data driven performance monitoring and modeling Challenged by increasingly complex systems and system requirements: Mechanics – electronics – control – software… get tightly interconnected. Performance demands become increasingly complex Page 10 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Digital Twin - A new age of computational paradigms Model Pioneers ~1985 Geometry Single-Physics ~2000 Computer-Aided Tools in R&D Multi-Physics Collaborative Model-based Engineering CAx: Computer Aided Design, Engineering, & Manufacturing Page 16 Digital Twin Era R&D Pervasion CAx Sunrise Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Industrial Metaverse ~2015 Model-based Operations X-as-a-service ~2022 TODAY Democratization Realtime++ Predictions Digital Twin – State of Industrial Adoption Today CAx Sunrise Model Pioneers ~1985 ~2000 Industrial Metaverse Digital Twin Era PLM Pervasion ~2015 ~2022 TODAY The Digital Twin market grows with annual CAGRs of 40-60% in maintenance, business optimization, performance monitoring, … Many companies struggle to implement Digital Twins. Road-blocks include company organizations change of business and processes, IT, … Sources: Page 17 Digital Twin Market by Technology, Type, Application, Industry, and Geography – Global Forecast to 2026, Markets and Markets Implementation Model in the Context of Use of Digital Twins, Digital Twin Readiness Assessment Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Design & Engineering 4.0+ requires novel paradigms Decreasing productivity per researcher User Number (log) Simulation User Trends Years 60s-70s 80s-90s 00s-10s 20s 30s? high Research & Development ≤ 3.0 Sequential design loops  Expert limited  Manufacturing constraint designs  Discipline limited optimization  Little software-based assistance Sources: Frank Piller (2021): Engineering and Design 4.0, Workshop at RWTH Page 20 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Simulation Experience low Research & Development ≥ 4.0 Interactive collaborative design environments  Democratized and compute limited  Usage and feedback driven design  Holistic product optimization  Autonomous design assistance Major Challenges in Digital Twin based Design  95% of costs are committed at the design stage  Serial processes are enforced due to expert focused tools  CAE does not support collaboration and fast iteration loops  Up to 80% during time-to-market is waste due to waiting Sources: Democratization of CFD, Industrial Design, User Interface Evolution, Innovation and Product Development, The Changing Role of Simulation Page 21 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Generative Artificial Intelligence No! We lack big Data in Engineering to achieve Generative AI for product design! Challenges @Engineering Image Analytics E.g., Google Open Images: 9 million images, 80 million annotations, and 600 classes Alpha Fold Trained on structures of around 100,000 unique and very different proteins as available from public data-bases Alpha Code Pre-training dataset contains a total of 715.1 GB of code from public GitHub repositories (C++, Python, …) Alpha Go Trained using reinforcement learning (semi-supervised) based on 44 million training games Sources: Towards Data Science, Nature, arXiv, Nature Page 23 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Manual data generation / annotation requires high expertise There are ‘no’ public engineering data sets Simulations are orders of magnitude more expensive than evaluation of a single Go move Major categories of Generative Design and Engineering applications Generative Design Category Generative Design Parameter Optimization Shape Optimization Topology Optimization Generative Engineering Co-Pilots Page 25 Use this application when you want to… Examples … quickly take an existing simulation driven design process to the next level. Optimization of design parameters such as position of cooling channels Optimization of control parameters Sizing of individual components … optimize an existing 2/3D design. Optimization of a car shape to reduce drag Optimization of a compressor blade configuration to improve efficiency … realize a completely new 2/3D design, e.g., exploiting by new manufacturing technologies. Additive Manufacturing: Optimization of a mechanical structure with minimal weight / costs Optimal flow ducts: Freeform design of flow ducts with minimal flow resistance … identify an optimal system configuration and architecture. Optimal eVTOL design balancing weight, flight distance, and safety Drive-cycle specific HEV powertrain architectures optimization … increase the CAE user base as well as efficiency of current users. Provide natural language based user interfaces Automate document as well as Q&A search Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) !! 10x faster Concept Design -90% cost Design for Additive Manufacturing CAE: Aided Design Page Computer 33 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) 0 experts CAE Democratization “Internet of Things” requires different technology than the “Internet of People”. “Internet of People” “Internet of Things” Chat Bots, Recommender Systems,… Diagnosis, Control, Service, …  High volume of similar data  People love to share data  People often “not well understood”  Data with very different context Page 36 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023)  Data availability is limited  Well understood machines (rigorous mathematical models) Major Challenges in Digital Twin based Operations 249 votes Digital Twins are slow and bespoke! Page 37 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) The Executable Digital Twin Digital Self-contained executable digital model of an asset Page 38 ! Real Digital Real Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Leveraged by anyone at any point in the lifecycle Why? The xDT is a unique concept allowing designers, manufacturers and operators to innovate and to meet engineering goals. Sustainability Continuously optimize performance based on live data to reduce waste. Cost optimization Drive efficiency and productivity with smart, adaptive control systems. Quality Detect and mitigate degradation or defects reliably and automatically. Page 43 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Major categories of Executable Digital Twin applications xDT Category Use this application when you want to… Virtual Testing & Commissioning … prepare for how your asset or system would interact with other assets, systems, or people. Testing of automation by virtual commissioning Testing new control strategies on gas turbines Operator training Virtual Sensing …measure something in your asset or system, but it isn’t feasible to put a sensor there. Temperature inside electric rotor Pressure distribution inside a gas turbine Free-flow inside a sewage network Diagnosis & Identification …know why your asset or system is behaving the way it is. Unbalance detection of large rotors Leakage detection in a water distribution network Predictive maintenance for machine tools Performance Prediction …know how your asset or system might behave in future operation. Remaining useful lifetime of electric motors Monitoring of coking in steam cracking furnaces Movement of people in emergencies Performance Optimization …inform actions on how to control the asset or system (with or without a Human-in-the-Loop). Model predictive control of a chemical reactor Pump schedule optimization of oil pipelines Operating point setting of catalyst modules Source: D. Hartmann (2021): Real-time Digital Twins. Page 45 Examples Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Enable the Impossible -90% error Manufacturing Quality CAM: Computer Aided Manufacturing Page 51 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) 0 experts CAM Democratization Volume of Relevant Data ML combined with Simulation enables the Digital Twin at scale Machine Learning Digital Twin Knowledge based Simulation Model Physical / Knowledge Courtesy to L. Horesh (2016): Should you derive? Or let the data derive - Towards a first-principles data-driven symbiosis Page 53 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) ! Digital Twins and the Industrial Metaverse Interoperability Functional economy $ Persistent virtual worlds Industrial Metaverse Unlimited live interactions