Pub Date : 2026-01-13DOI: 10.1016/j.advengsoft.2026.104102
Siqi Yin , Jiang Liu , Haidong Wang , Feiqi Wang , Yumin Huang , Xiangyang Cui , Yong Cai
Generating high-quality Cartesian grids for complex geometries remains a significant performance bottleneck in computational engineering. This paper presents a novel parallel framework to address this challenge, centered on the synergistic integration of two key innovations: (1) a distance-based adaptive strategy that, through its inherent smoothness, obviates the need for explicit grid balancing steps; and (2) a high-throughput subdivision and classification algorithm built upon a Linear Bounding Volume Hierarchy (LBVH) and an efficient parallel Union-Find algorithm. The entire workflow is efficiently parallelized using Threading Building Blocks (TBB). Evaluated on a 16-core desktop workstation, the framework demonstrates exceptional performance. The parallel subdivision engine achieves a throughput exceeding one million cells per second, while the interior/exterior classification process reaches ten million cells per second. In conclusion, the proposed method exhibits robust scalability and efficiency, paving the way for ultra-large-scale mesh generation and dynamic geometric reconstruction on commodity multi-core architectures.
{"title":"An efficient and robust parallel strategy for Cartesian grid generation on complex geometries","authors":"Siqi Yin , Jiang Liu , Haidong Wang , Feiqi Wang , Yumin Huang , Xiangyang Cui , Yong Cai","doi":"10.1016/j.advengsoft.2026.104102","DOIUrl":"10.1016/j.advengsoft.2026.104102","url":null,"abstract":"<div><div>Generating high-quality Cartesian grids for complex geometries remains a significant performance bottleneck in computational engineering. This paper presents a novel parallel framework to address this challenge, centered on the synergistic integration of two key innovations: (1) a distance-based adaptive strategy that, through its inherent smoothness, obviates the need for explicit grid balancing steps; and (2) a high-throughput subdivision and classification algorithm built upon a Linear Bounding Volume Hierarchy (LBVH) and an efficient parallel Union-Find algorithm. The entire workflow is efficiently parallelized using Threading Building Blocks (TBB). Evaluated on a 16-core desktop workstation, the framework demonstrates exceptional performance. The parallel subdivision engine achieves a throughput exceeding one million cells per second, while the interior/exterior classification process reaches ten million cells per second. In conclusion, the proposed method exhibits robust scalability and efficiency, paving the way for ultra-large-scale mesh generation and dynamic geometric reconstruction on commodity multi-core architectures.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"214 ","pages":"Article 104102"},"PeriodicalIF":5.7,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145950130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-05DOI: 10.1016/j.advengsoft.2025.104094
Nan Xu, Yanhui Liu, Hongjin Chen, Ming Gao
Concrete filled steel tube (CFST) columns are vulnerable to lateral impact from automobiles, ships and derailed trains as global transportation rapidly expands. The theoretical research about impact response is challenging as it involves a series of physical phenomena. This research aims to establish simplified mathematic model to evaluate CFST deflection and damage reliability through machine learning techniques, where input features contain 11 variables. 410 impact specimens were gathered and they were divided at 8:2 ratio for model training and testing. Nine hybrid algorithms were created, IGWO (Improved Grey Wolf Optimizer)-Ann (artificial neural networks) performed the best deflection prediction with correlation coefficient of 0.90. The simplified model of deflection was established, and its computational efficiency is substantially superior to theoretical methods. The credibility of IGWO-Ann was verified by Shapley Additive exPlanations analysis. Furthermore, the explicit equations of damage probability and damage reliability were proposed. Sensitivity analysis indicated that section outer diameter and steel strength influence dramatically damage probability. Finally, an impact-resistant procedure was suggested, which provides an efficient, user-friendly and precise method for structural engineers.
