Advances in Automotive Production Technology – Digital Product Development and Manufacturing

The accurate prediction of stress in fiber-reinforced laminates through finite element analysis is of critical importance for the design of lightweight automotive components, as it allows for better prediction of damage such as delamination. However, standard (Reissner–Mindlin) shell finite elements, while efficient, consider only a reduced stress state and neglect the transverse normal stress. This contribution investigates the potential of higher-order 3D-shell elements to improve stress prediction in laminates. In order to assess the efficacy of higher-order 3D-shell elements, their results are compared with those obtained from standard shell finite element simulations and fully 3D solid element simulations. The findings demonstrate that the use of cubic 3D-shell elements can be beneficial for simulating laminated materials, as they lead to a more accurate stress prediction in comparison to standard shell finite elements. In particular, the shear stress and normal stress in thickness direction of the laminate can be predicted with a higher degree of accuracy.

Isogeometric Analysis (IGA) uses higher-order and higher-continuity spline basis functions known from Computer Aided Design (CAD) to describe the geometry and the solution field of the simulation model (mainly Non-Uniform Rational B-splines). This leads to a more accurate geometry description, a smooth solution field and therefore superior simulation properties compared to traditional Finite Element Analysis (FEA). Using the same geometry description for CAD and IGA also speeds up the modeling process for the simulation. Real components consist of multiple sheet metal parts connected by point-like joints (spot-welds, rivets, screws). These joints significantly influence the component behaviour under crash load. Therefore, their properties must be accurately described within the component simulation. Due to minimum time step requirements a detailed modelling of every single joint is impossible. Thus, substitute models are used in component simulations, which describe the joints behaviour based on constrained conditions between the joining partners. However, these substitute models are developed for traditional FEA. To enable the application of IGA to vehicle simulations, this paper investigates whether existing constrained-based substitute models (e.g. *CONSTRAINED_SPR3 spotweld elements) can be combined with isogeometric shell models without further modifications. Therefore, specimen and component tests are simulated with IGA and FEA. The simulation results are compared to each other as well as experimental test results. It can be shown that IGA achieves a very good agreement with the experimental results, with a prediction quality comparable to the traditional FEA. This allows a straightforward replacement of existing FEA shell components with their IGA counterparts in vehicle simulations.

The automotive and aerospace industries have increasingly relied on Finite Element (FE) simulations to optimize their forming processes and reduce the costs and time associated with physical prototyping. However, ensuring the accuracy of these simulations in predicting actual forming defects remains a critical concern. To address this issue, a comprehensive convergence analysis, meticulously conducted, systematically compares the induced forming defects in FE-based simulations, point cloud representations, and actual formed parts. The study involves utilizing FE simulations to model various forming processes, capturing intricate details of formability, material-tool interactions, and varying key simulation parameters, such as material and process parameters. The study aimed to assess the sensitivity of the FE simulations to these parameters and to identify convergence criteria that lead to results closely resembling the actual formed parts and point cloud data. To achieve this, convergence studies were conducted to systematically vary these parameters and compare the simulated results against point cloud data obtained through advanced scanning technologies, providing a high-fidelity representation of the formed components. The study’s comparative analysis included a detailed examination of common forming defects such as wrinkles, bridging, and gaps, employing quantitative metrics to measure the deviation between the simulated, scanned, and actual formed parts.

Combining conventional with additive manufacturing processes opens up new possibilities for the flexible, economical production of complex automotive components and structures with varying geometries. This hybrid manufacturing approach can be realized by 3D-Printing functional structures onto pre-existing parts and surfaces with Robotic Screw Extrusion Additive Manufacturing (RSEAM). RSEAM combines the dexterity of industrial robotic arms with the productivity and extensive material range of a screw extruder to provide a highly flexible additive manufacturing process. This paper addresses a critical challenge in hybrid manufacturing: creating suitable manufacturing programs for printing on irregular, non-planar surfaces, that deviate significantly from ideal CAD input. Deviations are classified as macro or micro based on their magnitude relative to the desired printing layer height threshold. We propose a slicing and path-planning algorithm that adapts ideal paths to scanned surfaces and generates optimized trajectories to handle both macro and micro deviations via distinct methods. Macro deviations are handled by offsetting the scanned surface to create layers, smoothing to remove micro deviations and morphing each layer from the ideal CAD onto them. Micro deviations are then accounted for in the first layer by calculating the local non-uniform layer height and locally adjusting the robot velocity. The resulting smooth robot trajectories ensure accurate local bead geometry and correct over-all part geometry adaptation. The algorithm’s output is validated and optimized within a digital twin of the manufacturing system to avoid collisions and singularities. The implemented algorithms are validated and demonstrated on a real-world automotive application.