Source: MIT Technology Review Insights (2023): The emergent industrial metaverse Page 59 Unrestricted | © Siemens 2023 | Dirk Hartmann | Executable Digital Twin (November 2023) Decentral Internet Real-world relevance Lehrstuhl für Medientechnik Fakultät für Elektrotechnik und Informationstechnik Technische Universität München Applications of 3D reality capture - Inventory and modeling (how many of X, where is Y, …) - Progress documentation (e.g. construction sites) - Planning (e.g. modification of existing production line, planning of new ones) - Indoor navigation (e.g. for external service personel) - Quality control - Safety planning - Forensics - Real estate - Monument protection (Denkmalschutz) - And many more Eckehard Steinbach Digital Twinning of Indoor Spaces 4 Industrial Example   37 classes more than 1000 annotated 3D objects Eckehard Steinbach Digital Twinning of Indoor Spaces 24 ! Limitations of existing indoor solutions Wi-Fi-based localization Room-level localization requires up to 6 hotspots People, doors, etc. lead to large positioning errors Visible light communication Image source: Bytelight Ambient magnetic field No orientation estimation High infrastructure and maintenance costs Image source: IndoorAtlas Ltd. Eckehard Steinbach Digital Twinning of Indoor Spaces 31 Global Features 0.025 data base 0.02 Example (Color) histogram Similarity of two images 0.015 0.01 K  H1 , H 2    H1 (k )  H 2 (k ) 0.04 0 -50 0.035 k 1 0.03 query 0.005 0.025 0 50 100 150 200 250 300 0.03 0.025 0.02 0.02 0.015 0.015 0.01 0.01 Simple compact description Some invariance Discriminative power very limited 0.005 0.025 0.005 0 -50 0 50 100 150 200 0 -50 250 300 0 50 100 150 200 250 300 0 50 100 150 200 250 300 0.02 0.015 0.01 0.005 0 -50 Eckehard Steinbach Digital Twinning of Indoor Spaces 41 Local Features Largely invariant against perspective distance and focus illumination image resolution orientierung 50 100 Robust against noise 150 200 250 300 Eckehard Steinbach 50 Twinning 100 of Indoor 150 Spaces 200 Digital 250 300 350 400 42 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich The Goal of Ergonomics Spatial and temporal optimization of working conditions, workflow, arrangement of objects to be objects to be grasped, work equipment for a task by taking into account the human abilities and characteristics for an efficient and error-free work and for the protection of the human being from health damages also with long-term exercise of an activity. Design and Evaluation of Humane work situations Prof. Klaus Bengler | Human Digital Twins Development and Application 7 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Humane work 1. Feasibility of the work: Is the person capable of performing the work required of him? 2. Tolerability of the work: Can the work be performed regularly and for a working lifetime without risk of damage to health? 3. Reasonableness of work: Are minimum social requirements or statutes enacted by the legislature met? 4. Job satisfaction: Are the (individual) needs of people taken into account when designing the working conditions? Prof. Klaus Bengler | Human Digital Twins Development and Application 8 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Work types Work types Requirements Organs/ functions Example Apply forces Muscles Tendons … Load transport, Machine feeding Motoric Movement Sensory organs Muscles … Manual processing of hand-held tools, Precision mechanical assembly work Reactive React/Act Responsiveness Retention ability... Driving a vehicle, airplane or train Combinative Combine Information Thinking ability Retention ability Gear design, Development of an exoskeleton Creative Generate Information Thinking ability Retention ability Conclude Invention Mechanical Luczak, H.; Volpert, W. et al.(1987). Arbeitswissenschaft. Kerndefinition Gegenstandskatalog – Forschungsgebiete Prof. Klaus Bengler | Human Digital Twins Development and Application 9 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Modeling Aspects Anthropometric Modeling Biomechanical Modeling Physiological Modeling Cognitive Modeling Prof. Klaus Bengler | Human Digital Twins Development and Application 11 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Human reliability Work load The entirety of all external conditions and requirements in the working system, which could influence a person physically and/or psychologically. Working stress The effect of the work load on a person relative to his/ her individual characteristics and capabilities. Example for understanding: The same mountain tour is done by two people. One person is trained, the other person is not. In spite of the same work load (acting factor), the perceived working stress will be different between the two persons. => Stress beyond a person's capabilities causes errors, injuries and death. Prof. Klaus Bengler | Human Digital Twins Development and Application 14 ! Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Goals of Human Digital Twins in a Product Context Understand the variability of end users Anthropometry Sensoric capabilities Variability of usage contexts Maximize product suitability Due to the extreme variability of end users and usage contexts it is necessary to consider this in the early phases of product development Design products and interaction with respect to human variability Prof. Klaus Bengler | Human Digital Twins Development and Application 16 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Goals of Human Digital Twins in a Production Context Understand the worker Capabilities Behavior Load and Stress on the worker Reliability Maximize production output Due to demographic change it becomes more necessary to uphold productivity of individuals until retirement Human error, injury and death leads to loss productivity Design working environments prioritizing human needs Prof. Klaus Bengler | Human Digital Twins Development and Application 17 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Digital human models Cognitive digital human models Physical digital human models Anthropometric models Biomechanical models Highly purpose specific Prof. Klaus Bengler | Human Digital Twins Development and Application 19 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich RAMSIS Anthropometric digital human model Development of vehicle interior Reachability analysis Sight analysis Adaptation to anthropometric properties Prof. Klaus Bengler | Human Digital Twins Development and Application 20 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich RAMSIS Input Geometry of environment Task Test collective Prof. Klaus Bengler | Human Digital Twins Development and Application Output Anthropometry Posture Field of view Grasping envelope Visualization 21 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich ema Work Designer Anthropometric digital human model Visualization and Analysis of assembly workplaces Cycle times (MTM-UAS) Ergonomics (EAWS, reachability and sight analysis) Optimization of working paths Analysis of human-robot-interaction Prof. Klaus Bengler | Human Digital Twins Development and Application 25 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich ema Work Designer Input Product data Ressource data Layouts Task Output Prof. Klaus Bengler | Human Digital Twins Development and Application Working plan Visualization Layout Working time analysis (MTM-UAS) Ergonomics analysis (EAWS) Validation of task feasibility Movement data of human 26 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich OpenSim Biomechanical digital human model Open source software Modeling, simulating, and analyzing the biomechanics of human movement Create and modify musculoskeletal models, Perform dynamic simulations Evaluate muscle forces, joint loads, and kinematics Prof. Klaus Bengler | Human Digital Twins Development and Application 27 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich OpenSim Input Geometry Kinematics Dynamics Control of the system Motion of the human Prof. Klaus Bengler | Human Digital Twins Development and Application Output Muscle activation Joint reaction forces Load Stress 28 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich AnyBody Biomechanical digital human model Proprietary software Modeling, simulating, and analyzing the biomechanics of human movement Create and modify