随着全球交通运输的迅速扩张，钢管混凝土（CFST）柱容易受到汽车、船舶和出轨列车的横向冲击。由于冲击响应涉及一系列物理现象，其理论研究具有一定的挑战性。本研究旨在通过机器学习技术建立CFST挠度和损伤可靠性评估的简化数学模型，其中输入特征包含11个变量。收集410个冲击试件，按8:2的比例进行模型训练和试验。建立了9种混合算法，其中IGWO (Improved Grey Wolf Optimizer)-Ann （artificial neural networks）的偏转预测效果最好，相关系数为0.90。建立了挠度的简化模型，其计算效率大大优于理论方法。通过Shapley加性解释分析验证了IGWO-Ann的可信度。在此基础上，建立了损伤概率和损伤可靠度的显式方程。敏感性分析表明，截面外径和钢强度对损伤概率影响较大。最后，提出了一种抗冲击程序，为结构工程师提供了一种高效、方便和精确的方法。
{"title":"Prediction models of deflection and damage reliability for laterally impacted CFST based on data-driven technology","authors":"Nan Xu, Yanhui Liu, Hongjin Chen, Ming Gao","doi":"10.1016/j.advengsoft.2025.104094","DOIUrl":"10.1016/j.advengsoft.2025.104094","url":null,"abstract":"<div><div>Concrete filled steel tube (CFST) columns are vulnerable to lateral impact from automobiles, ships and derailed trains as global transportation rapidly expands. The theoretical research about impact response is challenging as it involves a series of physical phenomena. This research aims to establish simplified mathematic model to evaluate CFST deflection and damage reliability through machine learning techniques, where input features contain 11 variables. 410 impact specimens were gathered and they were divided at 8:2 ratio for model training and testing. Nine hybrid algorithms were created, IGWO (Improved Grey Wolf Optimizer)-Ann (artificial neural networks) performed the best deflection prediction with correlation coefficient of 0.90. The simplified model of deflection was established, and its computational efficiency is substantially superior to theoretical methods. The credibility of IGWO-Ann was verified by Shapley Additive exPlanations analysis. Furthermore, the explicit equations of damage probability and damage reliability were proposed. Sensitivity analysis indicated that section outer diameter and steel strength influence dramatically damage probability. Finally, an impact-resistant procedure was suggested, which provides an efficient, user-friendly and precise method for structural engineers.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104094"},"PeriodicalIF":5.7,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926010","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-27DOI: 10.1016/j.advengsoft.2025.104089
Shuangwei Hu , Qingshan Wang , Zhen Li
Doubly-curved shells are critical components in submarine bow hulls and unmanned underwater vehicles (UUVs). This study investigates their acoustic stealth performance—characterized by sound pressure level (SPL) and sound power level (SWL)—under external excitation in fluid environments. Based on the FSDT (first-order shear deformation theory), we establish coupled vibro-acoustic equations and employ the Kirchhoff-Helmholtz integral with CHIEF points to resolve non-uniqueness issues. A meshfree-Fourier spectral technique discretizes structural-acoustic variables, achieving high-fidelity predictions validated against FEM/BEM and literature. Parametric studies further quantify how geometry, material properties, excitation types, and boundary conditions affect acoustic responses, providing key insights for stealth-oriented shell design. The research results indicate: Frequency-amplitude reduction with increasing shape parameters, as structural softening lowers natural frequencies. Linear amplitude scaling with excitation force, leaving resonant frequencies unchanged in the linear regime. Fluid-dependent directivity, showing asymmetric patterns in light fluid and uniform distribution in heavy fluid.
