Alexander Kriwet, Fabian Urban, Dominik Er (Mercedes-Benz)
Markus Stommel (Leibniz-Institute of Polymer Materials Dresden)
Peter Middendorf (Institute of Aircraft Design University of Stuttgart)
The research's strong industrial relevance partly lies in its focus on short-fiber reinforced plastics, a widely used material, not just in automotive powertrain components. The authors’ innovative approach, leveraging a new test method for characterizing viscoelastic material data, opens new possibilities concerning the simulation of the vibration behaviour of these materials, which is critical for sound comfort. The validation of their results provided support for the reliability of their methods and findings. The paper's structure was praised for its clarity and coherence, making the complex concepts accessible and educational for readers. Overall, this paper presents a novel solution to an industry-relevant problem and provides validated educational content for its readers.
Malte Niehoff, Dieter Bestle (Brandenburg University of Technology) Philipp Kupijai (Rolls-Royce Deutschland)
This paper explores the optimization of aero engine design using surrogate models. Surrogate models replace the original complex design process, saving time and computational resources. However, they may mislead the optimization by offering non-viable designs, hence a classification model is used as an extra constraint to guide towards viable results. The researchers apply a regression-based bi-criterion optimization method, coupling an evolutionary algorithm with a classification surrogate to address this viability issue. The results indicate that surrogate-assisted optimization incorporating design viability is an effective approach, though further investigations are needed to incorporate more constraints and adaptive re-sampling strategies.
Marko Thiele, Martin Liebscher and Gordon Geißler (SCALE Gmbh)
This presentation showcased the possibilities for Simulation Data Management (SDM) tools to manipulate simulation runs in various physical domains and solved the problem effecting all engineering consultants in an elegant way. Instead of presenting snippets of highly confidential projects carried out for OEM-s, he illustrated the capabilities and methods on the example of LEGO® car models. This was awarded Best Presentation, owing to its stunning visuals, and the sound underlying technical background.
Philip Becht, Emre Kanpolat, Benjamin Marrant and Stefan Schmidt (ZF)
This interesting work on wind turbine development employs a comprehensive noise and simulation approach to minimize wind turbine noise. It received the Simulation for Sustainability award, due to its potential to impact the design of future wind turbines. The approach taken considers multiple aspects, including gear mesh excitation, vibro-acoustic transfer paths of the structure, and aero-acoustic noise from wind turbine rotors. The simulation platform used integrates with a library of drivetrain models and an optimization software for structural adjustments of drivetrain components. The study also covers optimization of gear excitation and structural dynamics to improve drivetrain noise, vibration, and harshness performance. The authors applied sensitivity and design-of-experiment studies for these optimizations. Overall, the researchers assert that this holistic approach enables the development of cost-effective, optimized, and well-integrated gearbox and drivetrain designs
Alexander Rogers, Laura Cannon, John Newman, Stephen Fay, Christopher Taggart, Alessandro Dimech, Chris Tilbury & David Munro (AWE)
Chris Metcalfe (Spurpark)
This study sought to understand the impact of large-scale explosions on storage tanks, using numerical analysis validated by scaled experiments due to the impracticality of full-scale trials. The numerical study explored the effects of various factors like tank geometry, fill level, and loading, revealing several key damage mechanisms primarily influenced by fill level.
A series of scaled physical tests, involving miniature tanks subjected to explosive charges, were conducted to confirm the identified damage mechanisms and provide data for model validation. High-speed video and 3D scanning were used to record the tanks' response and deformation. The results showed good agreement between numerical simulations and experimental tests, thereby enhancing confidence in the predictive modelling approach.
Markus Schatz (Ravensburg University of Cooperative Education) Tobias Klenk & Suwi Murugathas (newboxes GmbH)
In this research project, a novel approach is used in the additive manufacturing of fiber-reinforced plastics, which addresses the key challenge of load transmission between components. Metallic bushings are placed in critical load-bearing spots, designed for mechanical and thermal resilience and cost-effective placement during or after the manufacturing process.
In addition to enhanced mechanical properties, this method also enables easy repair by drilling out and re-gluing the bushing, if required. The authors use stress analysis to control the stress state near the metallic bushings.
Mariam Emara, Michael Balchanos, Burak Bagdatli, & Dimitri Mavris (Georgia Institute of Technology)
The authors here have developed a parametric decision-making environment that aids vehicle setup, focusing on the suspension system, using digital twin technology. The tool uses models of vehicle kinematic tests to calculate and visualise the results of different suspension setups, allowing for effective digital testing and tuning of vehicle options.
The decision-making platform, built in Python, allows users to select desired suspension parameters, while the backend uses surrogate models to predict the vehicle performance. An interactive user interface displays output plots and visualisation elements to provide insights about the performance of the chosen setup and allows comparison with a baseline configuration.
The authors demonstrated the tool with a use case of suspension tuning, highlighting its potential to quickly inform decisions without the need for expensive simulations or physical tests.
Todd Depauw, Alexandru Stere, Alan Byar, Sergey Fomin, John Dong (Boeing)
This presentation explored the application of Machine Learning (ML) for probabilistic material characterization in the aerospace industry, whilst adhering to regulatory certification requirements. The authors define "physics informed" machine learning as the inclusion of physical attributes and engineering principles in the ML training process.
They discuss studies conducted on compact-tension and compact-compression coupons to characterize composite material properties, considering test anomalies, and quantifying the uncertainty inherent in physical tests. They emphasize the importance of considering both physical test and numerical uncertainties in ML-driven predictions.
The authors conclude that with the right application of ML, the need for physical tests could be significantly reduced while still meeting regulatory standards.