musculoskeletal models, Perform dynamic simulations Evaluate muscle forces, joint loads, and kinematics Prof. Klaus Bengler | Human Digital Twins Development and Application 29 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich AnyBody Input Geometry Kinematics Dynamics Control of the system Motion of the human Prof. Klaus Bengler | Human Digital Twins Development and Application Output Muscle activation Joint reaction forces Load Stress 30 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Climate Comfort Humane climate conditions at workplaces Modeling climate comfort Posture Temperatures Information on usually comfortable perceived climate Prof. Klaus Bengler | Human Digital Twins Development and Application 31 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Current state Digital human models can represent Specific humans Generic, representative humans Purpose-specific Often not connection between the models Feedback to and from the real world often not automatic Prof. Klaus Bengler | Human Digital Twins Development and Application 33 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Connection of different types of models Connecting different types of human models requires: Standardized data exchange formats Understanding of which outputs from one model must be used as input for the other Understanding, which data needs to be availability at which level of detail Prof. Klaus Bengler | Human Digital Twins Development and Application 44 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Connection of different types of models Humans are highly complex systems Understanding requires different models: Movement data acquired from an anthropometric model can be used as input for the biomechanical model Prof. Klaus Bengler | Human Digital Twins Development and Application 45 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Connection of different types of models Automation of human simulation: Current simulations are prone to human errors, but require humans to set them up Prof. Klaus Bengler | Human Digital Twins Development and Application 46 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Data availability Human DTs face many data-related challenges, such as: Data availability and acquisition: high demand for data quantity and quality Need for special methods to collect data Lack of real-time data and automated data collection Prof. Klaus Bengler | Human Digital Twins Development and Application 47 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Data availability Data privacy and ethics: Unlike machines, humans have rights over their data Trade-off between personalized data and data protection, legal and moral implications of data use and sharing Prof. Klaus Bengler | Human Digital Twins Development and Application 48 Chair of Ergonomics TUM School of Engineering and Design Technical University of Munich Quo vadis? Future research on human digital twins will aim to create work environments that can adapt to individual needs and preferences This can enhance worker productivity and satisfaction, as well as reduce the chances of injuries and human errors in the workplace To achieve this, individualized data must be collected and processed, which requires addressing issues such as data availability, privacy protection, and data acquisition automation These challenges require a comprehensive approach that considers the technical, architectural, and collaborative aspects of digital Twins. The interplay between different modeling aspects has to be understood and implemented Prof. Klaus Bengler | Human Digital Twins Development and Application 50 Digitization Industrial Revolution up to Logistics 4.0 Industry 1.0 Mechanization, Steam power, loom Industry 2.0 Mass production, assembly line, electrical energy Industry 3.0 Automation, Computer und Electronics Industry 4.0 Logistics 4.0 Cyber-Physical Systems, Internet of Things, Networks Represents networking and integration of logistics processes Source: m-focus.co.th fml – Chair for Materials Handling, Material Flow, Logistics | TUM School of Engineering and Design | Technical University of Munich 3 Digitization Value of Data Role of Data Data as a product Data as an enabler of products Data as an enabler of processes Data as a result of processes 1970 1980/1990 2000 today Time Source: Otto, B., Fraunhofer-Gesellschaft (Hrsg.): Digitale Souveranitat. Beitrag des Industrial Data Space, Fraunhofer-Gesellschaft e. V., München 2016. fml – Chair for Materials Handling, Material Flow, Logistics | TUM School of Engineering and Design | Technical University of Munich 4 Digitization Transformation of Logistics 3.0 to Logistics 4.0 (1/2) Logistics and Production 3.0 Logistics and Production 4.0 Central control, rigid, complex Decentralized self-organization through networking Deterministic decisions Decisions contextually based on real-time simulations Established value chains Virtual ad-hoc organizations value networks Source: BVL, Digitization in Logistics fml – Chair for Materials Handling, Material Flow, Logistics | TUM School of Engineering and Design | Technical University of Munich 5 Digitization Transformation of Logistics 3.0 to Logistics 4.0 (2/2) Logistics and Production 3.0 Logistics and Production 4.0 Pre-planned systems Autonomous, self-organizing logistics and production units Expansion by significant redesign Expansion by modularization Carriers, work pieces, products as passive objects Intelligent carriers, work pieces, products actively support production and logistics processes Compulsory attendance of employees Flexible use of employees (availability calendar, expertise lists) Source: BVL, Digitization in Logistics fml – Chair for Materials Handling, Material Flow, Logistics | TUM School of Engineering and Design | Technical University of Munich 6 Tasks of fleet controllers for AGVs AGV fleet controllers are digital twins of mobile transport systems Incoming orders (e.g. from ERP system) Scheduling When is the job done? Routing Which route should be followed? Dispatching Using information about the current state of the transport system: vehicle positions routes order backlog Which vehicle does the job? www.opentcs.org/de/aufbau.html Executing Vehicle fml – Lehrstuhl für Fördertechnik Materialfluss Logistik | TUM School of Engineering and Design | Technische Universität München 18 Insight into Research Results: Concept Three steps form the concept of an automatic re-documentation Existing Conveyor System Source: Gebhardt Fördertechnik Data Acquisition How can raw data be collected with minimal effort? Data Interpretation How can movement and environmental data be derived from the raw data? Re-Documentation How can the modeling data be re-documented from movement and environment data? Data for Model generation fml – Chair for Materials Handling, Material Flow, Logistics | TUM School of Engineering and Design | Technical University of Munich 28 Use Case Summary Smart box passing over existing Conveyor Systems Automated Re-Documentation of Modelling Data Economically viable Application of Digital Models also for existing Conveyor Systems Result of the Research Project: Contribution to increasing the Usage of Digital Models fml – Chair for Materials Handling, Material Flow, Logistics | TUM School of Engineering and Design | Technical University of Munich 34 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München 1 Ursprünge M. Grieves K. Iwata 3 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München  1990er: „Doubleganger“, „Mirror Worlds“  1993: „Virtual Manufacturing System“  2002: „Conceptual Ideal for PLM“  2005: „Mirrored Spaces Model“  2006: „Information Mirroring Model“  2010: „Digital Twin“ Quelle: [Gelernter1993, Onosato1993, Iwata1995, Grieves2016, NASA2010] 5 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München 2 Definition und Interpretation M. Grievers (2002): „a set of virtual information constructs that fully describes a potential or actual physical manufactured product from the micro atomic level to the macro geometrical level. At its optimum, any information that could be obtained from inspecting a physical manufactured product can be obtained from its Digital Twin. The Digital Twin concept model contains three main parts: a) physical products in Real Space, b) virtual products in Virtual Space, c) and the connections of data and information that ties the virtual and real products together.