{"title":"Meshfree-based prediction for external acoustic radiation of laminated composite doubly-curved revolution shells","authors":"Shuangwei Hu , Qingshan Wang , Zhen Li","doi":"10.1016/j.advengsoft.2025.104089","DOIUrl":"10.1016/j.advengsoft.2025.104089","url":null,"abstract":"<div><div>Doubly-curved shells are critical components in submarine bow hulls and unmanned underwater vehicles (UUVs). This study investigates their acoustic stealth performance—characterized by sound pressure level (SPL) and sound power level (SWL)—under external excitation in fluid environments. Based on the FSDT (first-order shear deformation theory), we establish coupled vibro-acoustic equations and employ the Kirchhoff-Helmholtz integral with CHIEF points to resolve non-uniqueness issues. A meshfree-Fourier spectral technique discretizes structural-acoustic variables, achieving high-fidelity predictions validated against FEM/BEM and literature. Parametric studies further quantify how geometry, material properties, excitation types, and boundary conditions affect acoustic responses, providing key insights for stealth-oriented shell design. The research results indicate: Frequency-amplitude reduction with increasing shape parameters, as structural softening lowers natural frequencies. Linear amplitude scaling with excitation force, leaving resonant frequencies unchanged in the linear regime. Fluid-dependent directivity, showing asymmetric patterns in light fluid and uniform distribution in heavy fluid.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104089"},"PeriodicalIF":5.7,"publicationDate":"2025-12-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840458","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-27DOI: 10.1016/j.advengsoft.2025.104092
Jin Wang , Mingliang Zhu , Jiahao Cao , Jiamin Guo , Zhiwei Miao
The cable dome is widely recognized for its lightweight, structural efficiency, and ease of length adjustment for active displacement control. However, achieving effective active control critically depends on the optimal placement of actuators. This paper proposes a physics-informed neural network (PINN) and singular value decomposition (SVD) framework to optimize actuator placement. Specifically, we introduce a PINN-based loss function that simultaneously incorporates structural control accuracy and essential physical constraints, enhancing the optimization efficiency and accuracy. The use of SVD within the PINN framework systematically extracts dominant sensitivity directions from the structural response sensitivity matrix, significantly reducing computational complexity and improving predictive capability. The approach is validated through numerical studies involving four different cable dome configurations. Results demonstrate that the developed method achieves substantial reductions (over 85 %) in structural displacement, while requiring actuators on only 5–10 % of structural elements. The proposed method provides a practical, efficient, and reliable solution for actuator placement, addressing a critical engineering challenge in active control of cable domes.
{"title":"A physics-informed neural network and singular value decomposition framework for actuator placement optimization in cable dome","authors":"Jin Wang , Mingliang Zhu , Jiahao Cao , Jiamin Guo , Zhiwei Miao","doi":"10.1016/j.advengsoft.2025.104092","DOIUrl":"10.1016/j.advengsoft.2025.104092","url":null,"abstract":"<div><div>The cable dome is widely recognized for its lightweight, structural efficiency, and ease of length adjustment for active displacement control. However, achieving effective active control critically depends on the optimal placement of actuators. This paper proposes a physics-informed neural network (PINN) and singular value decomposition (SVD) framework to optimize actuator placement. Specifically, we introduce a PINN-based loss function that simultaneously incorporates structural control accuracy and essential physical constraints, enhancing the optimization efficiency and accuracy. The use of SVD within the PINN framework systematically extracts dominant sensitivity directions from the structural response sensitivity matrix, significantly reducing computational complexity and improving predictive capability. The approach is validated through numerical studies involving four different cable dome configurations. Results demonstrate that the developed method achieves substantial reductions (over 85 %) in structural displacement, while requiring actuators on only 5–10 % of structural elements. The proposed method provides a practical, efficient, and reliable solution for actuator placement, addressing a critical engineering challenge in active control of cable domes.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104092"},"PeriodicalIF":5.7,"publicationDate":"2025-12-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840545","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-24DOI: 10.1016/j.advengsoft.2025.104093
Chen Wu , Lin Wang , Tiantang Yu , Sundararajan Natarajan
In addressing seepage problems with free surfaces, conventional moving mesh methods often encounter numerical instability due to mesh distortion, whereas fixed mesh approaches typically fail to deliver high-accuracy solutions with controllable computational cost. To overcome these limitations, this study develops an adaptive isogeometric analysis framework for two-dimensional steady-state unconfined seepage, aiming to achieve efficient and stable numerical solutions. Combining the local refinement capability of truncated hierarchical NURBS with the permeability adjustment strategy, the proposed framework is shown to capture the free surface accurately while avoiding mesh reconstruction required by traditional methods. By introducing an adaptive criterion based on a posteriori error estimator, the system can intelligently identify and refine key regions near the free surface during the iteration process, significantly improving computational efficiency without compromising accuracy. Numerical results demonstrate that the proposed method exhibits excellent robustness under conditions with different seepage properties, and shows exceptionally good agreement with reference solutions. This method also provides a reliable foundation for subsequent multi-physics coupling analysis and seepage simulation in complex engineering scenarios.
