Publications & Patents

2025

1. Enhancing Coronary Artery Calcium Scoring via Multi-Organ Segmentation on Non-Contrast Cardiac Computed Tomography

   Jakub Nalepa, Tomasz Bartczak, Mariusz Bujny, Jarosław Gośliński, Katarzyna Jesionek, Wojciech Malara, Filip Malawski, Karol Miszalski-Jamka, Patrycja Rewa, and Marcin Kostur

   arXiv preprint arXiv:2501.11428, 2025
2. SMO

   Automatic differentiation-based level-set topology optimization for noise minimization in 3D domains considering acoustic-structure interaction

   Andreas Neofytou, Thiago Rios, Mariusz Bujny, Stefan Menzel, and H Alicia Kim

   Structural and Multidisciplinary Optimization, 2025

2024

1. Highly Accurate Multi-vendor AI-based Algorithm For Coronary Artery Calcium Scoring

   Karol Miszalski-Jamka, Filip Malawski, Mariusz Bujny, Wojciech Malara, Jarosław Gośliński, Marlena Witkowska, Patrycja Purgoł, Jakub Nalepa, and Marcin Kostur

   Journal of Cardiovascular Computed Tomography, 2024

   Abs HTML

   Introduction: Artificial intelligence (AI)-based coronary artery calcification (CAC) scoring algorithm applied to cardiac computed tomography (CT) scans can standardize image interpretation and quantification as well as improve clinical efficiency and accuracy. Methods: An automated CAC scoring algorithm, based on the recognition of cardiac structures, was developed and trained using a dataset of 293 cardiac CT scans (Graylight Imaging, Gliwice, Poland). The algorithm was tested on independent 434 non-contrast cardiac CT scans (seven scanner models, four manufacturers: 383 Siemens, 19 Canon, 17 GE, 15 Philips). We compared automated CAC scores to manual expert reader assessments. CAC scores were categorized into four risk categories following SCCT recommendations. Results: In assessing 434 scans (mean age 62.0±10.6 years, 49.5% males), the AI-model detected CAC in 361 scans (82.7%), which was comparable to the human reader’s detection rate of CAC in 360 scans (82.5%) (p = 1.0). There was an excellent correlation between AI-model and manual CAC scores, both for total score (Spearman’s r = 0.99, 95% CI 0.99-1.0, p \textless 0.001) and individual coronary arteries (Spearman’s r = 0.91, 0.98, 0.92, and 0.98 for left main, left anterior descending, left circumflex, and right coronary artery, respectively). Bland-Altman analysis revealed minimal bias of 7.1 AU with 95% limits of agreement ranging from -96.6 to 110.7 AU. The model correctly classified 426 patients (98.6%) into the same CAC risk category as the human reader. Among 74 patients (17.1%) with a CAC score of zero, only 2 patients (2.7%) were reclassified with a non-zero CAC score by the AI-model. The Cohen’s kappa value for CAC score risk categorization was 0.98 (95% CI 0.97-1.0), indicating excellent agreement. Conclusions: The developed multi-vendor, fully automated CAC scoring algorithm demonstrated almost perfect concordance with assessments by expert reader. Leveraging AI, novel high-precision algorithms have the potential to significantly reduce the time required for repetitive tasks in clinical practice.

2. Coronary artery segmentation in non-contrast calcium scoring CT images using deep learning

   Mariusz Bujny, Katarzyna Jesionek, Jakub Nalepa, Karol Miszalski-Jamka, Katarzyna Widawka-Żak, Sabina Wolny, and Marcin Kostur

   2024

   arXiv:2403.02544 [cs, eess]

   Abs HTML

   Precise localization of coronary arteries in Computed Tomography (CT) scans is critical from the perspective of medical assessment of coronary artery disease. Although various methods exist that offer high-quality segmentation of coronary arteries in cardiac contrast-enhanced CT scans, the potential of less invasive, non-contrast CT in this area is still not fully exploited. Since such fine anatomical structures are hardly visible in this type of medical images, the existing methods are characterized by high recall and low precision, and are used mainly for filtering of atherosclerotic plaques in the context of calcium scoring. In this paper, we address this research gap and introduce a deep learning algorithm for segmenting coronary arteries in multi-vendor ECG-gated non-contrast cardiac CT images which benefits from a novel framework for semi-automatic generation of Ground Truth (GT) via image registration. We hypothesize that the proposed GT generation process is much more efficient in this case than manual segmentation, since it allows for a fast generation of large volumes of diverse data, which leads to well-generalizing models. To investigate and thoroughly evaluate the segmentation quality based on such an approach, we propose a novel method for manual mesh-to-image registration, which is used to create our test-GT. The experimental study shows that the trained model has significantly higher accuracy than the GT used for training, and leads to the Dice and clDice metrics close to the interrater variability.