We propose an architecture for grasping on the move in order to facilitate pick and place operations while a mobile manipulator’s base is in motion. Our architecture eliminates the need for the robot to stop and localize in order to perform a pick and place operation which improves efficiency in setting such as industrial automation.Our framework applies a two-level planner where a base planner is used generate to trajectories for the base, and an arm planner is used to obtain a grasping trajectory given the planned motion of the base. We propose methods for combining and synchronizing these trajectories along with an online method for adjusting the trajectory of the arm to account for any deviation or error in the motion.We show application to a mobile manipulator robot, which consists of a UR5 arm mounted on a MiR mobile base. We show application to kitting and packing tasks in an industrial setting. Our results demonstrate that the robot is able to reach specified end-effector positions required to perform grasping while the base of the robot is in motion.

Automated sheet metal handling in the automotive industry using robot manipulators is a standard in modern production. However, the desire of automotive companies to speed up the production process on the assembly line and at the same time to reduce expensive hardware components poses new challenges for robotics. Excessively rapid movement of flexible parts or sheet metal can either lead to its plastic deformation or increase the decay time of the vibrations to such an extent that it is necessary to wait before the part can be further processed, for example by welding. The traditional approach to solving these problems is to add fixing points for the part and/or to allow for waiting times at the end of the robot movement to ensure that the sheet metal vibration subsides. This means that more effort than necessary has to be put into the hardware setup or a poor cycle time has to be accepted. In our work, we propose to improve virtual commissioning systems by conducting a deeper analysis of the dynamics of sheet metal parts during its movement by the robot. To achieve this, a three-step optimization strategy is proposed. The first step is a structural transient analysis of the thin metal part under disturbances that arise during its movements. The second step is the creation of a substitute model, which is trained on the basis of the data obtained in the first phase and considerably reduces the computing time required compared to the time needed for a finite element method (FEM) simulation. Subsequently, this is used to optimize robotic handling efficiency by optimizing the robot trajectory. By addressing the challenges posed by sheet metal dynamics, enhanced process control, reduced cycle times, and ultimately, improved manufacturing outcomes are anticipated.

In manufacturing processes involving industrial robots, collisions are a significant risk due to the high number of degrees of freedom. This is particularly the case in complex processes such as wire arc additive manufacturing, where metal is printed using a welding unit mounted on the robot. Simulations can be used before the operating phase to identify potential collisions between the industrial robot and itself, the workpiece clamping, or the workpiece. Even after intensive testing, a collision can still occur during the operating phase if any environmental condition differs from the simulation. This could be due to a change in motion caused by actual measured values or the modification of parameters by humans or an adaptive system. This work presents a concept for online real-time collision avoidance in robotic wire arc additive manufacturing. The concept is based on adopting simulations from virtual commissioning for the use in the operating phase. The concept is then realized with the help of a real industrial robot. The resulting solution’s novelty is the virtual commissioning-based online collision avoidance of wire arc additive manufacturing with industrial robots.

In industrial automation, simulation is extensively used during the engineering phases, but only rarely during the operational phase. The simulation of a manufacturing system in parallel to its operation, with the simulation system integrated into the mechatronic system, is known as online simulation and is beneficial for system monitoring, diagnosis, predictive analyses, decision support or online optimization. This paper focuses on the implementation of real-time online simulation at field level, reusing virtual commissioning simulation models. Main architectural aspects are the integration of the simulation system towards a control system and the management of the data distribution within a multi-functional edge device or control system. An exemplary realization in the form of a demonstrator is presented, showcasing the successful implementation of an online simulation using a comprehensive virtual commissioning model of a servo drive. The findings of this research contribute to the aim of using extended virtual commissioning simulation models in the operation phase of manufacturing systems and can be seen as indication for further realizations of online simulation systems.

