Clemens Zimmerling
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Clemens Zimmerling has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Luise Kärger
Karlsruhe Institute of Technology (KIT)
5 shared publications Dominik Dörr
2 shared publicationsFrank Henning
2 shared publicationsBenedikt Fengler
Karlsruhe Institute of Technology (KIT), Institute of Vehicle System Technology, Karlsruhe, Germany
1 shared publication- Constantin Krauß
Karlsruhe Institute of Technology (KIT)
1 shared publication - Felix Frölich
Institute of Vehicle System Technology - Lightweight Engineering, Karlsruhe Institute of Technology (KIT), Karlsruhe, Germany
1 shared publication
Latest Publications
- Open Access
- 27 December 2024
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- Vol. 136 (3-4) , 1561-1576
A submodeling approach for efficient prediction of local temperature profiles in component-scale additive manufacturing
To solve the multiscale problem of additive manufacturing of large components by material extrusion (MEX), this work utilizes the methodology of submodeling and shows the necessary modeling steps to apply the methodology to the material extrusion process. The research enables the coupling of high-resolution models and process simulations at the component level. It allows the investigation of local effects such as crystallization effects or the resulting interface strength and its dependence on process variables in critical zones of a component, considering the component geometry as well as global and component-specific process conditions. The principle is verified numerically and validated experimentally, showing good agreement. In addition, different specifications for boundary conditions and submodel sizes are compared and evaluated. Variable time increments are used to apply the submodels with a computational time independent of the component size. In addition, sensitivity analyses provide information on the necessary temporal and spatial discretization of the submodel for the most accurate prediction of the temperature profile.
- 27 December 2024
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- Vol. 136 (3-4) , 1561-1576
Open AccessA submodeling approach for efficient prediction of local temperature profiles in component-scale additive manufacturing
To solve the multiscale problem of additive manufacturing of large components by material extrusion (MEX), this work utilizes the methodology of submodeling and shows the necessary modeling steps to apply the methodology to the material extrusion process. The research enables the coupling of high-resolution models and process simulations at the component level. It allows the investigation of local effects such as crystallization effects or the resulting interface strength and its dependence on process variables in critical zones of a component, considering the component geometry as well as global and component-specific process conditions. The principle is verified numerically and validated experimentally, showing good agreement. In addition, different specifications for boundary conditions and submodel sizes are compared and evaluated. Variable time increments are used to apply the submodels with a computational time independent of the component size. In addition, sensitivity analyses provide information on the necessary temporal and spatial discretization of the submodel for the most accurate prediction of the temperature profile.
- Open Access
- 15 June 2024
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 429
Physics-informed MeshGraphNets (PI-MGNs): Neural finite element solvers for non-stationary and nonlinear simulations on arbitrary meshes
- 15 June 2024
- Published by Elsevier in Computer Methods in Applied Mechanics and Engineering
- Vol. 429
Open AccessPhysics-informed MeshGraphNets (PI-MGNs): Neural finite element solvers for non-stationary and nonlinear simulations on arbitrary meshes
- Open Access
- 12 February 2022
- Published by Elsevier in Composite Structures
- Vol. 287, 115347
Optimisation of layup type and fibre orientation in continuous-fibre reinforced components via anisotropy analysis
- 12 February 2022
- Published by Elsevier in Composite Structures
- Vol. 287, 115347
Open AccessOptimisation of layup type and fibre orientation in continuous-fibre reinforced components via anisotropy analysis
- 8 April 2021
- Published by University of Liege
Deep neural networks as surrogate models for time-efficient manufacturing process optimisation
Manufacturing process optimisation usually amounts to searching optima in high-dimensional parameter spaces. In industrial practice, this search is most often directed by human-subjective expert judgment and trial-and-error experiments. In contrast, high-fidelity simulation models in combination with general-purpose optimisation algorithms, e.g. finite element models and evolutionary algorithms, enable a methodological, virtual process exploration and optimisation. However, reliable process models generally entail significant computation times, which often renders classical, iterative optimisation impracticable. Thus, efficiency is a key factor in optimisation. One option to increase efficiency is surrogate-based optimisation (SBO): SBO seeks to reduce the overall computational load by constructing a numerically inexpensive, data-driven approximation („surrogate“) of the expensive simulation. Traditionally, classical regression techniques are applied for surrogate construction. However, they typically predict a predefined, scalar performance metric only, which limits the amount of usable information gained from simulations. The advent of machine learning (ML) techniques introduces additional options for surrogates: in this work, a deep neural network (DNN) is trained to predict the full strain field instead of a single scalar during textile forming („draping“). Results reveal an improved predictive accuracy as more process-relevant information from the supplied simulations can be extracted. Application of the DNN in an SBO- framework for blank holder optimisation shows improved convergence compared to classical evolutionary algorithms. Thus, DNNs are a promising option for future surrogates in SBO.