3. Seeing the Invisible: On Aortic Valve Reconstruction in Non-Contrast CT

   Mariusz Bujny, Katarzyna Jesionek, Jakub Nalepa, Tomasz Bartczak, Karol Miszalski-Jamka, and Marcin Kostur

   In International Conference on Medical Image Computing and Computer-Assisted Intervention, 2024
4. AI-based algorithm for assessing coronary artery calcium score on contrast-enhanced cardiac computed tomography scans

   Karol Miszalski-Jamka, Filip Malawski, Jarosław Goslinski, Marlena Witkowska, Patrycja Purgol, Mariusz Bujny, Tomasz Bartczak, Wojciech Malara, Maciej Banach, and Marcin Kostur

   European Heart Journal, 2024
5. Prevalence of high-risk plaque features on cardiac computed tomography scans in patients with a coronary artery calcium score of zero

   Karol Miszalski-Jamka, Katarzyna Jesionek, Patrycja Purgol, Mariusz Bujny, Maciej Banach, and Marcin Kostur

   European Heart Journal, 2024
6. SMO

   Level set topology optimization with sparse automatic differentiation

   Andreas Neofytou, Thiago Rios, Mariusz Bujny, Stefan Menzel, and H Alicia Kim

   Structural and Multidisciplinary Optimization, 2024
7. Method for structural optimization of a design and cost of a physical object

   Mariusz Bujny, Hung Lin, Nathan Zurbrugg, Duane Detwiler, and Stefan Menzel

   2024

   US Patent 12,124,995

   HTML

2023

1. Appl. Intell.

   Evaluation of geometric similarity metrics for structural clusters generated using topology optimization

   Nivesh Dommaraju, Mariusz Bujny, Stefan Menzel, Markus Olhofer, and Fabian Duddeck

   Applied Intelligence, 2023
2. JMD

   Similarity-Driven Topology Optimization for Statics and Crash via Energy Scaling Method

   Muhammad Salman Yousaf, Duane Detwiler, Fabian Duddeck, Stefan Menzel, Satchit Ramnath, Nathan Zurbrugg, and Mariusz Bujny

   Journal of Mechanical Design, 2023

   Abs DOI HTML

   Abstract Topology optimization (TO) is used in the initial design phase to optimize certain objective functions under given boundary conditions by finding suitable material distributions in a specified design domain. Currently available methods in the industry work very efficiently to get topologically optimized design concepts under static and dynamic load cases. However, conventional methods do not address the designer’s preferences about the final material layout in the optimized design. In practice, the final design might be required to have a certain degree of local or global structural similarity with an already present good reference design because of economic, manufacturing, and assembly limitations or the desire to re-use parts in different systems. In this article, a heuristic energy scaling method (ESM) for similarity-driven TO under static as well as dynamic loading conditions is presented and thoroughly evaluated. A 2D cantilever beam under static point load is used to show that the proposed method can be coupled with gradient-based and also heuristic, nongradient methods to get designs of varying similarity with respect to a reference design. Further testing of the proposed method for similarity-driven TO on a 2D crash test case and a large-scale 3D hood model of a car body indicates the effectiveness of the method for a wide range of problems in the industry. Finally, the application of similarity-driven TO is further extended to show that ESM also has the potential for sensitivity analysis of performance with respect to the extension of design domain.