{"title":"An adaptive isogeometric analysis framework for two-dimensional steady-state unconfined seepage problems","authors":"Chen Wu , Lin Wang , Tiantang Yu , Sundararajan Natarajan","doi":"10.1016/j.advengsoft.2025.104093","DOIUrl":"10.1016/j.advengsoft.2025.104093","url":null,"abstract":"<div><div>In addressing seepage problems with free surfaces, conventional moving mesh methods often encounter numerical instability due to mesh distortion, whereas fixed mesh approaches typically fail to deliver high-accuracy solutions with controllable computational cost. To overcome these limitations, this study develops an adaptive isogeometric analysis framework for two-dimensional steady-state unconfined seepage, aiming to achieve efficient and stable numerical solutions. Combining the local refinement capability of truncated hierarchical NURBS with the permeability adjustment strategy, the proposed framework is shown to capture the free surface accurately while avoiding mesh reconstruction required by traditional methods. By introducing an adaptive criterion based on a posteriori error estimator, the system can intelligently identify and refine key regions near the free surface during the iteration process, significantly improving computational efficiency without compromising accuracy. Numerical results demonstrate that the proposed method exhibits excellent robustness under conditions with different seepage properties, and shows exceptionally good agreement with reference solutions. This method also provides a reliable foundation for subsequent multi-physics coupling analysis and seepage simulation in complex engineering scenarios.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104093"},"PeriodicalIF":5.7,"publicationDate":"2025-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840544","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The paper presents the results of the development of an original mathematical model and a numerical algorithm for pore network modeling of non-Newtonian flows of power law fluids in microchannel media. The pore network model is based on an original method of connecting one-dimensional and three-dimensional CFD solvers into a conjugated hydrodynamic model with a common pressure field. The non-Newtonian properties of a liquid are taken into account in a pore network model by setting a generalized resistance coefficient. An original microfluidic chips simulating fractured and porous rock were created to test the developed numerical method of pore network modeling of non-Newtonian flows. The flow of polymer solutions in a microfluidic chip has been studied experimentally. Testing has shown that the pore network algorithm for the case of a non-Newtonian (power law) flow in a single circular tube coincides with the analytical solution with an error below 0.01 %. The average uncertainty in determining the pressure drop in a microfluidic chip by the proposed algorithm over a wide range of fluid flow rates and their rheological characteristics does not exceed 10 %. At the same time, the computational efficiency of the pore network model is demonstrated to be 180 times of magnitude higher than that of the CFD model for a fractured microchip model and to be 720 times for a porous microchip model, all other things being equal and with close accuracy compared to the experiment.
{"title":"Development and testing of a new pore network algorithm for modeling flows of power law fluids in porous media","authors":"S.A. Filimonov , A.I. Pryazhnikov , D.V. Guzei , A.V. Minakov","doi":"10.1016/j.advengsoft.2025.104091","DOIUrl":"10.1016/j.advengsoft.2025.104091","url":null,"abstract":"<div><div>The paper presents the results of the development of an original mathematical model and a numerical algorithm for pore network modeling of non-Newtonian flows of power law fluids in microchannel media. The pore network model is based on an original method of connecting one-dimensional and three-dimensional CFD solvers into a conjugated hydrodynamic model with a common pressure field. The non-Newtonian properties of a liquid are taken into account in a pore network model by setting a generalized resistance coefficient. An original microfluidic chips simulating fractured and porous rock were created to test the developed numerical method of pore network modeling of non-Newtonian flows. The flow of polymer solutions in a microfluidic chip has been studied experimentally. Testing has shown that the pore network algorithm for the case of a non-Newtonian (power law) flow in a single circular tube coincides with the analytical solution with an error below 0.01 %. The average uncertainty in determining the pressure drop in a microfluidic chip by the proposed algorithm over a wide range of fluid flow rates and their rheological characteristics does not exceed 10 %. At the same time, the computational efficiency of the pore network model is demonstrated to be 180 times of magnitude higher than that of the CFD model for a fractured microchip model and to be 720 times for a porous microchip model, all other things being equal and with close accuracy compared to the experiment.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104091"},"PeriodicalIF":5.7,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145840542","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-19DOI: 10.1016/j.advengsoft.2025.104090
Jie You , Yonghong Zhao , Liangyue Jia , Nan Wang , Zhibin Sun , Wenkai Zou , Yu Hu , Liang Liu , Chuanyang Zhang
Finite-element analysis (FEA) is the benchmark for crashworthiness evaluation, yet its prohibitive computational cost and labour-intensive re-meshing make it unsuitable for iterative structural optimization. Although surrogate models offer partial relief, they still demand large simulation datasets and frequent mesh updates. Focusing on the vehicle front crash condition, proposing a Sectional Force-Based Multi-Stage Physics Informed Surrogate Model (SFB-MSPISM) that integrates sectional force features with semi-empirical physical priors (i.e., the Gérard buckling formula) in a two-stage architecture, thereby reducing the training data requirement and virtually eliminating manual re-meshing. In Stage-1, a Physics-Informed XGBoost-CNN-Transformer ensemble (PI-XCT) is proposed to predict the peak sectional force and energy absorption of five key beams; In Stage-2, a multi-output XGBoost regressor is proposed to estimate the maximum crash acceleration and the Toe-board intrusion. Trained on fewer than 130 high-fidelity simulations (100 data for Stage-1 and 26 data for Stage-2), SFB-MSPISM attains a coefficient of determination of 0.97 for peak deceleration and a mean intrusion error of 2.525 mm (≤5 %), while reducing per-design evaluation time from 6.5 h to 0.038 s. These results show a speed-up exceeding five orders of magnitude and virtually eliminate human intervention, thereby enabling millisecond-scale, physically consistent crashworthiness assessment for rapid design exploration.
