For flow-related design optimization problems, e.g., aircraft and automobile aerodynamic design, com-putational fluid dynamics (CFD) simulations are commonly used to predict flow fields and analyze performance. While important, CFD simulations are a resource-demanding and time-consuming iterative process. The expensive simulation overhead limits the opportunities for large design space exploration and prevents interactive design. In this paper, we propose FlowDNN, a novel deep neural network (DNN) to efficiently learn flow representations from CFD results. FlowDNN saves computational time by directly predicting the expected flow fields based on given flow conditions and geometry shapes. FlowDNN is the first DNN that incorporates the underlying physical conservation laws of fluid dynamics with a carefully designed attention mechanism for steady flow prediction. This approach not only improves the prediction accuracy, but also preserves the physical consistency of the predicted flow fields, which is essential for CFD. Various metrics are derived to evaluate FlowDNN with respect to the whole flow fields or regions of interest (RoIs) (e.g., boundary layers where flow quantities change rapidly). Experiments show that FlowDNN significantly outperforms alternative methods with faster inference and more accurate results. It speeds up a graphics processing unit (GPU) accelerated CFD solver by more than 14000×, while keeping the prediction error under 5％.

Donglin CHEN;Xiang GAO;Chuanfu XU;Siqi WANG;Shizhao CHEN;Jianbin FANG;Zheng WANG

College of Computer,National University of Defense Technology,Changsha 410073,China;State Key Laboratory of High Performance Computing,National University of Defense Technology,Changsha 410073,China;School of Computing,University of Leeds,Leeds LS29JT,UK

信息与电子工程前沿（英文）

[1]Donglin CHEN;Xiang GAO;Chuanfu XU;Siqi WANG;Shizhao CHEN;Jianbin FANG;Zheng WANG-.FlowDNN:a physics-informed deep neural network for fast and accurate flow prediction)[J].信息与电子工程前沿（英文）,2022(02):207-219

FlowDNN,RoIs

physics,informed,deep,neural,network,accurate,flow,prediction,For,related,optimization,problems,aircraft,automobile,aerodynamic,fluid,dynamics,CFD,simulations,commonly,used,fields,analyze,performance,While,important,resource,demanding,consuming,iterative,expensive,overhead,limits,opportunities,large,space,exploration,prevents,interactive,In,this,paper,we,propose,novel,efficiently,learn,representations,from,results,saves,computational,by,directly,predicting,expected,given,conditions,geometry,shapes,first,that,incorporates,underlying,physical,conservation,laws,carefully,designed,attention,mechanism,steady,This,approach,not,improves,accuracy,but,also,preserves,consistency,predicted,which,essential,Various,metrics,derived,evaluate,respect,whole,regions,interest,boundary,layers,where,quantities,change,rapidly,Experiments,show,significantly,outperforms,alternative,methods,faster,inference,more,It,speeds,up,graphics,processing,GPU,accelerated,solver,than,while,keeping,error

0.611676