3. Sci. Rep.

   Learning hyperparameter predictors for similarity-based multidisciplinary topology optimization

   Mariusz Bujny, Muhammad Salman Yousaf, Nathan Zurbrugg, Duane Detwiler, Stefan Menzel, Satchit Ramnath, Thiago Rios, and Fabian Duddeck

   Scientific Reports, 2023

   HTML
4. Deep Learning Meets Particle Swarm Optimization For Aortic Valve Calcium Scoring From Cardiac Computed Tomography

   Jaroslaw Goslinski, Filip Malawski, Mariusz Bujny, Marcin Kostur, Karol Miszalski-Jamka, and Jakub Nalepa

   In 2023 IEEE International Conference on Image Processing (ICIP), 2023

   Abs DOI HTML

   Aortic stenosis is the most common primary valvular pathology requiring surgical or transcatheter intervention in developed countries. Quantification of aortic valve calcification with cardiac computed tomography (CCT) is used for assessment of aortic stenosis severity, disease progression and prediction of major cardiovascular events. The calcium deposits, however, commonly appear in different regions of the aorta and heart, leading to false-positive regions, and to an incorrectly calculated aortic valve calcium score. We tackle the issue of pruning such false-positive regions from CCT scans, and introduce a particle swarm optimization (PSO) algorithm for this task. In our approach, PSO optimizes the radius while benefiting from the evolved position of a sphere which would embrace those calcifications that are positioned near the aortic valve. Our experimental study, performed over 30 non-contrast CCT scans, showed that our results are in strong agreement with the experienced human reader, and indicate the potential of PSO in data-driven pruning of false-positive calcifications which are positioned in other parts of the aorta and heart. Additionally, PSO outperformed a geometricalbased approach for this task.

5. Optim. Eng.

   Topology optimization of periodic structures for crash and static load cases using the evolutionary level set method

   Hua-Ming Huang, Elena Raponi, Fabian Duddeck, Stefan Menzel, and Mariusz Bujny

   Optimization and Engineering, 2023

   Abs DOI HTML

   Assembly complexity and manufacturing costs of engineering structures can be significantly reduced by using periodic mechanical components, which are defined by combining multiple identical unit cells into a global topology. Additionally, the superior energy-absorbing properties of lattice-based periodic structures can potentially enhance the overall performance in crash-related applications. Recent research developments in periodic topology optimization (PTO) have shown its efficacy for tackling new design problems and finding advanced novel structures. However, most of these methods rely on gradient information in the optimization process, which poses difficulties for crash problems where analytical sensitivities are usually not directly applicable. In this paper, we present an effective periodic evolutionary level set method (P-EALSM) for the optimization of periodic structures. P-EA-LSM uses a low-dimensional level-set representation based on moving morphable components to parametrize a single unit cell, which is replicated in the design domain according to a predefined pattern. The unit cell is optimized using an evolutionary algorithm and the structural responses are calculated for the entire system. We initially assess the performance of P-EA-LSM using three 2D minimum compliance test cases with varying periodicities. Our results demonstrate that our approach produces solutions comparable to other state-of-the-art methods for PTO while keeping a low dimensionality of the optimization problem. Subsequently, we effectively evaluate the capabilities of P-EA-LSM in a crashworthiness scenario. This particular application highlights the significant potential of the method, which does not rely on analytical sensitivities.

6. Data-driven Evolutionary Optimization of eVTOL Design Concepts Based on Multi-agent Transport Simulations

   Sebastian Brulin, Mariusz Bujny, Tim Puphal, and Stefan Menzel

   In AIAA SCITECH 2023 Forum, 2023

   DOI HTML
7. A modularized level set topology optimization methodology for vibro-acoustic problems

   Andreas Neofytou, Jaeyub Hyun, Mariusz Bujny, Thiago Rios, Stefan Menzel, and Hyunsun A Kim

   In AIAA SCITECH 2023 Forum, 2023

   HTML
8. A method of calculating an aortic valve calcium score

   Mariusz Bujny, Jarosław Gośliński, Karol Miszalski-Jamka, and Marcin Kostur

   2023

   EU Patent Application No. EP4553761A1

   HTML

2022

1. IEEE TEVC

   CarHoods10k: An Industry-grade Data Set for Representation Learning and Design Optimization in Engineering Applications

   Patricia Wollstadt, Mariusz Bujny, Satchit Ramnath, Jami J. Shah, Duane Detwiler, and Stefan Menzel