- 8 April 2021
- Published by University of Liege
Deep neural networks as surrogate models for time-efficient manufacturing process optimisation
Manufacturing process optimisation usually amounts to searching optima in high-dimensional parameter spaces. In industrial practice, this search is most often directed by human-subjective expert judgment and trial-and-error experiments. In contrast, high-fidelity simulation models in combination with general-purpose optimisation algorithms, e.g. finite element models and evolutionary algorithms, enable a methodological, virtual process exploration and optimisation. However, reliable process models generally entail significant computation times, which often renders classical, iterative optimisation impracticable. Thus, efficiency is a key factor in optimisation. One option to increase efficiency is surrogate-based optimisation (SBO): SBO seeks to reduce the overall computational load by constructing a numerically inexpensive, data-driven approximation („surrogate“) of the expensive simulation. Traditionally, classical regression techniques are applied for surrogate construction. However, they typically predict a predefined, scalar performance metric only, which limits the amount of usable information gained from simulations. The advent of machine learning (ML) techniques introduces additional options for surrogates: in this work, a deep neural network (DNN) is trained to predict the full strain field instead of a single scalar during textile forming („draping“). Results reveal an improved predictive accuracy as more process-relevant information from the supplied simulations can be extracted. Application of the DNN in an SBO- framework for blank holder optimisation shows improved convergence compared to classical evolutionary algorithms. Thus, DNNs are a promising option for future surrogates in SBO.
- 1 September 2019
- Published by Elsevier in Composites Part A: Applied Science and Manufacturing
- Vol. 124, 105459
A machine learning assisted approach for textile formability assessment and design improvement of composite components
- 1 September 2019
- Published by Elsevier in Composites Part A: Applied Science and Manufacturing
- Vol. 124, 105459
A machine learning assisted approach for textile formability assessment and design improvement of composite components
- 1 January 2018
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 1960 (1) , 020042
A meta-model based approach for rapid formability estimation of continuous fibre reinforced components
Due to their high mechanical performance, continuous fibre reinforced plastics (CoFRP) become increasingly important for load bearing structures. In many cases, manufacturing CoFRPs comprises a forming process of textiles. To predict and optimise the forming behaviour of a component, numerical simulations are applied. However, for maximum part quality, both the geometry and the process parameters must match in mutual regard, which in turn requires numerous numerically expensive optimisation iterations. In both textile and metal forming, a lot of research has focused on determining optimum process parameters, whilst regarding the geometry as invariable. In this work, a meta-model based approach on component level is proposed, that provides a rapid estimation of the formability for variable geometries based on pre-sampled, physics-based draping data. Initially, a geometry recognition algorithm scans the geometry and extracts a set of doubly-curved regions with relevant geometry parameters. If the relevant parameter space is not part of an underlying data base, additional samples via Finite-Element draping simulations are drawn according to a suitable design-table for computer experiments. Time saving parallel runs of the physical simulations accelerate the data acquisition. Ultimately, a Gaussian Regression meta-model is built from the data base. The method is demonstrated on a box-shaped generic structure. The predicted results are in good agreement with physics-based draping simulations. Since evaluations of the established meta-model are numerically inexpensive, any further design exploration (e.g. robustness analysis or design optimisation) can be performed in short time. It is expected that the proposed method also offers great potential for future applications along virtual process chains: For each process step along the chain, a meta-model can be set-up to predict the impact of design variations on manufacturability and part performance. Thus, the method is considered to facilitate a lean and economic part and process design under consideration of manufacturing effects.
- 1 January 2018
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 1960 (1) , 020042
A meta-model based approach for rapid formability estimation of continuous fibre reinforced components
Due to their high mechanical performance, continuous fibre reinforced plastics (CoFRP) become increasingly important for load bearing structures. In many cases, manufacturing CoFRPs comprises a forming process of textiles. To predict and optimise the forming behaviour of a component, numerical simulations are applied. However, for maximum part quality, both the geometry and the process parameters must match in mutual regard, which in turn requires numerous numerically expensive optimisation iterations. In both textile and metal forming, a lot of research has focused on determining optimum process parameters, whilst regarding the geometry as invariable. In this work, a meta-model based approach on component level is proposed, that provides a rapid estimation of the formability for variable geometries based on pre-sampled, physics-based draping data. Initially, a geometry recognition algorithm scans the geometry and extracts a set of doubly-curved regions with relevant geometry parameters. If the relevant parameter space is not part of an underlying data base, additional samples via Finite-Element draping simulations are drawn according to a suitable design-table for computer experiments. Time saving parallel runs of the physical simulations accelerate the data acquisition. Ultimately, a Gaussian Regression meta-model is built from the data base. The method is demonstrated on a box-shaped generic structure. The predicted results are in good agreement with physics-based draping simulations. Since evaluations of the established meta-model are numerically inexpensive, any further design exploration (e.g. robustness analysis or design optimisation) can be performed in short time. It is expected that the proposed method also offers great potential for future applications along virtual process chains: For each process step along the chain, a meta-model can be set-up to predict the impact of design variations on manufacturability and part performance. Thus, the method is considered to facilitate a lean and economic part and process design under consideration of manufacturing effects.