Prediction of aerodynamic flow fields using convolutional neural networks

Bhatnagar, Saakaar; Afshar, Yaser; Pan, Shaowu; Duraisamy, Karthik; Kaushik, Shubham

doi:10.1007/s00466-019-01740-0

Cited by 395 publications

(216 citation statements)

References 46 publications

Supporting

Mentioning

195

Contrasting

Order By: Relevance

“…For this purpose, inputs (iii) and (iv) are provided to the CNN in the form of distance functions for objects and for monopoles. Such an approach has previously been used for CNNs to predict steady-state flow fields [ 3 , 7 ]. The distance functions are defined by i.e., for each cell with location ( i , j ) in a domain the minimal distances and to the boundary of an object and to a monopole M are determined.…”

Section: Methodsmentioning

confidence: 99%

“…For this purpose, inputs (iii) and (iv) are provided to the CNN in the form of distance functions Φ o for objects and Φ m for monopoles. Such an approach has previously been used for CNNs to predict steady-state flow fields [3,7]. The distance functions are defined by…”

Section: Machine Learning Techniquesmentioning

confidence: 99%

“…In this case, the period of averaging needs to be bridged before the results can be analyzed. To overcome this issue, methods to accelerate the prediction of steady flow fields using convolutional neural networks (CNNs) are studied [3,7]. In [7], the flow over simplified vehicle bodies is predicted with CNNs.…”

Section: Introductionmentioning

confidence: 99%

“…The corresponding surrogate model is considerably faster than traditional flow solvers. In [3], CNNs are successfully applied to predict flow fields around airfoils with varying angles of attack and Reynolds numbers. Lee and You [16] predict the unsteady flow over a circular cylinder using DL methods.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Prediction of Acoustic Fields Using a Lattice-Boltzmann Method and Deep Learning

Rüttgers

Koh

Jitsev

et al. 2020

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Using traditional computational fluid dynamics and aeroacoustics methods, the accurate simulation of aeroacoustic sources requires high compute resources to resolve all necessary physical phenomena. In contrast, once trained, artificial neural networks such as deep encoder-decoder convolutional networks allow to predict aeroacoustics at lower cost and, depending on the quality of the employed network, also at high accuracy. The architecture for such a neural network is developed to predict the sound pressure level in a 2D square domain. It is trained by numerical results from up to 20,000 GPU-based lattice-Boltzmann simulations that include randomly distributed rectangular and circular objects, and monopole sources. Types of boundary conditions, the monopole locations, and cell distances for objects and monopoles serve as input to the network. Parameters are studied to tune the predictions and to increase their accuracy. The complexity of the setup is successively increased along three cases and the impact of the number of feature maps, the type of loss function, and the number of training data on the prediction accuracy is investigated. An optimal choice of the parameters leads to network-predicted results that are in good agreement with the simulated findings. This is corroborated by negligible differences of the sound pressure level between the simulated and the network-predicted results along characteristic lines and by small mean errors.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Machine Learning Techniquesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Prediction of Acoustic Fields Using a Lattice-Boltzmann Method and Deep Learning

Rüttgers

Koh

Jitsev

et al. 2020

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…Despite the disadvantage of a large amount of data needed for training, ANNs appears in several works [4][5][6] in the field of loads and damage monitoring in the helicopter area. In the last few decades, these techniques have become widely used for force reconstruction [7,8] and aerodynamic flow interaction [9,10] problems. One of the main advantages of ANN is that no physical model is required behind the application of the methodology.…”

Section: Introductionmentioning

confidence: 99%

State and Force Estimation on a Rotating Helicopter Blade through a Kalman-Based Approach

Cumbo

Tamarozzi

Jiránek

et al. 2020

Sensors

View full text Add to dashboard Cite

The interaction between the rotating blades and the external fluid in non-axial flow conditions is the main source of vibratory loads on the main rotor of helicopters. The knowledge or prediction of the produced aerodynamic loads and of the dynamic behavior of the components could represent an advantage in preventing failures of the entire rotorcraft. Some techniques have been explored in the literature, but in this field of application, high accuracy can be reached if a large amount of sensor data and/or a high-fidelity numerical model is available. This paper applies the Kalman filtering technique to rotor load estimation. The nature of the filter allows the usage of a minimum set of sensors. The compensation of a low-fidelity model is also possible by accounting for sensors and model uncertainties. The efficiency of the filter for state and load estimation on a rotating blade is tested in this contribution, considering two different sources of uncertainties on a coupled multibody-aerodynamic model. Numerical results show an accurate state reconstruction with respect to the selected sensor layout. The aerodynamic loads are accurately evaluated in post-processing.

show abstract

Finite volume method network for the acceleration of unsteady computational fluid dynamics: Non‐reacting and reacting flows

Jeon

Lee

Kim

2022

Intl J of Energy Research

View full text Add to dashboard Cite

Summary Despite rapid improvements in the performance of the central processing unit (CPU), the calculation cost of simulating chemically reacting flow using CFD remains infeasible in many cases. The application of the convolutional neural networks (CNNs) specialized in image processing in flow field prediction has been studied, but the need to develop a neural network design fitted for CFD has recently emerged. In this study, a neural network model introducing the finite volume method (FVM) with unique network architecture and physics‐informed loss function was developed to accelerate CFD simulations. The developed network model, considering the nature of the CFD flow field where the identical governing equations are applied to all grids, can predict the future fields with only two previous fields unlike the CNNs requiring many field images (>10 000). The performance of this baseline model was evaluated using CFD time series data from non‐reacting flow and reacting flow simulation; counterflow and hydrogen flame with 20 detailed chemistries. Consequently, we demonstrated that (a) the FVM‐based network architecture provided significantly improved accuracy of multistep time series prediction compared to the previous MLP model (b) the physic‐informed loss function prevented non‐physical overfitting problem and ultimately reduced the error in time series prediction (c) observing the calculated residuals in an unsupervised manner could monitor the network accuracy. Additionally, under the reacting flow dataset, the computational speed of this network model was measured to be about 10 times faster than that of the CFD solver.

show abstract

Prediction of aerodynamic flow fields using convolutional neural networks

Cited by 395 publications

References 46 publications

Prediction of Acoustic Fields Using a Lattice-Boltzmann Method and Deep Learning

Prediction of Acoustic Fields Using a Lattice-Boltzmann Method and Deep Learning

State and Force Estimation on a Rotating Helicopter Blade through a Kalman-Based Approach

Finite volume method network for the acceleration of unsteady computational fluid dynamics: Non‐reacting and reacting flows

Contact Info

Product

Resources

About