   IEEE Transactions on Evolutionary Computation, 2022

   Abs DOI HTML

   Large-scale, high-quality data sets are central to the development of advanced machine learning techniques that increase the effectiveness of existing optimization methods or even inspire novel ones. Especially in the engineering domain such high quality data sets are rare due to confidentiality concerns and generation costs, be it computational or manual efforts. We therefore introduce the OSU-Honda Automobile Hood Dataset (CarHoods10k), an industry-grade 3D vehicle hood data set of over 10000 shapes along with mechanical performance data that were validated against real-world hood designs by industry experts. CarHoods10k offers researchers and practitioners the unique opportunity to develop novel methods on realistic data with relevance to real-world vehicle design. To illustrating central use cases, we first apply methods from geometric deep learning to learn a compact latent representation for design space exploration. Second, we use machine learning models to predict mechanical hood performance from the learned latent representation. We thus demonstrate the effectiveness of machine learning for building metamodels, which are used in design optimization whenever possible to replace costly engineering simulations. Third, we integrate CarHoods10k in a topology optimization approach based on evolutionary algorithms to demonstrate its capability to search for high-performing structures, while maintaining manufacturability constraints.

2. Deformation Clustering Methods for Topologically Optimized Structures under Crash Load based on Displacement Time Series

   Yasuyuki Shimizu, Nivesh Dommaraju, Mariusz Bujny, Stefan Menzel, Markus Olhofer, and Fabian Duddeck

   In 15th World Congress on Computational Mechanics (WCCM-XV), 2022
3. Multi-objective 3D Path Planning for UAVs in Large-Scale Urban Scenarios

   Nikolas Hohmann, Mariusz Bujny, Jurgen Adamy, and Markus Olhofer

   In 2022 IEEE Congress on Evolutionary Computation (CEC), 2022

   Abs DOI HTML

   In the context of real-world path planning applications for Unmanned Aerial Vehicles (UAVs), aspects such as handling of multiple objectives (e.g., minimizing risk, path length, travel time, energy consumption, or noise pollution), generation of smooth trajectories in 3D space, and the ability to deal with urban environments have to be taken into account jointly by an optimization algorithm to provide practically feasible solutions. Since the currently available methods do not allow for that, in this paper, we propose a holistic approach for solving a Multi-Objective Path Planning (MOPP) problem for UAVs in a three-dimensional, large-scale urban environment. For the tackled optimization problem, we propose an energy model and a noise model for a UAV, following a smooth 3D path. We utilize a path representation based on 3D Non-Uniform Rational BSplines (NURBS). As optimizers, we use a conventional version of an Evolution Strategy (ES), two standard Multi-Objective Evolutionary Algorithms (MOEAs) – NSGA2 and MO-CMAES, and a gradient-based L-BFGS-B approach. To guide the optimization, we propose hybrid versions of the mentioned algorithms by applying an advanced initialization scheme that is based on the exact bidirectional Dijkstra algorithm. We compare the different algorithms with and without hybrid initialization in a statistical analysis, which considers the number of function evaluations and quality features of the obtained Pareto fronts indicating convergence and diversity of the solutions. We evaluate the methods on a realistic 3D urban path planning scenario in New York City, based on real-world data exported from OpenStreetMap. The examination’s results indicate that hybrid initialization is the main factor for the efficient identification of near-optimal solutions.

4. Cooperative Multi-objective Topology Optimization Using Clustering and Metamodeling

   Nivesh Dommaraju, Mariusz Bujny, Stefan Menzel, Markus Olhofer, and Fabian Duddeck

   In 2022 IEEE Congress on Evolutionary Computation (CEC), 2022

   Abs DOI HTML

   Topology optimization optimizes material layout in a design space for a given objective, such as crash energy absorption, and a set of boundary conditions. In industrial applications, multi-objective topology optimization requires expensive simulations to evaluate the objectives and generate multiple Pareto-optimal solutions. So, it is more economical to identify preferred regions on the Pareto front and generate only the desired solutions. Clustering methods, a widely used subclass of machine learning methods, provide an unsupervised approach to summarize the dataset, which eases the identification of the preferred set of designs. However, generating solutions similar to the preferred designs based on different metrics is a challenging task. In this paper, we present an interactive method to generate designs similar to a preferred set using one of the state-ofthe-art weighted-sum approaches called scaled energy weighting - hybrid cellular automata (SEW-HCA). To avoid unnecessary computations, metamodels are used to predict the desired weight vectors needed by SEW-HCA. We evaluate an application of our method for cooperative topology optimization using a cantilever multi-load-case problem and a crashworthiness optimization problem. Using the proposed method, we could successfully generate designs that are similar to preferred solutions based on geometry and performance. We believe that this is a crucial component that will improve the usefulness of multi-objective topology optimization in real-world applications.