“ Quelle: [Grieves2002] 6 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München 3 Sicht der Produktionstechnik Digitaler Schatten > - ABB. der Prozesse !  Handlungsfeld von Industrie 4.0  Hinreichend genaues Abbild der Prozesse in der Produktion, der Entwicklung, und angrenzenden Bereichen  zur Schaffung einer echtzeitfähigen Auswertebasis. Quelle: [WGP2016] 9 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München Digitaler Zwilling  Baut auf digitalem Schatten auf  Zusammenspiel aus unterschiedlichen Prozessmodellen und Simulationen  Möglichst identisches Abbild der Realität Quelle: [WGP2016] 10 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München Digitaler Zwilling  Interagiert mit digitalem Schatten und/oder realem System  Zusammenspiel aus unterschiedlichen Prozessmodellen und Simulationen  Möglichst identisches Abbild der Realität Reales System Korrelation Digitaler Schatten Kausalität Digitaler Zwilling 11 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München 4 Einsatz im Kontext der Umformtechnik Ebenen  Produktionsnetzwerk  Umformmaschine  Werkzeug- und Prozessebene 12 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München ! Zwecke auf Produktionsnetzwerkebene  Produktionsüberwachung  Produktionssteuerung  Logistikplanung  Restrukturierung und Fabrikplanung 13 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München Zwecke auf Anlagenebene  Vorbeugende Instandhaltung  Schulung  Virtuelle Inbetriebnahme  Fehlersuche 14 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München Zwecke auf Werkzeug- und prozessebene  Vorbeugende Instandhaltung  Prozessführung und -regelung  Qualitätsmanagement  Fehlersuche 15 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München I Digitaler Zwilling  Minimierungsprogramm mit kostenbasiertem Zielfunktional 𝐾𝑜𝑠𝑡𝑒𝑛 = 𝐾𝑜𝑠𝑡𝑒𝑛 + 𝐾𝑜𝑠𝑡𝑒𝑛 + 𝐾𝑜𝑠𝑡𝑒𝑛  Bilanzbedingungen für Standorte über Zeithorizont 𝐿𝑎𝑔𝑒𝑟 + 𝑃𝑟𝑜𝑑𝑢𝑘𝑡𝑖𝑜𝑛 + 𝐿𝑖𝑒𝑓𝑒𝑟𝑢𝑛𝑔 + 𝐹𝑒ℎ𝑙𝑡𝑒𝑖𝑙𝑒 ≥ 𝐵𝑒𝑑𝑎𝑟𝑓 + 𝑉𝑒𝑟𝑠𝑎𝑛𝑑  Nebenbedingungen der Betriebspraxis Quelle: [Opritescu2018] 23 Lehrstuhl für Umformtechnik und Gießereiwesen TUM School of Engineering and Design Technische Universität München ! Motivation Local controller Global controller Erhebliche Wartezeit, da die Geometrie erst nach dem Umformen bewertet werden kann Entwickelt, um inline und in Echtzeit zu arbeiten Abweichungen können nur an lokalen Biegemerkmalen (Radius und Winkel) behandelt werden Abweichung der gesamten Kontur kann kompensiert werden (Krümmung und Verdrehung) Fehler, die in bereits gebogenen Abschnitten entstehen, können nicht kompensiert werden und übertragen sich auf die nächsten Abschnitte Frühere Fehler können in den nächsten Durchläufen durch Minimierung eines globalen Qualitätsmaßes reduziert werden Quelle: [Lechner2023] 37 Definition AR/VR/MR/XR Mensch-Maschine-Kommunikation (MMK) – Eingebettet in die physikalische Welt Menschliche Intelligenz & Motivation MMK-Ebenen 1. Funktion 2. Ergonomie 3. Kognitive Nutzbarkeit 4. UX und Physikalität Nutzer Physikalische Welt Wahrnehmung, Interpretation Eingabegerät Gamification Intelligence Amplification (IA) IA > AI [Fred Brooks] Interpretation Ausgabegerät Virtuelle Welt System Künstliche Intelligenz (KI) Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. 5 Definition AR/VR/MR/XR XR: Multimediale/multimodale Kombination aus “Physikalität” und “Virtualität” Verteilte Systeme Drahtlose Kommunikation Echtzeit Systemarchitektur Tragbare Geräte Digitaler Zwilling Virtuelle Welt Datensicherheit Physikalische Welt Multimodale Displays Multi-modale Sensordaten Manipulation physikalischer XR-System Objekte Information Dynamische Modellierung Robotik Maschinelles Lernen Dynamiche Visualisierung Multimodale MMK Multimodale Darstellung Simulation Spiele Sicherheit Szenenanalyse Tracking Nutzer Adaptation Ergonomie Nutzbarkeit Motivation Kognition Nutzerzentrierte Anforderungsanalyse Psychologie Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. 6 Definition AR/VR/MR/XR !! Die XR Erfahrung (UX) Wir sind von Information umgeben. Information ist echt, aber wir können sie nicht immer vollständig erinnern Kritische Technologien: mit unseren begrenzten Sinnen wahrnehmen. youtu.be/drAvbw1G3ZQ stationär Computer können uns dabei helfen, Information zu explorieren, zu analysieren, zu verstehen und zu genießen. Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. VR 2D PC/Laptop 3D AR mobil Phone/Tablet 7 Definition AR Augmented Reality [Azuma 98] Mischung: Real + virtuell Interactiv in Echtzeit Dreidimensionale Registerung [ECRC 93] Mixed Reality Continuum: [Milgram and Kishino 94] Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. 8 1 Hardware Verschiedene Ansätze Kopf-basiert Hand-basiert Auf realen Oberflächen Multi-touch Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. Tisch-basiert Hybrid und Tangibel 11 ! AR/VR und Digitale Zwillinge in der Prozessindustrie EvolutionAnlagenstruktur + Auslegungsdaten Mensch Anlagenstruktur + Auslegungsdaten - TIR 003 TIR 002 TIR 003 FIRC 001 TIR 002 - FIRC 001 Anlagenstruktur + Auslegungsdaten Prozessdaten Prozessdaten TIR 003 TIR 002 Konsistente Daten FIRC 001 Equipmenthistorie ? V-001 Anlagenstruktur Anlagenstruktur + & Auslegungsdaten Designdaten Prozessdaten V-002 Equipmenthistorie V-003 Daten der Wartungsdaten Instandhaltung V-004 TIR 003 V-361 TIR 002 Defekt: Wartung: FIRC 001 V-001 ? ? Analyse V-002 ? V-002 V-003 Analyse V-004 Historie V-361 Defekt: Wartung: V-001 Prozessdaten Maschine V-003 Equipment Wartungsdaten Equipmenthistorie Prozessdaten Ortsunabhängiger (in Anlage, Leitwarte, Werkstatt, ?Home Office?) Datenzugriff Kollaborativ (Handwerker, Zulieferer, Betriebspersonal) Analyse - Möglichst automatisiert (ML) Heterogene Daten (Informationsmodelle) Steigende Datenmenge (IoT/Big Data) Simulierbarkeit V-004 Wartungsdaten V-361 Defekt: Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. Wartung: 22 Zusammenfassung Durch die Bündelung (das Mapping) von Datentöpfen einer Prozessanlage zu einem Digitalen Zwilling können in Zukunft AR/VR Anwendungen für die Industrie einfacher, schneller und damit kosteneffizienter implementiert werden können. Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. 25 Wie profitieren Digitale Zwillinge von AR/VR? (1/3) Datenselektion Anstatt funktionale Strukturen (Hierarchien) zu durchsuchen, kann der Nutzer direkt räumlich das Teil auswählen, zu dem er Informationen braucht Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. 26 Wie profitieren Digitale Zwillinge von AR/VR? (2/3) Datenvisualisierung Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. - AR: Die funktionalen Informationen können direkt in der Anlage visualisiert werden - VR: Alle nötigen Informationen können ortsunabhängig visualisiert werden 28 Zusammenfassung ! AR/VR und Digitale Zwillinge sind in der Industrie Schwesternkonzepte, die sich gegenseitig unterstützen können ! AR/VR Technologien sind nicht nur geeignete Visualisierungsmöglichkeiten für komplexe Digitale Zwillinge, sie helfen auch bei der Erstellung und Aktualisierung von Digitalen Zwillingen ! Standardisierte Digitale Zwillinge, die die heterogene Datenlandschaft von Prozessanlagen mappen und strukturieren, erleichtern die Erstellung von AR/VR Anwendungen für die Industrie immens. Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. 38 Forschungsmethoden FrontendTechnologien Mensch IST-Prozesse dokumentieren Mit Werkern Use Cases erarbeiten Demonstratoren entwickeln Potential identifizieren Konzepte für AR/VR entwickeln Machbarkeit und Skalierbarkeit evaluieren Digitale Zwillinge und Augmented Reality | Linda Rudolph, M.Sc. & Prof. Gudrun Klinker, Ph.D. AnlagenInfrastruktur IST-Daten analysieren Matchingstrategien entwickeln Plattform konzipieren / testen 40 What is a digital twin? Kritzinger, W., Karner, M., Traar, G., Henjes, J., Sihn, W., 2018. Digital twin in manufacturing: A categorical literature review and classification. 