{"title":"A fast crashworthiness assessment framework: Sectional force-based multi-stage physics informed surrogate model","authors":"Jie You , Yonghong Zhao , Liangyue Jia , Nan Wang , Zhibin Sun , Wenkai Zou , Yu Hu , Liang Liu , Chuanyang Zhang","doi":"10.1016/j.advengsoft.2025.104090","DOIUrl":"10.1016/j.advengsoft.2025.104090","url":null,"abstract":"<div><div>Finite-element analysis (FEA) is the benchmark for crashworthiness evaluation, yet its prohibitive computational cost and labour-intensive re-meshing make it unsuitable for iterative structural optimization. Although surrogate models offer partial relief, they still demand large simulation datasets and frequent mesh updates. Focusing on the vehicle front crash condition, proposing a Sectional Force-Based Multi-Stage Physics Informed Surrogate Model (SFB-MSPISM) that integrates sectional force features with semi-empirical physical priors (i.e., the Gérard buckling formula) in a two-stage architecture, thereby reducing the training data requirement and virtually eliminating manual re-meshing. In Stage-1, a Physics-Informed XGBoost-CNN-Transformer ensemble (PI-XCT) is proposed to predict the peak sectional force and energy absorption of five key beams; In Stage-2, a multi-output XGBoost regressor is proposed to estimate the maximum crash acceleration and the Toe-board intrusion. Trained on fewer than 130 high-fidelity simulations (100 data for Stage-1 and 26 data for Stage-2), SFB-MSPISM attains a coefficient of determination of 0.97 for peak deceleration and a mean intrusion error of 2.525 mm (≤5 %), while reducing per-design evaluation time from 6.5 h to 0.038 s. These results show a speed-up exceeding five orders of magnitude and virtually eliminate human intervention, thereby enabling millisecond-scale, physically consistent crashworthiness assessment for rapid design exploration.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104090"},"PeriodicalIF":5.7,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790899","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-19DOI: 10.1016/j.advengsoft.2025.104087
Teoman Toprak , Michael Loibl , Guilherme H. Teixeira , Irina Shishkina , Chen Miao , Josef Kiendl , Benjamin Marussig , Florian Kummer
In the field of scientific computing, one often finds several alternative software packages (with open or closed source code) for solving a specific problem. These packages sometimes even use alternative methodological approaches, e.g., different numerical discretizations. If one decides to use one of these packages, it is often not clear which one is the best choice. To make an informed decision, it is necessary to measure the performance of the alternative software packages for a suitable set of test problems, i.e., to set up a benchmark. However, setting up benchmarks ad-hoc can become overwhelming as the parameter space expands rapidly. Very often, the design of the benchmark is also not fully set at the start of some project. For instance, adding new libraries, adapting metrics, or introducing new benchmark cases during the project can significantly increase complexity and necessitate laborious re-evaluation of previous results. This paper presents a proven approach that utilizes established Continuous Integration tools and practices to achieve high automation of benchmark execution and reporting. Our use case is the numerical integration (quadrature) on arbitrary domains, which are bounded by implicitly or parametrically defined curves or surfaces in 2D or 3D.