5. Using Evolutionary Algorithms for eVTOL Design Optimization based on Multi-agent Simulations

   Sebastian Brulin, Mariusz Bujny, Tim Puphal, and Stefan Menzel

   In MATSim User Meeting, 2022

   HTML
6. OSU-Honda automobile hood dataset (CarHoods10k)

   Satchit Ramnath, Jami J. Shah, Patricia Wollstadt, Mariusz Bujny, Stefan Menzel, and Duane Detwiler

   2022

   [Dataset]

   DOI HTML

2021

1. Hybrid Evolutionary Approach to Multi-objective Path Planning for UAVs

   Nikolas Hohmann, Mariusz Bujny, Jürgen Adamy, and Markus Olhofer

   In 2021 IEEE Symposium Series on Computational Intelligence (SSCI), 2021

   Abs DOI

   The goal of Multi-Objective Path Planning (MOPP) is to find Pareto-optimal paths for autonomous agents with respect to several optimization goals like minimizing risk, path length, travel time, or energy consumption. In this work, we formulate a MOPP for Unmanned Aerial Vehicles (UAVs). We utilize a path representation based on Non-Uniform Rational B-Splines (NURBS) and propose a hybrid evolutionary approach combining an Evolution Strategy (ES) with the exact Dijkstra algorithm. Moreover, we compare our approach in a statistical analysis to state-of-the-art exact (Dijkstra’s algorithm), gradient-based (L-BFGS-B), and evolutionary (NSGA-II) algorithms with respect to calculation time and quality features of the obtained Pareto fronts indicating convergence and diversity of the solutions. We evaluate the methods on a realistic 2D urban path planning scenario based on real-world data exported from OpenStreetMap. The examination’s results indicate that our approach is able to find significantly better solutions for the formulated problem than standard Evolutionary Algorithms (EAs). Moreover, the proposed method is able to obtain more diverse sets of trade-off solutions for different objectives than the standard exact approaches. Thus, the method combines the strengths of both approaches.

2. Eng. Optim.

   Similarity control in topology optimization under static and crash loading scenarios

   Muhammad Salman Yousaf, Mariusz Bujny, Nathan Zurbrugg, Duane Detwiler, and Fabian Duddeck

   Engineering Optimization, 2021

   Abs DOI HTML

   Topology Optimization (TO) redistributes the material within a design space to optimize certain objective functions under given constraints. Currently available TO methods do not consider the designer’s preferences about the final material layout in the optimized design. Contrarily, often an improved design similar to a reference design is required because of the economic, manufacturing or assembly restrictions. In this article, the proposed heuristic similarity control methods like the Energy Scaling Method (ESM), weak passive material method, and an intuitive method of modified design domain are compared with the formal mathematical method of Optimality Criteria (OC)-based Solid Isotropic Material with Penalization (SIMP) with a similarity constraint. Initially, the methods are coupled with Hybrid Cellular Automata (HCA) and OC-based SIMP for the TO of a cantilever beam under a static point load. The ESM is found to be the most effective and further tested for similarity-based TO with HCA in crash loading scenarios. The results show that ESM can be used to control the TO process effectively.