In the context of deploying robust solutions in production environments, evaluating the performance of different machine learning models on a specific application is crucial. While common metrics like the mean square error offer a concise summary of model accuracy, they may overlook nuanced differences in prediction performance. In this paper, we propose a visual comparison method based on analyzing the distribution of residuals across multiple machine learning models. By sorting, visualizing, and defining a cutoff value, we aim to provide machine learning engineers with a comprehensive yet practical approach to assess model performance, enabling informed decision-making for real-world applications. The corresponding code is published on GitHub.

Resistance spot welding is a widely used joining method in automotive manufacturing, known for its efficiency and reliability. With growing regulatory scrutiny on vehicles’ carbon footprint, the body-in-white faces higher weight requirements. Therefore, aluminum alloys, particularly 5xxx and 6xxx series aluminum, are increasingly favored over steel due to their strength-to-density ratio. Despite this shift, research on the resistance spot welding of aluminum in large-scale production remains limited compared to steel. Furthermore, with the advent of big data and industry 4.0, the development and optimization of aluminum resistance spot welding may be addressed as a machine learning problem. In this study, we establish a first baseline model through machine learning and big data to predict the nugget diameter. The model is trained on tens of thousands of weld spots for both mentioned alloys across a wide process parameter range. Moreover, we provide a workflow for identifying the optimal hyperparameters for a suitable training algorithm. We identify a reasonable algorithm, determine the most important process parameters, and visualize their effects with shapley additive explanations. The performance of the baseline model serves as a benchmark for future machine learning models and provides a cornerstone for the experimental design decisions within the field. To ensure the reproducibility of results, we make available our baseline models and code on GitHub ( https://github.com/JanAlexanderZak/SCAP_2024 ).

Fuel cells are an environmentally friendly alternative to combustion engines since they use hydrogen to generate electric power.To assure the fulfilment of the lifetime targets, extensive aging tests are required. In order to reduce the amount and costs of these tests, it is necessary to use a predictive simulation model.A performance model has been developed by ZSW using Matlab®. This model is now being extended to include the degradation model.There are many processes which cause degradation in the cell components. The focus of this work is on the electrochemical aging behavior of catalyst layer, including also first aspects of carbon corrosion and dehydrophobization effects.The catalyst consists of the support material carbon and the catalyst material platinum. On the surface platinum and carbon oxidation and corrosion occurs, Pt ions dissolve and move into the membrane, particles detach and agglomerate. The result of these mechanisms is the reduction of the electrochemical active surface area which leads to performance loss. Additionally, the reduced ionomer coverage of the catalyst and carbon corrosion results in a decreasing contact angle, leading to dehydrophobization and thereby to a limited O2 mass transport.The degradation is calculated using an ODE system based on the particle radii and size distribution for every time step. At the beginning, the size of the particle radii and the distribution are defined. The change in these two variables is then calculated in the ODE system and the active surface area of platinum is computed as one influencing aging factor.

In today’s rapidly changing manufacturing environment, it is increasingly important to have dynamic process reconfiguration through adaptation control mechanisms. Digital Product Passports are also being developed and will become mandatory, such as for batteries in the European Union by 2027. Although this may initially require additional work, it can lead to synergies from the information obtained. By using $$\mathrm {CO_2~}$$ CO 2 values, e.g., from a Digital Product Passport of components, you can optimize the $$\mathrm {CO_2~}$$ CO 2 content of the end product to achieve a specific target and improve competitiveness. This can be achieved by adapting processes, such as choosing between high or low dynamics to influence overall energy consumption. A framework is necessary to extract specific information from DPP and make decisions for adaptation. This paper presents a framework architecture based on OPC UA and the AAS, accompanied by an illustrative example of a battery-packing handling process. The speed of the packing process is determined by the energy consumption values of the individual cells to reduce the total energy consumption value of the battery.