16th IFAC Symposium on Information Control Problems in Manufacturing INCOM 2018. UAV Fleet Personalized Tumor Treatment Potential of quantifying uncertinties in DT approach for nuclear power systems Deeper understanding of sources of uncertainty and their propagation throughout the entire lifecycle Kapteyn et al (2021) A probabilistic graphical model foundation for enabling predictive digital twins at scale Nuclear Power Systems Kochunas et al. (2021) Digital twin concepts with uncertainty for nuclear power applications Chaudhuri et al. (2023) Predictive digital twin for optimizing patient-specific radiotherapy regimens under uncertainty in high-grade gliomas Additive Manufacturing Nath et al. (2022) Probabilistic digital twin for additive manufacturing process design and control. Probabilistic Digital Twins Physical Environment Digital Environment ”updating” Property Data Modelling Digital State ”updating” Behavior Data Behavioral Prediction data used directly Data Physical State Straub & Cotoarbă | Probabilistic Digital Twins “collect information” ”control/action” Decisions Property Data Physical Environment Digital Environment ”updating” Property Data Direct observation of property Intrusive Modelling Digital State Non-Intrusive ”updating” Behavior Data Behavioral Prediction data used directly (Usually) Expensive Data è Sparse measurements Physical State “collect information” ”control/action” Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Decisions Behavior Data Physical Environment Digital Environment ”updating” Observation of behavior of physical asset (Mostly) Non-Intrusive Cheap Expensive è Integration could be challenging Property Data Modelling Digital State ”updating” Behavior Data Behavioral Prediction data used directly Data Physical State “collect information” ”control/action” Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Decisions Modelling Physical Environment Spross, J., & Larsson, S. (2021) Digital Environment ”updating” Inference At unknown points Transformation Property Data Modelling Observation ó Properties Accuracy Amount of training data Modelling* assumptions *traditionally deterministic à stochastic Digital State ”updating” Behavior Data Behavioral Prediction data used directly Data Physical State “collect information” ”control/action” Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Decisions Example Visualization Model Example Data Model Digital State Physical Environment Spross, J., & Larsson, S. (2021) Digital Environment ”updating” Data Model Database storing all properties Stochastic and deterministic description Property Data Modelling Digital State ”updating” Behavior Data Level of Detail Accuracy dependent on Level of Detail Behavioral Prediction data used directly Data Physical State “collect information” ”control/action” Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Decisions Behavioral Prediction Spross, J., & Larsson, S. (2021) Physical Environment Physics-Based Models e.g. Finite Element Analysis Data-Driven Models e.g. Regression ”updating” Property Data E.g., Informed ML “best of both” Modelling Digital State Hybrid Models Accuracy + Explainability Digital Environment Computational efficiency ”updating” Behavior Data Behavioral Prediction data used directly Surrogate Models e.g. AI or ML methods Data Physical State “collect information” ”control/action” Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Decisions Decisions Bismut, E., Cotoarbă, D., Spross, J., & Straub, D. (2023). Optimal adaptive decision rules in geotechnical construction considering uncertainty. Géotechnique, 1-12. Physical Environment Digital Environment ”updating” ”collect information” “control/perform action” 1st Stage support platform à copilot Property Data Digital State 2nd Stage Fully automated Modelling ”updating” Behavior Data Behavioral Prediction data used directly Data à Sequential decision making under uncertainty Physical State “collect information” ”control/action” Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Decisions Learning and updating the model via Bayesian analysis Bayesian statistics provides a consistent framework to combine data and models from different sources The principle: an a-priori probabilistic model is combined with a likelihood function describing the data: Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Sequential decision making under uncertainty Decisions/actions taken based on the PDT can be optimized Also known as stochastic control Challenge: not only the future but also the current state is only known with uncertainty (hence the PDT) Approaches to find optimal decision sequences: POMDP (Partially observable Markov decision process) solvers Reinforcement learning Heuristic strategy optimization Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Outlook & discussion PDT is essential when uncertainties can affect decisions Potential key application areas include: Geotechnical design Existing structural systems Systems with structural health monitoring Optimization of inspections and maintenance (predictive maintenance) Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Take home message The knowledge of engineering systems is not perfect and this can affect decision making Uncertainty is represented by augmenting the system‘s digital representation with stochastic parameters Bayesian analysis allows to efficiently integrate data and update the digital representation PDT technology is ready for basic settings (as in classical stochastic control), but still requires R&D for applications where decisions are taken by humans Straub & Cotoarbă | Probabilistic Digital Twins | Lecture series "Digital Twin in Engineering and Design“ | 20.12.2023 Definition eines Digitalen Zwillings Vom Modell zum Digitalen Zwilling Aus der Industrieperspektive Ein digitaler Zwilling bezeichnet die virtuelle Abbildung eines physischen Gegenstands oder Systems über dessen gesamten Lebenszyklus.1 Aus der Forschungsperspektive Eine dynamische, virtuelle Abbildung eines physischen Objekts über dessen gesamten Lebenszyklus unter Verwendung von Echtzeitdaten, um Verständnis, Erkenntnisse und Schlussfolgerungen zu ermöglichen2,3 Mikell, Matthew, and Jen Clark. "Cheat sheet: what is Digital Twin." IBM Internet of things blog. See https://www. ibm. com/blogs/internet-of-things/iot-cheat-sheet-digital-twin (2018). Bolton, Ruth N., et al. "Customer experience challenges: bringing together digital, physical and social realms." Journal of service management 29.5 (2018): 776-808 National Infrastructure Commission. "Data for the public good NATIONAL INFRASTRUCTURE COMMISSION." (2017). © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften 2 Digitale Darstellungen realer Objekte Vom Modell zum Digitalen Zwilling Digitales Modell eine digitale Version eines Objektes ohne automatisierten Datenfluss zwischen dem Objekt und dessen digitalem Modell digitale Version reales Objekt digitales Abbild reales Objekt digitales Abbild reales Objekt Digitaler Schatten ein dynamisches, digitales Abbild eines Objektes, welches nur einen einseitig automatisierten Datenfluss vom oder zum realen Objekt aufweist Digitaler Zwilling ein dynamisches, digitales Abbild eines Objekts, welches beidseitig einen automatisierten Datenfluss vom und zum realen Objekt aufweist © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften manueller Datenfluss automatisierter Datenfluss 3 Vorgehensweise Betriebsschwingungsformanalyse Erstellung eines Messmodelles der Maschine Messung bei Fremderregung (Betrieb) Auswertung und Schwachstellenanalyse Erfassung aller für die Charakterisierung der Schwingungsformen notwendigen Strukturpunkte Messung der Schwingungen an allen relevanten Strukturpunkten in Bezug auf einen Referenzpunkt Bestimmung der Schwingungsformen, Interpretation der Schwingungsformen © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften 26 Die Modellgenauigkeit kann durch Schwingungsanalysen an der Maschine verbessert werden. Vom Modell zum Digitalen Zwilling Übertragungsverhalten der Werkzeugmaschine Prozessmodell des Ratterwirkungskreises reales Objekt manuelle Messungen erforderlich Durchführung der Schwingungsanalyse präzise Bestimmung des Übertragungsverhaltens Werkzeugeigenschaften digitales Abbild Beschleunigungsdaten Stabilitätsdiagramm neue Aufstellelemente veränderliches Übertragungsverhalten © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften GROB-WERKE GMBH & CO. KG 28 Durch die automatisierte Anpassung an die Maschine wird das Modell zum Digitalen Zwilling. Vom Modell zum Digitalen Zwilling automatisierte Anpassung des Übertragungsverhaltens an Komponentenänderungen Digitaler Zwilling Werkzeugeigenschaften