{"title":"Employing Continuous Integration inspired workflows for benchmarking of scientific software — A use case on numerical cut element quadrature","authors":"Teoman Toprak , Michael Loibl , Guilherme H. Teixeira , Irina Shishkina , Chen Miao , Josef Kiendl , Benjamin Marussig , Florian Kummer","doi":"10.1016/j.advengsoft.2025.104087","DOIUrl":"10.1016/j.advengsoft.2025.104087","url":null,"abstract":"<div><div>In the field of scientific computing, one often finds several alternative software packages (with open or closed source code) for solving a specific problem. These packages sometimes even use alternative methodological approaches, e.g., different numerical discretizations. If one decides to use one of these packages, it is often not clear which one is the best choice. To make an informed decision, it is necessary to measure the performance of the alternative software packages for a suitable set of test problems, i.e., to set up a benchmark. However, setting up benchmarks ad-hoc can become overwhelming as the parameter space expands rapidly. Very often, the design of the benchmark is also not fully set at the start of some project. For instance, adding new libraries, adapting metrics, or introducing new benchmark cases during the project can significantly increase complexity and necessitate laborious re-evaluation of previous results. This paper presents a proven approach that utilizes established Continuous Integration tools and practices to achieve high automation of benchmark execution and reporting. Our use case is the numerical integration (quadrature) on arbitrary domains, which are bounded by implicitly or parametrically defined curves or surfaces in 2D or 3D.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104087"},"PeriodicalIF":5.7,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790985","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-18DOI: 10.1016/j.advengsoft.2025.104088
Wei Chen , Dingding Wang , Ping Xiang , Peng Shi , Junsong Hu
In this research, a concise and efficient numerical approach is presented to explore the size-dependent free vibration behavior of functionally graded (FG) microplates composed of graphene origami (GOri)-enabled auxetic metamaterial (GOEAM), supported by Winkler, Pasternak, or Kerr foundations. The FG microplate is modeled as a multilayered structure composed of isotropic and homogeneous GOEAM layers, with a stepwise GOri dispersion through the thickness. At the same time, AI-assisted micromechanical modeling using genetic programming (GP) techniques are employed to precisely describe the complex material behavior. In addition, a refined plate theory involving four independent variables is adopted to incorporate both bending responses and shear effects. To address size-dependent phenomena, the modified couple stress theory (MCST), which introduces an intrinsic material length scale parameter (MLSP), is embedded within the conventional continuum mechanics framework. Thereafter, applying Hamilton’s principle, the weak formulation governing the size-dependent free vibration of the FG-GOEAM microplate placed on Winkler, Pasternak, or Kerr-type foundations is established. The corresponding numerical results are then acquired using the isogeometric analysis (IGA) technique. After validating the convergence and efficacy of the methodology presented herein, an extensive investigation was conducted to examine how several factors such as GOri dispersion pattern, weight fraction, folding degree, MLSP, and foundation stiffness affect the free vibration performance of the FG metamaterial microplates. The study demonstrates that the inclusion of MLSP alters how the frequencies of FG-GOEAM microplates varies with changes in GOri weight fraction and folding degree. Specifically, in general, an increase in the MLSP accentuates the increasing trend of frequency as GOri weight fraction rises, and gradually transforms the decreasing trend of frequency with the reduction of GOri folding degree into an increasing one. Additionally, the Pasternak shear layer coefficient and the Kerr foundation's intermediate shear layer coefficient dominantly influence the microplates' frequency.