Due to flexibility requirements, the strictly horizontal communication of the automation pyramid is converging. This includes communication at the field level. Due to the long lifetime of machines in manufacturing, it is important to support converged communication for existing fieldbuses such as EtherCAT. This enables the implementation of brownfield approaches for gradually implementing converged networks in existing plants. EtherCAT is a widely used industrial Ethernet-based fieldbus protocol for communication between programmable logic controllers and field devices. This work analyzes the tunneling concept of EtherCAT through a Time-Sensitive Networking (TSN) network from the literature and contributes an empirical evaluation based on a test setup with multiple EtherCAT networks and EtherCAT slaves. The tunneling of EtherCAT through TSN based on Virtual Local Area Networks (VLANs) is demonstrated to be a viable option, allowing the utilization of existing EtherCAT devices without the necessity for adaptation. A comparison of the Linux features SO_TXTIME and a simple RAW_SOCKET reveals that both introduce jitter, which is compensated by the EtherCAT slaves to a few microseconds.

In the domain of reconfigurable production systems, Cyber-Physical Matrix Production Systems (CPMPS) are recognized for their advanced levels of operational flexibility. Given the inherent flexible material flow, these loosely coupled systems are characterized by dynamic interdependencies and rapid changes in order sequencing and allocation. This leads to major challenges in production flow control including the emergence of instable behaviors decreasing robustness and threatening overall performance.Traditional methodologies for assessing and enhancing the reliability and ensuring the robustness of the system do not tackle the dynamic behavior of re-configurable production systems. Due to rigid probabilistic assumptions, efficiency decreases and reasoning in fault propagation is not apparent. For this reason, dependable systems engineering embraces formal descriptions of the systems’ dynamical behaviors and continuous monitoring of system properties.This paper proposes the application of distributed artificial intelligences in the form of holonic multi-agent system (MAS) that integrate the concepts of dependability as part of the system design. Multi-level monitoring of state properties and fault-tolerant control mechanisms are used to minimize deviation between the modelled and observed behavior, therefore ensuring robustness and securing the system’s intended operation. The presented framework demonstrates feasibility by first implementations of dynamic interaction mechanisms for subsidiary decision improving makespan while remaining flexible.

In automated production lines, process automation devices with high individual availability are rigidly interlinked. The resulting risk of interlinking losses is minimized by decoupling buffers, offering more flexibility by the cost of additional space. Instead, flexible production concepts with more degrees of freedom could be enabled by applying intelligent control systems.To optimize and control industrial autonomous production systems with different classes of changeability, IT system architectures are required that can support strategic design as well as tactical and operational control decisions. Agent-based modeling and simulation can be used to implement the concept of a digital twin for applying intelligent planning and control. The findings on the interactions of individual system components can support planners in the design of production systems and can also be used as a basis for optimizing decision-making behavior in operations.In this paper a simulation study is conducted, in which an agent-based system architecture is applied to control an automated production process in a body shop embracing the different degrees of freedom for altering scenarios. The results show a robust production by enabling a different production concept with reconfigurable static robot cells connected via flexible mobile robots. Whilst reducing the number of production resources and buffer capacities in comparison to rigidly linked scenarios, productivity could be improved. In addition, the integration of the reconfigurable system with more flexibility into an automated production line increases the availability and thus the productivity of the overall production line.

Measuring the resilience of industrial production facilities during the production is very complex due to the variety of possible influences and interactions. In addition, production systems are characterized by a dynamic environment with increasing adjustments and changes. A metric for measuring the resilience of individual production facilities during usage phase is intended to help overcome these challenges while meeting requirements such as applicability and comparability. The objective of the metric focuses on the evaluation of measures to remedy resilience relevant events. For this purpose, classified events are considered as a total quantity of changes, adjustments and disruptions together with the measures taken, each in isolation with regard to the productivity of a production facility. An adapted OEE approach with the three weighted influencing factors of quality, performance and availability is developed as the basis for evaluating productivity. This approach evaluates quality based on the direct and indirect costs incurred for rework and post-production. Performance is determined by the cycle time of the facility. The time share of value-adding processes in the period under review is evaluated as a measure of availability. Including the time parameter, which is recorded alongside the resilience event, the metric results in a comparable resilience value. By classifying the event in terms of speed of occurrence and degree of awareness, it is possible to derive the characteristics of adaptability, innovation capability, robustness and improvisation capability of a production facility. In addition, the resilience value determined enables further analysis and the application of resilience strategies.