3. Hybrid Strategy Coupling EGO and CMA-ES for Structural Topology Optimization in Statics and Crashworthiness

   Elena Raponi, Mariusz Bujny, Markus Olhofer, Simonetta Boria, and Fabian Duddeck

   In Computational Intelligence, 2021

   Series Title: Studies in Computational Intelligence

   Abs DOI HTML

   Topology Optimization (TO) represents a relevant tool in the design of mechanical structures and, as such, it is currently used in many industrial applications. However, many TO optimization techniques are still questionable when applied to crashworthiness optimization problems due to their complexity and lack of gradient information. The aim of this work is to describe the Hybrid Kriging-assisted Level Set Method (HKG-LSM) and test its performance in the optimization of mechanical structures consisting of ensembles of beams subjected to both static and dynamic loads. The algorithm adopts a low-dimensional parametrization introduced by the Evolutionary Level Set Method (EA-LSM) for structural Topology Optimization and couples the Efficient Global Optimization (EGO) and the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) to converge towards the optimum within a fixed budget of evaluations. It takes advantage of the explorative capabilities of EGO ensuring a fast convergence at the beginning of the optimization procedure, as well as the flexibility and robustness of CMA-ES to exploit promising regions of the search space Precisely, HKG-LSM first uses the Kriging-based method for Level Set Topology Optimization (KG-LSM) and afterwards switches to the EALSM using CMA-ES, whose parameters are initialized based on the previous model. Within the research, a minimum compliance cantilever beam test case is used to validate the presented strategy at different dimensionalities, up to 15 variables. The method is then applied to a 15-variables 2D crash test case, consisting of a cylindrical pole impact on a rectangular beam fixed at both ends. Results show that HKG-LSM performs well in terms of convergence speed and hence represents a valuable option in real-world applications with limited computational resources.

4. SMO

   Topology Optimization of 3D-printed joints under crash loads using Evolutionary Algorithms

   Mariusz Bujny, Markus Olhofer, Nikola Aulig, and Fabian Duddeck

   Structural and Multidisciplinary Optimization, 2021

   Abs DOI HTML

   In order to take full advantage of the enormous design freedom offered by Additive Manufacturing (AM) technologies, the use of Topology Optimization (TO) methods becomes essential. Although TO is well-established in many disciplines, the problems in vehicle crashworthiness pose severe difficulties for standard, gradient-based approaches, due to high noisiness, multi-modality, and discontinuous nature of the nonlinear simulation responses considered typically as objectives and constraints. In this article, we propose to use Evolutionary Algorithms (EAs) together with a suitable low-dimensional representation in an extended version of the Evolutionary Level Set Method (EA-LSM), able to address complex 3D crash TO problems. The method is used to optimize a 3D-printed metal joint in a hybrid S-rail structure under axial crash loading, inspired by novel frame design concepts in industry. The obtained results show that the method is capable of handling optimization problems with multiple constraints, including challenging acceleration responses, and yields significantly better solutions than the state-of-the-art methods. Finally, the robustness of the obtained designs is studied, demonstrating the ability of EA-LSM to find designs of low sensitivity w.r.t. small variations of the loading conditions, which is crucial from the perspective of industrial applications of the proposed method.

5. Clustering Topologically-Optimized Designs based on Structural Deformation

   Ernest Hutapea, Nivesh Dommaraju, Mariusz Bujny, and Fabian Duddeck

   In Proceedings of the Munich Symposium on Lightweight Design, 2021

   HTML

2020

1. High Dimensional Bayesian Optimization Assisted by Principal Component Analysis

   Elena Raponi, Hao Wang, Mariusz Bujny, Simonetta Boria, and Carola Doerr

   In 16th International Conference on Parallel Problem Solving from Nature, 2020
2. Simultaneous Exploration of Geometric Features and Performance in Design Optimization

   Nivesh Dommaraju, Mariusz Bujny, Stefan Menzel, Markus Olhofer, and Fabian Duddeck

   In 16th International LS-DYNA Conference, 2020
3. Level Set Topology Optimization for Crashworthiness using Evolutionary Algorithms and Machine Learning

   Mariusz Bujny

   2020

   HTML
4. Multi-material Topology Optimization in LS-TaSC using Ordered SIMP Interpolation

   Satchit Ramnath, Mariusz Bujny, Nathan Zurbrugg, Stefan Menzel, and Duane Detwiler

   In 16th International LS-DYNA Conference, 2020

2019

1. CMAME

   Kriging-Assisted Topology Optimization of Crash Structures

   Elena Raponi, Mariusz Bujny, Markus Olhofer, Nikola Aulig, Simonetta Boria, and Fabian Duddeck

