Pre-processing DNS data to improve statistical convergence and accuracy of mean velocity fields in invariant data-driven turbulence models

Fonseca, Eduardo; Rangel, Victor B.; Brener, Bernardo Pereira; Cruz, Matheus; Thompson, Roney L.

doi:10.1007/s00162-022-00603-4

Cited by 6 publications

(6 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The success in this strategy lies in the fact that the RFV is computed exclusively using the velocity data, by avoiding the use of RST data considerable gains in accuracy were achieved. The disparity between using the RFV over the RST was demonstrated by Brener et al 26 to be independent of numerical strategy employed in the propagation to the RANS equations, which was corroborated by multiple works 40–42 …”

Section: Introductionmentioning

confidence: 83%

“…Generally speaking, it has been extensively demonstrated that the improvements of the data-driven correction techniques are attached to the use of high-fidelity mean velocity data. 26,31,[39][40][41][42] That is to say that the greater the participation of the mean velocities in calculating the ML targets, the lower the error propagation is in the corrections. Our model does not completely avoid the use of the second-order statistics of the RST but rather combines it with first-order statistics of the mean flow and with the benefits of using it.…”

Section: Contributions Of the Present Workmentioning

confidence: 99%

“…The disparity between using the RFV over the RST was demonstrated by Brener et al 26 to be independent of numerical strategy employed in the propagation to the RANS equations, which was corroborated by multiple works. [40][41][42] Alternate methods of using or predicting the RFV were recently evaluated, Berrone and Oberto 43 adapted the TBNN 29 into a vector basis NN to predict it and correct the baseline RANS mean flows in the SD and PH. Amarloo et al 42 assessed the a priori performance of two frozen propagation methods for the RFV in flows with secondary motion.…”

Section: Introduction 11 Reynolds Stress Transport Modelmentioning

confidence: 99%

“…The methodology introduced in this work is a step further in complexity and physical stringency, analogous to the RSM in RANS modeling approaches. Generally speaking, it has been extensively demonstrated that the improvements of the data‐driven correction techniques are attached to the use of high‐fidelity mean velocity data 26,31,39–42 . That is to say that the greater the participation of the mean velocities in calculating the ML targets, the lower the error propagation is in the corrections.…”

Section: Introductionmentioning

confidence: 99%

“…To evaluate and demonstrate the model, NN were trained using the DNS database of the SD flow by Pinelli et al, 51 preprocessed by Fonseca et al 41 The achievements of the proposed methodology are compared with the direct prediction and injection of the RST into the RANS equations.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A data‐driven turbulence modeling for the Reynolds stress tensor transport equation

Macedo,

Cruz,

Brener

et al. 2024

Numerical Methods in Fluids

Self Cite

View full text Add to dashboard Cite

The long lasting demand for better turbulence models and the still prohibitively computational cost of high‐fidelity fluid dynamics simulations, like direct numerical simulations and large eddy simulations, have led to a rising interest in coupling available high‐fidelity datasets and popular, yet limited, Reynolds averaged Navier–Stokes simulations through machine learning (ML) techniques. Many of the recent advances used the Reynolds stress tensor or, less frequently, the Reynolds force vector as the target for these corrections. In the present work, we considered an unexplored strategy, namely to use the modeled terms of the Reynolds stress transport equation as the target for the ML predictions, employing a neural network approach. After that, we solve the coupled set of governing equations to obtain the mean velocity field. We apply this strategy to solve the flow through a square duct. The obtained results consistently recover the secondary flow, which is not present in the baseline simulations that used the model. The results were compared with other approaches of the literature, showing a path that can be useful in the seek of more universal models in turbulence.

show abstract

Section: Introductionmentioning

confidence: 83%