In an era characterized by the rapid evolution of consumer preferences and diminishing product life cycles, the significance of reconfigurable manufacturing systems (RMS) has surged. For businesses, swiftly adapting to these fluctuating demands is imperative. Conjoint analysis has emerged as a prevalent method for capturing consumer preferences within market research. Subsequently, to delineate an optimal product-line that aligns with these preferences, product-line optimization strategies, such as profit maximization models, are employed. These models comprehensively consider various costs associated with the product line. RMS endeavors to minimize expenses, duration, waste, CO2 emissions, and other objectives within the production process through a multi-objective framework. This study introduces a novel approach by integrating a profit maximization model from product-line optimization with the cost function of RMS, facilitating a multi-objective optimization framework that adeptly accommodates customer preferences within RMS. While not aiming to delineate the Pareto frontier, this paper elucidates the potential implications of the new framework on the configuration of the Pareto frontier. Additionally, it provides a comprehensive overview of RMS applications in the automotive industry.

This study investigates the effectiveness of interaction modalities for working within 2-dimensional and 3-dimensional content in autonomous vehicles (AVs). In a Virtual Reality (VR) simulation, three interaction concepts - space mouse, gesture control, and touchpad - are evaluated independently and comparatively based on task completion, user feedback, User Experience Questionnaire (UEQ), and NASA Task Load Index (TLX) assessments. The results show that the space mouse concept is positively received for both 2D and 3D interactions, with high task completion rates and reduced task times. Gesture control is favorable when executed intuitively, while the touchpad performs well in 2D but struggles in 3D due to spatial disparities and technical limitations. UEQ and TLX results align with qualitative findings. Consequently, a hybrid system integrating gesture control and space mouse input is developed to address limitations and enhance user experience within autonomous vehicles.

Looking towards a possible future in which fully autonomous vehicles are commonplace, it is imperative to already begin evaluating vehicle concepts. As of today level 5 autonomous vehicles are still under active research, mainly with a more technical perspective. Therefore evaluations often take place in virtual worlds, crafted for this specific use-case. However, stationary simulations fail to adequately represent the dynamic movements of a vehicle, which has inherent drawbacks, particularly a low sense of immersion and the development of motion sickness. To address this issue, this paper proposes the integration of Holoride technology, which leverages OpenStreetMap data and real-time sensor feedback from a vehicle to create a dynamic environment that closely mimics the real-world driving experience in Virtual Reality (VR). The hypothesis is a reduction in the risk of motion sickness and an increase in the perceived immersion. To test this hypothesis, a dynamic simulation is developed and evaluated in a study involving 18 participants, comparing it to a static simulation across various metrics including motion sickness, immersion and overall user experience. While the results did not show a noticeable reduction in motion sickness in the dynamic simulation, potentially attributable to technical inaccuracies of the holoride technology, there was a significant increase in perceived immersion and presence. Moreover, participants consistently rated the user experience higher in the dynamic simulation compared to the static counterpart.

The Industrial Internet of Things (IIoT) generates vast amounts of data, often unstructured and heterogeneous. Processing and analyzing this data to gain insights and drive industrial automation requires sophisticated tools. This paper investigates the potential of various platforms for handling unstructured IIoT data, focusing on a qualitative comparative analysis of Node-RED, a visual programming tool, against established industrial solutions such as Apache NIFI, KNIME Analytics Platform, and Microsoft Azure Logic Apps. By evaluating these tools, we highlight their respective capabilities and limitations in managing unstructured data. Through this comparative analysis, we demonstrate the distinct features and performance of each tool and discuss their applicability in IIoT data management. The study shows that Node-RED is the most effective tool for handling unstructured IIoT data. Accordingly, we illustrate its use for a specific use case, i.e., value stream analysis.