automatisierte Schwingungsanalyse Übertragungsverhalten der Werkzeugmaschine digitales Abbild reales Objekt automatisierte Übermittlung der Ergebnisse Beschleunigungsdaten aktuelles Stabilitätsdiagramm Prozessmodell des Ratterwirkungskreises Stabilitätsdiagramm Der automatisierte Signalfluss von der Maschine zum Modell und umgekehrt ist umgesetzt. Das Modell passt sich entlang des Lebenszyklus der Maschine an. © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften GROB-WERKE GMBH & CO. KG 29 Beispiel: CNC-Wälzstoßmaschine Betriebsschwingungsformanalyse Aufbau des Messmodells: 63 Messpunkte, 1 Referenzsignal (fallabhängig) Aufnahme der Beschleunigungssignale für 400, 500 und 600 Doppelhübe (fallabhängig) © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften 32 Finite-Elemente-Berechnungen Berechnung von Gestellbauteilen – Finite-Elemente-Methode Geometriedaten aus CAD-Modell Materialeigenschaften FE-Modell Lastfälle Dichte E-Modul Querkontraktionszahl Wärmeübergangskoeffizient Wärmedehnungskoeffizient … Verformungsbilder Wärmekarten Frequenzgänge Kräfte Momente Druck Beschleunigung Verlagerung Wärmequellen Frequenz in Hz © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften 37 Typen finiter Elemente Berechnung von Gestellbauteilen – Finite-Elemente-Methode geometrische Dimension φy Elemententyp y z φy x φx Freiheitsgrade pro Knotenpunkt x, y, z, φx, φy, φz 0 diskrete Feder, Masse, Dämpfer 1 Stabelement Wellenelement Balkenelement x x, φx x, y, z, φx, φy, φz* 2 Scheibenelement Plattenelement Schalenelement x, y, φz* z, φx, φy* x, y, z, φx, φy, φz 3 Hexaederelement Pentaederelement Tetraederelement x, y, z x, y, z x, y, z * die Aufteilung der Freiheitsgrade ist bei diesen Elementen teilweise programmspezifisch © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften Quelle: MILBERG (1995), S. 148 38 Virtuelle Inbetriebnahme von Steuerungstechnik Virtuelle Inbetriebnahme (VIBN) simulativ durchführbar ohne VIBN Konstruktion Konstruktion Fertigung/ Montage Inbetriebnahme (IBN) Fertigung/ Montage IBN Nutzen mit VIBN Modellierung Δt VIBN Aufwand Zeit © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften Quelle: W ÜNSCH (2008) 43 Virtuelle Inbetriebnahme: Hardware-in-the-Loop-Ansatz Virtuelle Inbetriebnahme (VIBN) Virtuelle Maschine Verhaltenssimulation Visualisierung Feldbus TCP/IP Visualisierung Geometrie Form, Lage, Abmessungen Kinematik Art, Länge, Anzahl von Achsen Verfahrbereichsgrenzen Kollisionsrechnung Materialfluss Physikalisches Verhalten von Maschine und Prozess Hardware-in-the-LoopAnsatz Schaltverhalten Bewegungsverhalten Materialfluss blockorientierte Simulation Steuerungstechnische Sicht © iwb – Institut für Werkzeugmaschinen und Betriebswissenschaften Buskonfiguration Eingangs-Ausgangs-Abbild Quelle: W ÜNSCH (2008) Steuerungshardware Steuerungssoftware frühzeitig entwickelte Steuerungssoftware auf geplanter, realer Steuerungshardware 44 Digitaler Zwilling: Definition und Beispiele NASA Definition The Digital Twin A concept which combines as-built vehicle components, as-experienced loads and environments, and other vehicle-specific characteristics to enable ultrahigh fidelity modeling of aircraft and spacecraft or their components throughout their service lives. Name 05.32.001 National Aeronautics and Space Administration © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack 3 Digitaler Zwilling: Definition und Beispiele WiGeP Definition & Zusammenhänge Legende: Name 05.32.001 Digital Master / Models Development (BoL) © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack Digital Shadow / Data Storage Operation and Condition Data, Process Data, … Guidance Adapts Adapts Constructs Virtual Prototype Digitaler Zwilling 1 Digitaler Zwilling 1 Digital Twin Data Virtual Prototype n Operating Data Virtual Prototype I Physical Twin Guidance Begin of Life Mid of Life End of Life Instantiation BoL MoL EoL User Digital Master / Models Production/Usage/Service (MoL) EoL Digitaler Zwilling: Definition und Beispiele Vorteile z.B. Predictive Maintenance Begin of Life Mid of Life End of Life Physical Twin Name 05.32.001 Digital Master / Models Development (BoL) © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack + Guidance + System Feedback Verschiedene Zwecke &Digitaler Arten Zwilling 1 Digitaler Zwilling 1 Digital Twin von Operation and Condition Data, Process Data, … Data Auswertung Digitalen Zwillingen Felddaten Digital Shadow / Data Storage Adapts Adapts Constructs Virtual Prototype Instantiation Virtual Prototype n Operating Data Virtual Prototype I User Guidance Legende: BoL MoL EoL + Digital Master / Models Production/Usage/Service (MoL) + Modell Training EoL Digitaler Zwilling: Definition und Beispiele Klassen und Eigenschaften Es lassen sich anhand verschiedener Kriterien (Datenstand, Modellqualität, Echtzeitfähigkeit, etc.) unterschiedliche Klassen von Digitalen Zwillingen unterscheiden. Über den Anwendungsfokus lassen sich am geeignetsten Aussagen über erforderte Aufwände zur Implementierung und Umfang des Digitalen Zwillings treffen. Name 05.32.001 Informational Digital Twin Supporting Digital Twin Dient der Informationsbeschaffung Dient der Entscheidungsunterstützung Betriebsdaten werden im Kontext vorhandener Modelle aufbereitet und dem Nutzer dargestellt Betriebsdaten werden ausgewertet und anhand von Entscheidungsmodellen in Handlungsempfehlungen an den Nutzer weitergegeben © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack Autonomous Digital Twin Autonomer Digitaler Zwilling Der Digitale Zwilling gibt die Handlungsanweisungen unmittelbar an sein physisches Gegenstück weiter (z.B. Anweisungen zur Optimierung) Digitaler Zwilling: Definition und Beispiele Szenarien entlang des Produktlebenszyklus BoL EoL MoL Prototyping Zustandsüberwachung Feldtests Prozessoptimierung Predictive Maintenance Effizienzanalyse Name 05.32.001 Produktionsplanung Konzept © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack Entwicklung Produktion Betrieb Entsorgung 11 Digitale Zwilling im Toleranzmanagement Anwendungsszenario Toleranzsimulation Simulierte FertigungsprozessAbweichungen simulation Adäquater Montagepartner und angepasste Montagedaten Fu 1. Spiegeln der Sensordaten in Produktionsprozesssimulation 2. Analyse geometrischer Abweichungen Entwicklung Name 05.32.001 Messdaten Sensordaten Produktion Prüfung 3. Auswahl von adäquaten Montagepartnern durch die Toleranzsimulation Montage Testen Quelle: Schleich2017 © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack 21 Der Digitale Zwilling im Toleranzmanagement Montagestrategien Bauteil 1 1 5 2 1 3 … 2 4 Bauteil 3 5 … 6 4 4 3 Potential für Digitalen Zwilling 6 Selektiv Selektiv Baugruppe 3 1 6 5 2 2 1 4 1 2 6 5 4 5 4 … … … 6 3 3 #1 #2 #3 #4 #5 … n Rein zufällige Kombination der Bauteile in der Montage Häufigste Montageart in der Serienfertigung Roth 2023 Name 05.32.001 5 6 Zufällig © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack 1 3 … 2 Bauteil 2 Individuell Individuell 3 5 1 2 1 4 2 1 4 4 2 6 3 5 6 3 5 6 #1 #2 #3 #1 #2 #3 #1 #2 #3 2 4 #1 – #2 – #1 #2 – #1 – #3 5 1 #3 – #3 – #2 4 5 2 3 1 3 5 2 4 2 3 6 1 6 1 6 4 5 … … … #1 #2 #3 #4 #5 #6 … 100%-Messung aller Bauteile Einteilung in Gruppen gleicher Qualität Paarung der Gruppen und zufällige Kombination der Bauteile der Paarungen 3 2 100%-Messung aller Bauteile Individuelle 1:1 Paarung 23 Digitale Zwilling im Toleranzmanagement Anwendungsszenario SMART Assembly 4.0 Realisierung einer autonomen, selbst-optimierenden Montagefabrik. Ergebnis ist die „Smart Assembly 4.0“ Systemsoftware für die Sortierung und Montageplanung, die Qualität erhöht ohne Toleranzen einzuengen. TEIL A Scan Sortieren Pairwise Matching Eintreffende Teile Anpassungen: Locators Clamps Sequence Datenbank Digital Twin Variationen treten in Eigenschaften der Teile auf Name 05.32.001 TEIL B Scan Sortieren Scan Pairwise Matching Quelle: Söderberg2017 © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack 24 Zusammenfassung Was sollten Sie aus dieser Lehreinheit mitnehmen? Im Allgemeinen beschreibt das Konzept des Digitalen Zwillings jedoch die Verbindung eines existierenden, physischen Objektes mit einem virtuellen Gegenstück. In der industriellen Praxis werden Bauteile rein zufällig, selektiv oder individuell zu Baugruppen assembliert. Auch im Toleranzmanagement gewinnt der Digitale Zwilling zukünftig an Bedeutung! Hier können Digitale Zwillinge genutzt werden, um ideale Fügepartner zu identifizieren oder Montageanpassungen zu bestimmen. Fertigungsprozess- Simulierte Dies erlaubt es, die Baugruppenqualität zu erhöhen, ohne die Toleranzen einzuengen! simulation Abweichungen Schleich 2021 Name 05.32.001 Adäquater Montagepartner und angepasste Montagedaten Fu Messdaten Sensordaten Entwicklung © LEHRSTUHL FÜR KONSTRUKTIONSTECHNIK Friedrich-Alexander-Universität Erlangen-Nürnberg Prof. Dr.