Linz Center of Mechatronics (Austria)

Accurately modeling the viscosity of shear-thinning elastomer compounds is crucial for optimizing their performance in industrial applications. Classical models (Power Law, Carreau, Cross) cannot directly account for temperature, and when coupled with temperature models (Arrhenius, WLF, VTF), they become mathematically complex and computationally costly for simulations. This study introduces the Multi-Variable Implicit (MVI) viscosity model, derived using machine learning-based symbolic regression. It offers a simplified mathematical structure while maintaining high predictive accuracy. The MVI model implicitly incorporates both shear-rate and temperature dependence in a single expression, while also capturing zero-shear viscosity through a Newtonian plateau. This feature ensures numerical stability and yields physically meaningful results across a wide range of shear rates and temperatures. The model was trained on high pressure capillary-rheometer data of an elastomer (70–120 °C; 10-5000 1/s), corrected for entrance losses and non-Newtonian profiles, and extended with synthetic points from a Carreau-Arrhenius fit to include the low-shear plateau. On test data, MVI achieved R² = 0.99 while preserving finite zero-shear viscosity. Independent validation on different materials across wider ranges gave R² = 0.957–0.990 without retraining, confirming strong generalization. To demonstrate its practical applicability, capillary-flow CFD simulations were carried out using both the MVI and Carreau-Arrhenius models. While each gave accurate predictions of the pressure-flow response, the MVI model required less computational effort because it avoids repeated exponential calculations. These findings highlight the MVI model as a novel, efficient, and practical solution for viscosity modeling in polymer processing and other fluid flow applications.

The rise of global temperature, the reasoning of the International Panel for Climate Change, and the political calls for carbon neutrality urge society and its players for immediate action. Short-term thinking and its advocation need to be altered. One way out is the circular economy as it combines economic perspectives with sustainable actions. Yet, its implementation demands to break up current system boundaries by focussing on overarching business models and product design across company boundaries. Here, we give an outline of how these company boundaries can be overcome. First, motivation. Material scarcity and far-reaching regulations motivate cross-company business models, which in turn alter the priorities of the product properties. Second, new networks. For cross-company design, new eco-systems need to be found and new cooperations with new partners need to be set up. Third, dynamic design. With increasing scientific depth, sophistication and new technologies we need to have diverse expert know-how at hand – in a scalable manner. Fourth, individualisation; with growing complexity, individual problem statements change dynamically, so do potential solution trajectories. By deploying digital workflow orchestration, we have the ability to provide instantly tailored responses that resonate with the user. To address the challenges of our times, we need to rethink established societal and engineering systems and reassemble economic building blocks for new definitions and dimensions of economic benefits. We need to reorientate as a society, by resonating with the consequences of our actions. The proposed harnessing of technologies is contributing key feedback loops for this paradigmatic shift.

The discrete element method (DEM) is a highly accurate and versatile approach for modeling large-scale particulate and fluid-mechanical systems critical to industrial processes. Additionally, DEM offers integration with grid-based computational fluid dynamics, making DEM a key ingredient for the modeling of many multi-physics systems. However, its computational demands, driven by the multiscale nature of these systems, limit simulation scale and duration. To address this, we introduce NeuralDEM, a fast and adaptable deep learning surrogate that captures long-term transport processes across various regimes using macroscopic observables, without relying on microscopic model parameters. NeuralDEM is a deep learning approach scalable to real-time industrial applications. Such scenarios have previously been challenging for deep learning models. NeuralDEM will open many doors to advanced engineering and much faster process cycles. The discrete element method (DEM) is crucial for modeling complex particulate systems but is limited by high computational demands. Here, the authors introduce NeuralDEM, a deep learning surrogate that enables real-time simulations by capturing fine-grained particle dynamics using a neural field model, significantly advancing engineering applications and accelerating process cycles across industries.

Digital twins are sophisticated software systems for the representation, monitoring, and control of cyber-physical systems, including automotive, avionics, smart manufacturing, and many more. Existing definitions and reference models of digital twins are overly abstract, impeding their comprehensive understanding and implementation guidance. Consequently, a significant gap emerges between abstract concepts and their industrial implementations. We analyze popular reference models for digital twins and combine these into a significantly detailed unifying reference model for digital twins that reduces the concept-implementation gap to facilitate their engineering in industrial practice. This enhances the understanding of the concepts of digital twins and their relationships and guides developers to implement digital twins effectively.

In this paper, we present a pallet detection system utilizing the You only look once X (YOLOX) machine learning method to identify two different types of pallets in various environments. The sensor technology employed in this system is capable of capturing both RGB and depth images, providing a comprehensive view of the environment. YOLOX, an advanced object detection algorithm, is used to enhance the accuracy and efficiency of pallet recognition. Despite the challenges posed by a relatively small dataset available for training, our system demonstrates robust performance in both offline simulations and live operational environments. The presence of two different types of pallets further adds complexity to the classification process, highlighting the effectiveness of YOLOX in handling diverse scenarios.

The research project ALICIA – Assembly Lines in Circulation – will provide a marketplace platform that enables the reuse of second-hand assembly resources. Based on a formalized resource description and AI-based decision-making support, the platform suggests suitable second-hand resources for new assembly lines. For these resources, an Asset Administration Shell (AAS) is developed to integrate them into Industrial Internet of Things (IIoT) systems. In return, the legacy resources can be connected to their Digital Twins (DTs) while complying with data security standards. Further, ALICIA evaluates environmental sustainability aspects measured by Key Performance Indicators (KPIs) as well as worker-centric aspects, e.g., worker skills, when selecting second-hand equipment to reuse in new lines. ALICA also identifies potential ethical impacts and risks caused by the technologies, stakeholders, or the ecosystem's processes to be addressed in designing digital solutions. Besides ALICIA’s core services, developed within the project, the platform will be open to partners to offer their services in the context of a second-hand equipment market, such as equipment health state assessment, repair, and Life Cycle Assessment (LCA), among others. To evaluate the platform‘s business models, a value proposition analysis is conducted, showing how Robotics as a Service (Raas) concepts can address potential stakeholder challenges in the project.

Accurately estimating a sequence of latent variables in state observation models remains a challenging problem, particularly when maintaining coherence among consecutive estimates. While forward filtering and smoothing methods provide coherent marginal distributions, they often fail to maintain coherence in marginal MAP estimates. Existing methods efficiently handle discrete-state or Gaussian models. However, general models remain challenging. Recently, a recursive Bayesian decoder has been discussed, which effectively infers coherent state estimates in a wide range of models, including Gaussian and Gaussian mixture models. In this work, we analyze the theoretical properties and implications of this method, drawing connections to classical inference frameworks. The versatile applicability of mixture models and the prevailing advantage of the recursive Bayesian decoding method are demonstrated using the double-slit experiment. Rather than inferring the state of a quantum particle itself, we utilize interference patterns from the slit experiments to decode the movement of a non-stationary particle detector. Our findings indicate that, by appropriate modeling and inference, the fundamental uncertainty associated with quantum objects can be leveraged to decrease the induced uncertainty of states associated with classical objects. We thoroughly discuss the interpretability of the simulation results from multiple perspectives.