International Journal of Computational Fluid Dynamics最新文献_第4页

Prediction of Flow Field Over Airfoils Based on Transformer Neural Network 基于变压器神经网络的翼型流场预测

4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2023-10-12 DOI: 10.1080/10618562.2023.2259806

Jianbo Zhou, Rui Zhang, Lyu Chen

{"title":"Prediction of Flow Field Over Airfoils Based on Transformer Neural Network","authors":"Jianbo Zhou, Rui Zhang, Lyu Chen","doi":"10.1080/10618562.2023.2259806","DOIUrl":"https://doi.org/10.1080/10618562.2023.2259806","url":null,"abstract":"AbstractAirfoil flow field data acquisition is pivotal to the study of aerodynamics, traditionally relying on time-consuming computational fluid dynamics simulations or expensive wind tunnel tests. Herein, we introduce a new methodology leveraging Transformer Neural Network (TNN), which differs from conventional methodologies by employing self-attention mechanisms, to effectively predict these critical flow field data using historical data. A comprehensive set of experiments demonstrates the TNN model’s exceptional predictive accuracy, achieving over 95% across various airfoils under various operating conditions. Beyond accuracy and efficiency, we introduce an attention principle in our TNN model enhancing its interpretability. By aligning the TNN model’s attention distribution with the aerodynamic principles of airfoils, we illustrate how it utilises these geometric attributes in its predictions, thereby offering theoretical backing to its predictive outcomes. Our TNN model’s commendable accuracy, efficiency and interpretability illuminate the pathway for continued exploration in the fusion of deep learning with computational fluid dynamics.KEYWORDS: Deep learningTransformer Neural Networkairfoilflow field Disclosure statementNo potential conflict of interest was reported by the author(s).Data Availability StatementThe data that support the findings of this study are available from the corresponding author upon reasonable request.Additional informationFundingThis work was supported by Scientific Research Project of Department of Education of Hunan Province [grant number 21C1577]; Natural Science Foundation of Hunan Province [grant number 2022JJ60090].","PeriodicalId":56288,"journal":{"name":"International Journal of Computational Fluid Dynamics","volume":"42 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"136012556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Unified Grid Approach Using Hamiltonian Paths for Computing Aerodynamic Flows 基于哈密顿路径的统一网格法计算气动流动

4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2023-10-04 DOI: 10.1080/10618562.2023.2264198

Yong Su Jung, Bharath Govindarajan, James Baeder

{"title":"A Unified Grid Approach Using Hamiltonian Paths for Computing Aerodynamic Flows","authors":"Yong Su Jung, Bharath Govindarajan, James Baeder","doi":"10.1080/10618562.2023.2264198","DOIUrl":"https://doi.org/10.1080/10618562.2023.2264198","url":null,"abstract":"AbstractA solution algorithm using Hamiltonian paths is presented as a unified grid approach for rotorcraft applications. Hidden line structures are robustly identified on general two- and three-dimensional unstructured grids with mixed elements, providing a framework for line-based solvers. A pure quadrilateral/hexahedral mesh is a prerequisite for line identification and enables approximate factorisation along the lines. The numerical efficiency obtained using the line-implicit method on various unstructured grids is better than that of the point-implicit method. Both finite-difference and gradient reconstructions are possible regardless of grid type. A combined reconstruction method is applied, which uses different reconstructions simultaneously but for different grid directions. Finally, the solution convergence rate is further improved using a preconditioned generalized minimal residual method (GMRES), where the preconditioning step is performed using the efficient line-implicit method.Keywords: Rotorcraft aerodynamicsunstructured gridcomputational efficiencyline-implicit methodgeneralized minimal residual method AcknowledgementsThe authors would like to acknowledge Dr. Roger Strawn (Army AFDD) and Dr. Rajneesh Singh (ARL) for their continued support of HAMSTR development. This work was supported by the National Supercomputing Center with supercomputing resources including technical support (KSC-2022-CRE-0197).Disclosure statementNo potential conflict of interest was reported by the author(s).","PeriodicalId":56288,"journal":{"name":"International Journal of Computational Fluid Dynamics","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135646263","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Fluid Flow Modelling Using Physics-Informed Convolutional Neural Network in Parametrised Domains 在参数化域使用物理信息卷积神经网络的流体流动建模