   Computer Methods in Applied Mechanics and Engineering, 2019
2. Hybrid Kriging-assisted Level Set Method for Structural Topology Optimization

   Elena Raponi, Mariusz Bujny, Markus Olhofer, Simonetta Boria, and Fabian Duddeck

   In 11th International Conference on Evolutionary Computation Theory and Applications, 2019
3. Identifying Topological Prototypes using Deep Point Cloud Autoencoder Networks

   Nivesh Dommaraju, Mariusz Bujny, Stefan Menzel, Markus Olhofer, and Fabian Duddeck

   In 2019 International Conference on Data Mining Workshops (ICDMW), 2019
4. Load Case Preference Patterns based on Parameterized Pareto-Optimal Vehicle Design Concept Optimization

   Satchit Ramnath, Nikola Aulig, Mariusz Bujny, Stefan Menzel, Imtiaz Gandikota, and Kurtis Horner

   In European LS-DYNA Conference 2019, 2019

2018

1. Learning-based Topology Variation in Evolutionary Level Set Topology Optimization

   Mariusz Bujny, Nikola Aulig, Markus Olhofer, and Fabian Duddeck

   In Proceedings of the Genetic and Evolutionary Computation Conference, 2018

   Abs DOI HTML

   The main goal in structural Topology Optimization is to find an optimal distribution of material within a defined design domain, under specified boundary conditions. This task is frequently solved with gradient-based methods, but for some problems, e.g. in the domain of crash Topology Optimization, analytical sensitivity information is not available. The recent Evolutionary Level Set Method (EA-LSM) uses Evolutionary Strategies and a representation based on geometric Level Set Functions to solve such problems. However, computational costs associated with Evolutionary Algorithms are relatively high and grow significantly with rising dimensionality of the optimization problem. In this paper, we propose an improved version of EA-LSM, exploiting an adaptive representation, where the number of structural components increases during the optimization. We employ a learning-based approach, where a pre-trained neural network model predicts favorable topological changes, based on the structural state of the design. The proposed algorithm converges quickly at the beginning, determining good designs in low-dimensional search spaces, and the representation is gradually extended by increasing structural complexity. The approach is evaluated on a standard minimum compliance design problem and its superiority with respect to a random adaptive method is demonstrated.

2017

1. IJOC

   Identification of optimal topologies for crashworthiness with the evolutionary level set method

   Mariusz Bujny, Nikola Aulig, Markus Olhofer, and Fabian Duddeck

   International Journal of Crashworthiness, 2017

   DOI
2. Kriging-guided level set method for crash topology optimization

   Elena Raponi, Mariusz Bujny, Markus Olhofer, Nikola Aulig, Simonetta Boria, and Fabian Duddeck

   In 7th GACM Colloquium on Computational Mechanics for Young Scientists from Academia and Industry, 2017
3. Topology optimization of crash structures with the hybrid evolutionary level set method

   Mariusz Bujny, Nikola Aulig, Markus Olhofer, and Fabian Duddeck

   In 12th World Congress of Structural and Multidisciplinary Optimisation, 2017
4. Optimal structures for crash by additive manufacturing

   Mariusz Bujny, Markus Olhofer, and Fabian Duddeck

   In 1st ECCOMAS Thematic Conf. on Simulation for Additive Manufacturing (Sim-AM), 2017
5. Topology optimization methods based on nonlinear and dynamic crash simulations