-Ing. Sandro Wartzack Toleranzsimulation Produktion Prüfung Montage Testen 38 WITTENSTEIN SE Digitaler Zwilling ermöglicht Konnektivität im gesamten Produktlebenszyklus Digitaler Zwilling im Engineering Typ Informationen Produktmodelle Algorithmen Technische Daten & Vorgabewerte Annahmen zum weitern Life-Cycle … Digitaler Zwilling in der Produktion und Nutzung Instanz Informationen Auslegungsinformationen Fertigungs- und Qualitätsdaten Nutzungsdaten After-Sales Services … Der Digitale Zwilling ist über den Digitalen Faden (Asset ID nach IEC 61406) eindeutig mit dem realen Produkt verbunden und macht so die relevanten Daten zugreifbar. WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec 22.01.2024 5 WITTENSTEIN SE Was ist der Digitale Zwilling technical data Operating history Carbon Footprint Timeseries data Dynamics Model Maintenance history FMEA Thermal model Product Models / Algorithms Operational / Process Data WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec Engineering data Friction model Production data Product Data 22.01.2024 6 WITTENSTEIN SE Digitaler Zwilling als Enabler für Dienstleistungsangebote Kinematik Berechnung Statische Auslegung Dynamische Auslegung Energetische & Thermische Berechnung Virtuelle Inbetriebnahme Kinematik Transformation Leistung & Bauraum Schwingungsberechnung Verluste, Erwärmung & Lebensdauer Regler Parametrisierung Bewegungsüberlagerung inkl. Nutzlast definieren Geometrie definieren Eigengewicht definieren Leistungsdaten (Drehmoment & Drehzahl) Montierbarkeitsüberprüfung Bauraum Check WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec Eigen-/Resonanzfrequenzen Dynamikanalyse Präzision Komponentenverluste Thermische Auslastung Schmierstoffgebrauchsdauer Energie/ CO2-Ausstoß Inbetriebnahme-Parameter Parameteroptimierung 22.01.2024 7 WITTENSTEIN SE Digitaler Zwilling als Entwicklungspfad zum cybertronischen Produkt/System I4.0 konform Proaktiver Digitaler Zwilling Vernetzter Digitaler Zwilling (aktiv) Harmonisierter Digitaler Zwilling Fragmentierter Digitaler Zwilling Digital Twin Produkt Mechanisch / Elektrisch WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec Sensorik Firmware Mechatronisch Vernetzung Eigenintelligenz Cybertronisch 22.01.2024 8 WITTENSTEIN SE Twin Feedback Produkt Zugriff Entwicklung eines Stufenmodells und einer Checkliste im Unternehmen WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec 22.01.2024 9 WITTENSTEIN SE Standardisierung des Digitalen Zwillings Als Gründungsmitglied der Industrial Digital Twin Association e. V. (IDTA) bringt WITTENSTEIN die Asset Administration Shell als standardisierten digitalen Zwilling in die Anwendung: Neue Serviceangebote und digitale Geschäftsmodelle auf Basis eines standardisierten Datenaustauschs Interoperabilität als Voraussetzung für skalierbare Integration Wiederverwendung von Daten entlang des Produktlebenszyklus und über Unternehmensgrenzen hinweg Investitionsschutz durch IEC-Standard und Nutzerorganisation mit mehr als 100 Mitgliedern WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec 22.01.2024 11 Challenges in today's digital twin ecosystem Il Apple, PROSTEP 2019 Note: exemplary content, not complete Asset Administration Shell in der Praxis | Bernd Vojanec 13 22.01.2024 15 Evolution of Industry 4.0 Industry 4.0 2013 Start of Industry 4.0 Asset Administration Shell in der Praxis | Bernd Vojanec Asset Administration Shell 2015 Concepts and Specifications Industrial Implementation Technology Spin-off 2019 2020 Introduction of the Digital Nameplate Founding of IDTA with 23 Organisations 15 22.01.2024 Data and content for Asset Administration Shell Engineering 3D geometry models Drawings Simulation data Properties (e.g. via ECLASS) … Who we Asset Administration are, what weShell do. in der Praxis | Bernd Vojanec Documentations Operation Installation and operating instructions Certificates Declaration of conformity … Order data Operating data, Service Notice (e.g. via OPC UA) … 22 22.01.2024 ! Asset Administration Shell In A Nutshell – Motivation & Nutzen Realität: - Keine Transparenz über die Datenbasis - Daten werden singulär genutzt (Siloeffekt) - Fehlende Interoperabilität & Defizite in der Datenqualität - Zentralle und starre Architekturen Zielbild: - Transparenz der Daten - Mindset für Data-Sharing zwischen Unternehmen - Interoperabilität und Datenqualität - Dynamische Architekturen und digitale Services für jedes Asset WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec 22.01.2024 27 ! Asset Administration Shell In A Nutshell - Grundlegende Konzepte AAS Metamodell (Konzept der I4.0 Komponente) Submodel Templates - Teilmodelle beschreiben Aspekte und Services des Assets (also die Inhalte des digitalen Zwillings) Asset Administration Shell Submodels Shell Asset Administration Referenzieren auf Data Dictionary - Asset und AAS bilden eine I4.0-Komponente Interfaces Logisch verbunden mit Asset - Teilmodell Templates sind Vorlagen und können Aspekte „standardisiert“ abbilden - Merkmale und deren Definitionen sollten aus öffentlichen Wörterbüchern stammen (ECLASS) - Assets, AAS, Teilmodelle und Merkmale sind global eindeutig identifizierbar - Interfaces bilden standardisierte Schnittstellen auf die Inhalte des digitalen Zwillings (bspw. fürs Engineering in AML und für den Betrieb mit REST oder OPC UA) WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec 22.01.2024 28 Asset Administration Shell In A Nutshell – Implementierungsformen & Infrastruktur Services Statische Inhalte Offline Datenaustausch via AASX, AutomationML, JSON, XML, … Dynamische Inhalte API für online Datenaustausch via HTTP REST, OPC UA, MQTT, … I4.0 Infrastruktur Services AASX Server „Wörterbücher“ wie ECLASS AAS Repository Authentication & Authorization WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec Fähigkeiten und KI Peer-to-peer Interaktion via I4.0 Interaktionsabläufen wie autonomen Anbieterverfahren AAS Runtimes AAS Registry 22.01.2024 29 Digitales Typenschild Grundlage für weitere Services, welche Typenschildinformationen benötigen Condition Monitoring und Anomalie-Erkennung durch den smarten Antrieb Sensor- und Felddaten werden durch Parameter und Produkteigenschaften aus dem Digitalen Zwilling angereichert Standardisierte Schnittstellen ermöglichen die Nutzung auf gängigen IIoT-Plattformen und IIoTGateways Data Gateway cynapse Teach-in Anomalieerkennung cynapse Monitor WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec 22.01.2024 36 DPP4.0 Überblick Konzept DPP4.0 basiert auf zwei neuen IEC-Normen DPP4.0 = DNP4.0 Identifikation + AAS Inhalte + Security AAS Digital Product Passport 4.0 Digital Nameplate 4.0 (auf Basis IEC61406-1) WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec Asset Administration Shell (auf Basis IEC63278) 22.01.2024 44 DPP4.0 Demonstrator Datensouveränität – Datenhoheit und Zugriffsschutz  Identifizierende Informationen => Registry  Daten/Inhalte => Repository WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec  Verwaltung von Berechtigungen => Anbieter  Öffentlicher & Eingeschränkter Zugang 22.01.2024 47 Wrap Up Asset Administration Shell als Standardisierter Digitaler Zwilling AAS ermöglicht es Unternehmen die Fähigkeiten Digitaler Zwillinge Stufenweise auszubauen AAS standardisiert Datensemantik, Interfaces und Infrastruktur-Services IDTA als Nutzerorganisation stellt Spezifikationen und Software Open Source bereit Standardisierte „Core“ Spezifikation ermöglicht darauf aufbauende Innovationen und Services zur Differenzierung Skalierung von Geschäftsmodellen und KI-Anwendungen nur mit Standardisierung und Datenqualität AAS ist für den herstellerübergreifenden Informationsaustausch konzipiert WITTENSTEIN | Asset Administration Shell in der Praxis | Bernd Vojanec 22.01.2024 66 We are an Impact Company ! Our mission is to be a digital partner for sustainability and efficiency. In doing so, we pave the way for a smarter and greener future. DIGITALIZATION For Efficiency Eliminate waste Increase efficiency Optimize overall processes + ELECTRIFICATION For Decarbonization Most efficient energy The best vector of decarbonization Property of Schneider Electric | Page 4 Internal = SUSTAINABILITY Smarter & Greener The global Digital Twin market Source: IoT Analytics “Digital Twin Market Report 2023–2027”, 2023 Spending for digital twin software that are used to model, setup, and manage digital twins Manufacturing is the largest market (74%) in 2022 and is expected to show the strongest growth until 2027 with a CAGR* of 29.9%  PAM** in 2027 of 1.1$B for manufacturing market * Compound annual growth rate ** Potential available market Property of Schneider Electric | Page 7 General Adoption rates manufacturing companies start adopting globally 63% are currently developing or have already developed their digital twin strategy. 