4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2023-01-02 DOI: 10.1080/10618562.2023.2260763

Ondřej Bublík, Václav Heidler, Aleš Pecka, Jan Vimmr

{"title":"Fluid Flow Modelling Using Physics-Informed Convolutional Neural Network in Parametrised Domains","authors":"Ondřej Bublík, Václav Heidler, Aleš Pecka, Jan Vimmr","doi":"10.1080/10618562.2023.2260763","DOIUrl":"https://doi.org/10.1080/10618562.2023.2260763","url":null,"abstract":"AbstractWe design and implement a physics-informed convolutional neural network (CNN) to solve fluid flow problems on a parametrised domain. The goal is to compare the effectiveness of training based solely on CFD-generated training data with purely physics-informed training and training based on a combination of both. We consider the problem of incompressible fluid flow in a convergent-divergent channel with variable wall shape. A speciality of the designed neural network is the incorporation of the boundary condition directly in the CNN. A physics-informed CNN that uses a non-Cartesian mesh poses a challenge when evaluating partial derivatives. We propose a gradient layer that approximates the first and second partial derivatives by finite differences using Lagrange interpolation. Our analysis shows that the convergence of purely physics-informed training is slow. However, using a small training dataset in combination with physics-informed training can achieve results comparable to physics-uninformed training with a considerably larger training dataset.Keywords: Physics-informed neural networkconvolutional neural networkU-Netincompressible fluid flowfluid dynamics Disclosure statementNo potential conflict of interest was reported by the author(s).Additional informationFundingThis research is supported by project GA21-31457S ‘Fast flow-field prediction using deep neural networks for solving fluid-structure interaction problems’ of the Grant Agency of the Czech Republic.","PeriodicalId":56288,"journal":{"name":"International Journal of Computational Fluid Dynamics","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135798728","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Wall-Resolved Large-Eddy Simulation of Flow Over a Three-Dimensional Gaussian Bump 三维高斯凸起上流动的壁面分辨大涡模拟

4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2023-01-02 DOI: 10.1080/10618562.2023.2246391

Donald P. Rizzetta, Daniel J. Garmann

{"title":"Wall-Resolved Large-Eddy Simulation of Flow Over a Three-Dimensional Gaussian Bump","authors":"Donald P. Rizzetta, Daniel J. Garmann","doi":"10.1080/10618562.2023.2246391","DOIUrl":"https://doi.org/10.1080/10618562.2023.2246391","url":null,"abstract":"AbstractWall-resolved large-eddy simulations were carried out for the flow over a Gaussian bump configuration. The geometry and flow conditions were motivated by an experimental investigation, which was conducted in order to provide data for validating numerical modelling. The present computations were initiated as benchmark results that are accessible via wall-resolved large-eddy simulation. It was found that by increasing the bump height, the Reynolds number could be reduced and flow separation would occur. The modified bump then serves as a surrogate for the original Gaussian bump producing a smooth separated flow. Solutions to the unsteady three-dimensional compressible Navier-Stokes equations were obtained utilising a high-fidelity computational scheme and an implicit time-marching approach. Large-eddy simulations were performed and grid resolution studies were carried to ensure quality of computed results. Features of the flowfields are elucidated, and it was found that the time-mean surface streamline pattern had similar features to that of the experiment.Keywords: Smooth-Body separationGaussian bumplarge-eddy simulationhigh-order numerical methodcompact-differencing scheme Disclosure statementNo potential conflict of interest was reported by the author(s).Additional informationFundingThis material is based upon work supported by the Air Force Office of Scientific Research under an award monitored by G. Abate. Computational resources were supported in part by grants of supercomputer time from the U. S. Department of Defense Supercomputing Resource Centers at the Stennis Space Center, MS, Vicksburg, MS, and Wright-Patterson AFB, OH.","PeriodicalId":56288,"journal":{"name":"International Journal of Computational Fluid Dynamics","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135799622","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A Lattice Boltzmann Front-Tracking Interface Capturing Method based on Neural Network for Gas-Liquid Two-Phase Flow 基于神经网络的气液两相流晶格Boltzmann前跟踪界面捕获方法