{"title":"Isogeometric free vibration analysis of size-dependent functionally graded graphene origami-enabled auxetic metamaterial microplates supported by Winkler/Pasternak/Kerr foundation","authors":"Wei Chen , Dingding Wang , Ping Xiang , Peng Shi , Junsong Hu","doi":"10.1016/j.advengsoft.2025.104088","DOIUrl":"10.1016/j.advengsoft.2025.104088","url":null,"abstract":"<div><div>In this research, a concise and efficient numerical approach is presented to explore the size-dependent free vibration behavior of functionally graded (FG) microplates composed of graphene origami (GOri)-enabled auxetic metamaterial (GOEAM), supported by Winkler, Pasternak, or Kerr foundations. The FG microplate is modeled as a multilayered structure composed of isotropic and homogeneous GOEAM layers, with a stepwise GOri dispersion through the thickness. At the same time, AI-assisted micromechanical modeling using genetic programming (GP) techniques are employed to precisely describe the complex material behavior. In addition, a refined plate theory involving four independent variables is adopted to incorporate both bending responses and shear effects. To address size-dependent phenomena, the modified couple stress theory (MCST), which introduces an intrinsic material length scale parameter (MLSP), is embedded within the conventional continuum mechanics framework. Thereafter, applying Hamilton’s principle, the weak formulation governing the size-dependent free vibration of the FG-GOEAM microplate placed on Winkler, Pasternak, or Kerr-type foundations is established. The corresponding numerical results are then acquired using the isogeometric analysis (IGA) technique. After validating the convergence and efficacy of the methodology presented herein, an extensive investigation was conducted to examine how several factors such as GOri dispersion pattern, weight fraction, folding degree, MLSP, and foundation stiffness affect the free vibration performance of the FG metamaterial microplates. The study demonstrates that the inclusion of MLSP alters how the frequencies of FG-GOEAM microplates varies with changes in GOri weight fraction and folding degree. Specifically, in general, an increase in the MLSP accentuates the increasing trend of frequency as GOri weight fraction rises, and gradually transforms the decreasing trend of frequency with the reduction of GOri folding degree into an increasing one. Additionally, the Pasternak shear layer coefficient and the Kerr foundation's intermediate shear layer coefficient dominantly influence the microplates' frequency.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104088"},"PeriodicalIF":5.7,"publicationDate":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790898","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To overcome the limitations of single Graphics Processing Unit (GPU) configurations in terms of computational resources and acceleration performance, this study develops a multi-GPU parallel computing framework for the explicit finite element method (FEM) that incorporates a parallel contact algorithm. A hybrid parallelization approach is adopted, combining coarse-grained parallelism with subdomains mapped to GPUs and fine-grained parallelism with elements mapped to threads, along with a stream-per-element-type concurrency technique to achieve efficient multi-GPU computation of element internal forces. For the global contact search phase, a GPU-to-GPU contact node communication algorithm is designed, and a GPU-parallelized bucket sort algorithm is developed. To address inter-GPU contact node drift after sliding, a communication and reorganization strategy for remote nodes is proposed. A complete inter-GPU contact force communication scheme is constructed based on the penalty contact algorithm. The performance of the proposed multi-GPU explicit FEM framework is evaluated through a series of benchmark simulations, demonstrating a maximum speedup of 223.29 on four GPUs, significantly enhancing the computational efficiency for drop-test simulations.
{"title":"A Multi-GPU explicit finite element framework with a parallel contact algorithm for drop testing of electronic products","authors":"Xinggang Cao, Xiang Zhao, Zhenhui Liu, Yongjie Pei, Yong Cai, Xiangyang Cui","doi":"10.1016/j.advengsoft.2025.104086","DOIUrl":"10.1016/j.advengsoft.2025.104086","url":null,"abstract":"<div><div>To overcome the limitations of single Graphics Processing Unit (GPU) configurations in terms of computational resources and acceleration performance, this study develops a multi-GPU parallel computing framework for the explicit finite element method (FEM) that incorporates a parallel contact algorithm. A hybrid parallelization approach is adopted, combining coarse-grained parallelism with subdomains mapped to GPUs and fine-grained parallelism with elements mapped to threads, along with a stream-per-element-type concurrency technique to achieve efficient multi-GPU computation of element internal forces. For the global contact search phase, a GPU-to-GPU contact node communication algorithm is designed, and a GPU-parallelized bucket sort algorithm is developed. To address inter-GPU contact node drift after sliding, a communication and reorganization strategy for remote nodes is proposed. A complete inter-GPU contact force communication scheme is constructed based on the penalty contact algorithm. The performance of the proposed multi-GPU explicit FEM framework is evaluated through a series of benchmark simulations, demonstrating a maximum speedup of 223.29 on four GPUs, significantly enhancing the computational efficiency for drop-test simulations.</div></div>","PeriodicalId":50866,"journal":{"name":"Advances in Engineering Software","volume":"213 ","pages":"Article 104086"},"PeriodicalIF":5.7,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145738712","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号