29% have fully implemented or are implementing a digital twin strategy for a portion of their operational assets Sources: 48% of those began to invest in digital twins within the last year Property of Schneider Electric | Page 8 General IoT Analytics “Digital Tw in Market Report 2023–2027”, 2023 IoT Analytics “Decoding Digital Tw ins: Exploring the 6 main applications and their benefits”, 2023 Concept of Digital Twins in our domain ! Digital twin is a virtual representation of a physical object, process, or system Combines historical, real-time, test data with algorithms to replicate the real-world behavior data structure Visualizations, 3D (names, data types, units of measurement, and other metadata) (derived from the data structure and the processed data) Populate the model with data Behavioral model Behavioral simulation (based on algorithms and causality) (Descriptive, diagnostic, predictive) Property of Schneider Electric | Page 9 Most common applications IoT Analytics looked at 100 digital twin case studies and classified each project Source: IoT Analytics “Decoding Digital Twins: Exploring the 6 main applications and their benefits”, 2023 Property of Schneider Electric | Page 10 General Industry applications orient around certain use cases Focus on system level Interoperability “standards” still under definition Full life-cycle coverage rather an exception External market drivers meet internal challenges Demand for more intelligent products in a connected world Shortened product life cycle reduce time-to-market enable decision making and robust change management improve quality, compliance, mitigate risk Manufacturing industry strives to create MORE innovative products, combined with FASTER time-to-market supported by BETTER decision making, quality and cost control. Traditional machine design involves multiple iterations and complex adjustments 75% Engineers from different disciplines need to work closely together 30% Physical testing and validation of machines is costly, complex and time-consuming 60% Unforeseen commissioning issues affect the project schedule Property of Schneider Electric | Page 15 General of projects exceed budget of projects are delayed of commissioning time is used to identify and eliminate problems with controls, protocols, and integration Digital Twins reshape machine building Digital Twin and smart manufacturing transform traditional sequential project workflows Design Engineering Classic workflow: Commission Mechanics Electrics Development follows a sequence step by step The domains push their results to the PLC programmers Program testing with all domains starts with the physical commissioning Pneum atic & Fluids Programming & Testing Machine Prototyping & Assembly VS. Design Commission Engineering Mechanics Virtual commission-driven workflow: Faster time to market Electrics 50% Pneum atic & Fluids Digital Twin / Virtualized Machine Reduced quality costs Programming & Testing Machine Assembly Commissioning time reduction Property of Schneider Electric | Page 16 General 20% 60% Development follows an iterative and agile approach The digital twin pulls progress from the domains Continuous integration and validation of the domain’s interoperation Deliver value incrementally in a collaborative manner & reduce uncertainties before doing the physical commission Agile methodology facilitates complex machine development Use cases throughout the complete development phases Design Virtual Design + Concept assessment during early design phase + compare solution concepts based on reached and targeted KPIs + Experimentation & proof of concept at minimal cost Engineering Virtual Engineering + Detailed collision and movement path visualization analysis + Integration of engineering domains in one validation model + Test performance at the mechanical edge Commission Virtual Commission + Visualize and simulate product processing + Validation of multi-domain system design ensuring high-quality integration and performance + Set up as many virtual machine instances as needed to parallelize workstreams Limit uncertainties before proceeding with a design Increased flexibility due to virtual design and testing Increased SW quality, test coverage Improved predictability of (non)achievable results early on Proof of design with virtual control testing and Improved design quality Reduced uncertainties and business risk at the physical commission Confidential Property of Schneider Electric | Page 18 General Digital Twin features enable more use cases Use cases throughout the complete machine lifecycle Sales Presentation Remote Operator Training + Demonstrate products and performance with a 3D-model based digital twin + Showcase in the customer‘s environment + Easily tailor demonstration with options and addons (Moving from 2D to 3D presentation) + Train operators remotely & virtually on a 3D-model based digital twin + No need to bring trainer, operators, asset together + no risk of damage + Extended training content with injected errors and malfunctions including troubleshooting Remote Maintenance Support + Guide local staff remotely on a shared virtual machine model + Connect your vendor’s experts remotely with staff on premise + Guide local staff in maintenance and troubleshooting tasks Limited risk of misunderstandings and complaints 24/7 availability to train on demand Be at the center of your customer’s operations Bring the virtual machine to the customer Automate tracking of trained operators Connect operator login with training status Deliver best in class After-Sales Design Engineering Commission Confidential Property of Schneider Electric | Page 19 General Operation Maintenance Digital Twin features enable more use cases Use cases throughout the complete machine lifecycle Performance Optimization + + + + Asset Health Monitoring Detect deviation from expected behavior during operation Use the DT as training agent for AI powered analysis Simulate production before executing it Gather real-time data from a fleet of operational machines + Tailor maintenance activities in addition to regular activities + Real time component health monitoring and health check of critical components + Predictive maintenance to prevent a possible breakdown before it actually occurs Continuous analysis operational data Increased uptime Optimize machine operation on the fly Reduction in downtimes and maintenance costs Flexibly adjust machine setup and let it learn Valuable insights on the key components can ensure least or low bottlenecks and smooth operations Operation Maintenance Confidential Property of Schneider Electric | Page 20 General Remanufacture Real-world outcomes drive adoption in various fields Lifecycle Management Advanced Analytics and AI Integration Trend towards full life cycle coverage Modularization to tailor DTs to specific use cases along the life-cycle enabling better decision-making, process optimization, and scenario planning Trend towards self-learning Simulation and Optimization Enhanced simulation and optimization capabilities test different scenarios, improve processes, and optimize resource utilization Property of Schneider Electric | Page 37 Integration with IIoT and Edge Computing more granular data collection and analysis Distributed digital twinning based on use case specific needs (computational performance, latency, connectivity, …)