IF 1.3 4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2023-01-02 DOI: 10.1080/10618562.2023.2246398

Bozhen Lai, Zhaoqing Ke, Zhiqiang Wang, Ronghua Zhu, Ruifeng Gao, Yu Mao, Ying Zhang

{"title":"A Lattice Boltzmann Front-Tracking Interface Capturing Method based on Neural Network for Gas-Liquid Two-Phase Flow","authors":"Bozhen Lai, Zhaoqing Ke, Zhiqiang Wang, Ronghua Zhu, Ruifeng Gao, Yu Mao, Ying Zhang","doi":"10.1080/10618562.2023.2246398","DOIUrl":"https://doi.org/10.1080/10618562.2023.2246398","url":null,"abstract":"This paper presents a new method that accurately captures the interface of gas–liquid two-phase flow using a neural network-based lattice Boltzmann front-tracking interface capturing method. The motion of a single free-falling droplet is simulated using the Front Tracking Method (FTM), enabling the acquisition of information regarding the velocity field and interface points. The velocity field and interface points from the simulations are then utilised to generate input and output datasets for training the neural network (NN) models. Subsequently, the trained Bayesian regularised Back Propagation Neural Network (BRBPNN) model is integrated into the Lattice Boltzmann method (LBM), utilising the velocity field obtained from LBM simulation as input. The predicted LBM interface exhibits remarkable agreement with the FTM interface, as evidenced by a high correlation coefficient of 0.99945 for the ordinate value of the interface point in both methods. Therefore, the proposed method achieves precise positioning of the phase interface of LBM.","PeriodicalId":56288,"journal":{"name":"International Journal of Computational Fluid Dynamics","volume":"11 1","pages":"49 - 66"},"PeriodicalIF":1.3,"publicationDate":"2023-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"73241043","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Verification of a Pressure-Based Compressible Flow Solver 基于压力的可压缩流动求解器的验证

IF 1.3 4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2023-01-02 DOI: 10.1080/10618562.2023.2242271

J. Muralha, L. Eça, C. Klaij

引用次数: 0

Influence of the Downstream Vehicle Length on Train Aerodynamics Subjected to Crosswind 侧风作用下下游车辆长度对列车空气动力学的影响

IF 1.3 4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2022-12-31 DOI: 10.36959/717/661

Zhuang Tianci, Li Wenhui, Liu Tanghong

引用次数: 0

Thermal Radiation, Chemical Reaction and Viscous Dissipation Effects on MHD Mixed Convection Flow of Micro Polar Fluid with Stretching Surface in the Presence of Heat Generation/Absorption 热辐射、化学反应和粘滞耗散对生热/吸热条件下具有拉伸表面的微极流体MHD混合对流的影响

IF 1.3 4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2022-12-31 DOI: 10.36959/717/662

Zigta Binyam

引用次数: 0

Shape Optimization and Flow Analysis of Supersonic Nozzles Using Deep Learning 基于深度学习的超声速喷管形状优化与流动分析

IF 1.3 4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2022-11-26 DOI: 10.1080/10618562.2023.2225416

Aref Zanjani, A. Tahsini, Kimia Sadafi, Fatemeh Ghavidel Mangodeh

引用次数: 1

Finite Element Numerical Simulation of Local Scour of a Three-Dimensional Cylinder under Steady Flow 定常流动条件下三维圆柱局部冲刷的有限元数值模拟

IF 1.3 4区工程技术

International Journal of Computational Fluid Dynamics Pub Date : 2022-11-26 DOI: 10.1080/10618562.2023.2221645

Dawei Peng, Lanhao Zhao, Chuanyuan Zhou, Jia Mao

引用次数: 0