   Fabian Duddeck, Mariusz Bujny, and Duo Zeng

   In 11th European LS-DYNA Conference, 2017

   Abs

   Topology optimization for crashworthiness has been investigated during the last years, starting from methods based on linear elastic and static simulations [1] or so-called equivalent static loads (ESL) obtained by a single nonlinear crash simulation with a subsequent optimization loop based on the linear stiffness matrix and the corresponding sensitivities [2, 3]. Both methods do not consider material nonlinearities in their optimization process, which are essential for structural components designed for energy absorption, although it is well-known that plasticity and failure play an important role. As alternative, optimization methods have been proposed, which use fully nonlinear and dynamic crash simulations. The first method, presented for example in Patel’s PhD thesis, is based on a hybrid cellular automata approach (HCA) and derives optimal structures using a homogenized energy density approach where each finite element (cell) is modified until the highest degree of homogeneity is achieved [4, 5]. Because this is not fully appropriate for thin-walled structures, Hunkeler modified the approach (HCA-TWS – Hybrid Cellular Automata for Thin-walled Structures) such that deformation energy is only homogenized between larger structural entities (i.e. thin ribs / walls) [6, 7]. The most recent method, the EA-LSM, a combined level set method (LSM) and evolutionary approach, was then proposed by Bujny et al. where more appropriate objectives and constraints can be used with the drawback of higher computational costs [8, 9]. In this paper, the latest results for HCA-TWS and EA-LSM will be presented, see also [10]. Special focus is here the investigation of the influence of different material models for plasticity. Examples are inspired by recent material model developments for magnesium alloys with a characteristic anisotropy in the plasticity model [11]. As a result, it is shown that the optimal topologies depend on the material model and that it is necessary to use nonlinear and dynamic finite elements for crash topology optimization.

2016

1. Evolutionary Crashworthiness Topology Optimization of Thin-Walled Structures

   Mariusz Bujny, Nikola Aulig, Markus Olhofer, and Fabian Duddeck

   In 11th ASMO UK / ISSMO / NOED2016: International Conference on Numerical Optimisation Methods for Engineering Design, 2016
2. Hybrid Evolutionary Approach for Level Set Topology Optimization

   Mariusz Bujny, Nikola Aulig, Markus Olhofer, and Fabian Duddeck

   In IEEE 2016 Congress on Evolutionary Computation (CEC), 2016

   Abs DOI

   Although Topology Optimization is widely used in many industrial applications, it is still in the initial phase of development for highly nonlinear, multimodal and noisy problems, where the analytical sensitivity information is either not available or difficult to obtain. Since for such problems, gradient-based methods cannot be applied, a search for alternative approaches is inevitable. One possibility is a use of Evolutionary Algorithms, which are well-suited for this type of problems, but involve considerable computational costs. In this paper we propose a hybrid evolutionary optimization method using a geometric Level-Set Method for an implicit representation of mechanical structures. The hybrid approach aims for superior global search properties by using gradient approximations that can be obtained from structural state, equivalent state or any known heuristics. In order to evaluate the proposed method, a minimum compliance problem for a standard cantilever beam benchmark case is considered. The results show that the hybridization can be very beneficial in terms of convergence speed and performance of the optimized designs.

3. Evolutionary Level Set Method for Crashworthiness Topology Optimization

   Mariusz Bujny, Nikola Aulig, Markus Olhofer, and Fabian Duddeck

   In ECCOMAS Congress, 2016

   Abs

   Vehicle crashworthiness design belongs to one of the most complex problems considered in the design optimization. Physical phenomena that are taken into account in crash simulations range from complex contact modeling to mechanical failure of materials. This results in high nonlinearity of the optimization problem and involves remarkable amount of numerical noise and discontinuities of the objective functions that are being optimized. Consequently, the sensitivity information can be obtained analytically only for considerably simplified problems, which, in most cases, excludes the use of the gradient-based optimization methods. As a result, in the state-of-the-art methods for crashworthiness Topology Optimization, very strong assumptions about the properties of the optimization problem are made and heuristic approaches are used to optimize structures. As a result, in each of those methods, the optimality criterion is arguable. This problem can be solved with use of Evolutionary Algorithms, where no assumptions about the optimization problem have to be made and which perform very well even for highly nonlinear and discontinuous problems. In this paper we propose an evolutionary optimization approach using a geometric Level-Set Method for an implicit representation of mechanical structures. In order to evaluate the proposed method, an energy maximization problem for a 2D transverse bending of a rectangular beam is considered. The results show that the evolutionary optimization methods can be efficiently used for an optimization of crash-loaded structures. In most experiments, the optimized topologies show many similarities to the designs obtained with the Hybrid Cellular Automata technique.

2014

1. Aerodynamic Shape Optimisation of a High Speed Train Locomotive using Adjoint Solver

   Mariusz Bujny, Armen Jaworski, and Jacek Rokicki

   In Proceedings of the Second International Conference on Railway Technology: Research, Development and Maintenance, 2014

   DOI HTML