Today’s production is characterized by shorter innovation cycles, leading to a highly dynamic shop floor environment. To tackle the need for fast and flexible adaptation, it is aimed for a complete digital integration of machines and a comprehensive data management. Benefiting from the vast amounts of heterogeneous data collected requires a thorough understanding of their dependencies and correlations as well as a human-assisted, yet automated analysis process. Still, up to this date, the realization of this vision proves to be very challenging in practice and requires a myriad of different new methods and technologies.Therefore, the scope of this paper is to provide an outline of how such an end-to-end connectivity can be achieved along the real-world example of a lightweight construction process. As such, it is shown how its machines are upgraded and connected to the network in order to allow for collection of time series data using a central agent. A context provider is introduced, with which external data sources, such as wearables, can be integrated. After collecting and merging the data, its storage and sharing are managed using a Data Lakehouse. The data is analyzed by using a human-assisted feedback loop, effectively integrating expert knowledge into the learning process. Using the so-gained knowledge, optimization of the shop floor, intralogistics and IT infrastructure can be achieved.Preliminary results from parts of this loop already in operation suggest significant potential benefits, paving the way for more comprehensive integration across production systems.

This paper presents an AI-based system for improving cycle time measurement in industrial environments, leveraging YOLOv8 for object detection and ByteTrack for tracking. Our non-invasive approach analyzes video from an Azure Kinect camera to calculate cycle times by detecting objects and monitoring their state changes. Tested at the University of Applied Sciences Kempten’s demo plant, the system showcased high accuracy against ground truth data, highlighting its potential to enhance production line monitoring and efficiency significantly. This work contributes to industrial automation by offering a real-time, accurate method for cycle time analysis, promising substantial advancements in manufacturing process optimization.

Object detection is crucial in many industrial computer vision applications. However, the reliance on manually annotated data prevents a cost-effective deployment of supervised models in domain-specific industrial tasks.This research work presents an evaluation of the real-world performance of leading algorithms in the context of industrial multi-class object detection while being trained solely on synthetic data. We train widely-used models on an order-of-magnitude less synthetic industrial data than the current state-of-the-art and demonstrate a mAP@50-95 of 75% under high-variability environments. Our work also offers an ablation study to narrow the sim-to-real domain gap based on context-aware identification of synthetic features that contribute the most to closing the sim-to-real gap. We show that by employing guided domain randomization based on low-level and on semantic contextual features, it is possible to reduce the amount of required synthetic images by a factor of three while affecting mAP@50-95 on real data by only 2%.

Machines operating in a production environment inevitably undergo changes throughout their usage life span, especially due to environmental influences, modifications or wear and tear. This introduces complexities in leveraging machine learning to monitor their condition for maintenance purposes. Depending on the significance of the changes, the performance of the condition monitoring system could degrade drastically. Especially, if changes occur frequently, retraining the system from scratch is not feasible.Therefore, we investigated active learning and self-training as adaptation strategies to enhance the robustness of the condition monitoring process. We evaluated both strategies using a demonstrator consisting of several electric motors and a single vibration sensor. The classification task is to identify, which motors are running based on their superimposed vibration data. We emulated different changes occurring in typical production environments and rated them as small, medium and large scale changes.For a comparative analysis of both strategies, we trained a benchmark model based on a convolutional neural network and evaluated the performance of active learning and self-training for the different changes in relation to this reference model.The results indicate that employing both active learning and self-training in a production environment to adapt the model can enhance its robustness. Self-training is the preferred option for small changes, as it adapts the model without the need for user interaction. For medium and large scale changes, on the other hand, self-training can fail, while active learning is a sensible strategy, despite the operator having to label some of the data.

The automation of wire harness handling and installation in the automotive industry presents a challenge due to the inherent flexibility of cables, the high variance in wire harnesses and plug combinations, and the intricate spatial configurations required for accurate installation. Addressing this challenge requires the integration of sensors for accurate pose estimation with high-dexterity robotic systems. This work introduces a novel approach to automate the process of grasping of wiring harnesses for autonomous installation using a robotic arm. The methodology encompasses several stages. Initially, a multi-stereo camera setup creates a high-accuracy representation of the working area. Next, a deep learning model predicts a spline representing the segment with the biggest connector attached to it for 6D grasp pose estimation. The final stage uses a skill-based robot program to perform the grasping of the wiring harness, which is evaluated using 50 random configurations inside a bin. As a result, the proposed solution achieves an accuracy of 82% of successful wiring harness bin picking grasps, where success is defined when the result is that a specific connector on the wiring harness is in a predefined spot after grasping. Future work will use another robot to grasp the connector from the first robot to install it in an automotive demo door using reinforcement learning.

Advancements in technology and numerical methods have shifted from slow, resource-intensive software to faster predictive solutions powered by artificial intelligence (AI). An exemplary case is the analysis of interference fit connections between a cylindrical shaft and hub, which has the potential to redefine optimal design, minimizing stress and maximizing torque transmission. Traditional experimental analysis using Finite Element Method (FEM) simulations is undeniably time-consuming, inefficient, and complex, thus necessitating the deployment of AI as a pivotal tool in industrial applications. This paper unequivocally introduces a cutting-edge technique that harnesses two powerful AI approaches: Supervised Learning and Reinforcement Learning. The Reinforcement Learning approach expounded in this paper impeccably predicts the shaft-hub geometry set, eliminating the need for iterative simulations and drastically streamlining the optimization process. In order to address this challenge, a Supervised Learning model is rigorously trained using limited data obtained from experimental structural analysis. Subsequently, the predictions from this model serve as the environment for the Reinforcement Learning (RL) algorithm. The customized environment in Reinforcement Learning ingeniously employs the model to refine predictions by adjusting the input parameters for different geometric sets through respective actions on the environment.

The design of fibre composite parts spans across various design phases, starting with requirements collection and decomposition, going on with geometry definition, mechanical calculations, optimization of fibre arrangements, production planning and so on and ending with manufacturing. Typically, changes in one design phase may affect all subsequent phases or even affect the entire design process. Graph-based design languages (GBDLs) offer a unified and consistent digital data model capable to store and propagate relationships of design objects via linked graph nodes, thereby expressing the different mutual dependencies for design modifications within the model. By defining the contents of the knowledge domains in a vocabulary, rules and a rule sequence, the entire design process can be described and executed automatically. This paper shows how GBDLs can be applied to the design process of fibre composite parts by using them to automatically generate a FEM simulation for a given geometry and then using the result to model and plan the production sequence for the created composite part.

The importance of sustainability continues to grow. Considering the entire life cycle of a product is of great importance in evaluating and optimizing sustainability criteria. The greatest impact on sustainability is in the early phase of product development, but the information about the developed system is low. However, resolving the trade-offs between economic, environmental, and technical criteria requires information about the product over its entire life cycle. Many options need to be evaluated, especially in the early stages of product development. It is time-consuming to create detailed simulation models for each option and each stage of the life cycle. Therefore, the knowledge about the product in the phases of the product development process was analyzed, and measurable sustainability indicators for the phases were identified. Furthermore, the necessary models for the evaluation of the sustainability indicators are selected. Based on these findings, the paper presents a framework for the development of sustainable handling systems, which are part of nearly every automated production system, across all life phases. Starting from the requirements of the handling system, this framework helps to identify suitable solutions, evaluate them, and optimize them with respect to their sustainability score.

Environmental performance is increasingly being emphasized during the design phase. Although the lack of quantity and quality of data leads to inaccurate environmental results in the early stages of design, the ability of life cycle assessment (LCA) to influent environmental performance of the product is significant during this phase. This article aims to evaluate the data uncertainty of LCA during the design phase, integrate the data quality assessment results into the sustainability-oriented development process (NEP), and provide more accurate environmental influence information in the early stages of the design process to avoid major changes later. To that purpose, the life cycle inventory (LCI) data of hydrogen tank production based on pedigree matrix method is used to analyse the data quality of existing data in the early design stages. Methodologies are investigated which link these pedigree matrix methods to data of contribution analysis and probability distributions in order to predict data uncertainty in life cycle impact assessment (LCIA) results. A framework will then be proposed which allows to decide if further data updating or process refinement is needed based the calculated uncertainty for the predicted LCIA result. Through data quality analysis the reliability of LCA results can be improved at early design stage, and according to combination with the sustainable development proc ess LCA can identify the environmental influence during design phase, to avoid the large-scale change due to that in the late stages of design.

This paper provides an overview and examples of the potential, but also the challenges of Metal Additive Manufacturing in the future Circular Economy paradigm. The Circular Economy, aimed at reducing resource consumption, waste generation, and the overall carbon footprint of our industries, presents significant challenges to traditional, linear production systems. In this context, Additive Manufacturing emerges as a highly flexible technology with the potential to overcome many of these hurdles: The ability to reduce material usage through optimized lightweight and functional designs, the use of recycled materials, Additive Manufacturing-based repair and remanufacturing of tools and products or the optimization of production through hybrid manufacturing approaches (e.g. combinations of subtractive and additive manufacturing) are just a few examples.However, the lack of automated and digitalized production processes, complex process chains with potential health and safety concerns, and high demands on material and component qualification still pose significant hurdles to the widespread adoption of Metal Additive Manufacturing in the Circular Economy framework. In addition, the Circular Economy alone does not necessarily mean a lower carbon footprint over the entire product life cycle. On the contrary, depending on the specific process used, the carbon footprint of metal additive part production is often higher than in conventional manufacturing scenarios due to the high energy requirements of the additive process and long production times. Therefore, in this study, a comprehensive literature review is performed to outline the potential of Additive Manufacturing for Circular Economy principles, providing several beneficial examples of the so-called “R-strategies” covered with the help of Additive Manufacturing. In addition, an outlook is given on the development of new technologies to increase the potential of Additive Manufacturing to make an important contribution to circular production scenarios.

Hydrogen emerges as a pivotal element in achieving fossil-free transportation, offering a sustainable alternative to conventional fossil fuels and significantly reducing environmental impacts. This report outlines the development of a cutting-edge Type V high-pressure vessel employing carbon fiber composite material, aimed at tackling the challenges associated with liner less hydrogen pressure vessel. Current research focuses on addressing two major challenges, namely manufacturing complexity and leak tightness of the tank. Manufacturing complexity of the tank is thoroughly addressed, covering various removable mandrel technologies and further a novel integral mandrel technique is introduced, featuring a carbon fiber-reinforced polymer (CFRP) structure. Another significant challenge is ensuring the leak tightness of the tank without a polymer liner. The current article explains this issue in detail and discusses various solutions for a liner less, leak-tight tank structure. While Type V pressure vessels offer advantages in weight and storage capacity, their integration into automotive applications requires addressing manufacturing complexity, leak tightness, material compatibility, and cost-effectiveness. By overcoming these obstacles, the full potential of Type V tanks in hydrogen vehicles for the transportation sector can be realized, fostering sustainable mobility solutions.

Due to rising global temperature, mobility sector is witnessing a transition phase. It is essential to cut tailpipe emissions significantly to reduce rising global temperature and avoid changing climate patterns. In response to climate change, innovations in automotive sector, sustainable mobility solutions and increased public awareness for global warming is needed. While significant progress has been made in reducing greenhouse gas emissions, the transportation industry remained largely unchanged. Europe aims to achieve carbon neutrality by 2050, necessitating a major shift in traditional transportation practices. Conventional means of transport burning fossil fuel in urban areas are major contributors to emissions, posing environmental and health risks. Emerging technological advancements in automotive sector offer a promising way to improve urban environments, in line with Europe’s Green Deal and Paris Agreement. This study aims to suggest techniques to create more sustainable and liveable cities through affordable mobility innovations. The research explores the potential of integrating modern vehicle technologies with existing road infrastructure to reduce transport emissions.

Software-Defined Value Networks (SDVN) aim to increase flexibility, value-chain resilience through redundancy, scalability through standardized linkages and data formats as well as data consistency for sustainability requirements such as the carbon footprint, as opposed to current Global Value Chains. For a potential success of implementing SDVNs, the extent to which current research technologies are being utilized in companies and challenges of the companies regarding SDVNs need to be elaborated. This article presents a survey to identify industrial requirements, roadblockers and successfactors for implementing SDVN. The survey was conducted among industrial partners of ARENA2036, a research campus for future mobility and automotive production in Stuttgart, with two representatives, development and management, from each participating company The survey covers the interviewee’s experience with automation systems, the use of emerging technologies throughout the company’s value chains, perceived risks, sustainability factors, the potential of existing value chains, and requirements for further automation in SDVNs. The results of the survey are summarized and interpreted in the context of current technological developments, such as the Eclipse Dataspace Components, the Asset Administration Shell and solutions of Software-defined Manufacturing. They show the need of value-chain orchestration across companies including strategies for solving goal conflicts with e.g. sustainability requirements, as well as applicability of scientific developments and standardization processes to a broad industrial implementation.