4.7 Article

Sensing the turbulent large-scale motions with their wall signature

Journal

PHYSICS OF FLUIDS
Volume 31, Issue 12, Pages -

Publisher

AIP Publishing
DOI: 10.1063/1.5128053

Keywords

-

Funding

  1. Spanish State Research Agency (SRA) [DPI2016-79401-R]
  2. European Regional Development Fund (ERDF)

Ask authors/readers for more resources

This study assesses the capability of extended proper orthogonal decomposition (EPOD) and convolutional neural networks (CNNs) to reconstruct large-scale and very-large-scale motions (LSMs and VLSMs respectively) employing wall-shear-stress measurements in wallb-ounded turbulent flows. Both techniques are used to reconstruct the instantaneous LSM evolution in the flow field as a combination of proper orthogonal decomposition (POD) modes, employing a limited set of instantaneous wall-shear-stress measurements. Due to the dominance of nonlinear effects, only CNNs provide satisfying results. Being able to account for nonlinearities in the flow, CNNs are shown to perform significantly better than EPOD in terms of both instantaneous flow-field estimation and turbulent-statistics reconstruction. CNNs are able to provide a more effective reconstruction performance employing more POD modes at larger distances from the wall and employing lower wall-measurement resolutions. Furthermore, the capability of tackling nonlinear features of CNNs results in estimation capabilities that are weakly dependent on the distance from the wall. Published under license by AIP Publishing.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.7
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

Article Thermodynamics

Inter-scale interaction in pipe flows at high Reynolds numbers

Xiaobo Zheng, Gabriele Bellani, Lucia Mascotelli, Ramis Orlu, Andrea Ianiro, Carlos Sanmiguel Vila, Stefano Discetti, Jacopo Serpieri, Marco Raiola, Alessandro Talamelli, Ye Li, Nan Jiang

Summary: Hot-wire measurements were conducted in the large-scale pipe-flow facility CICLoPE, showing Re-tau-independent modulation features and geometric characteristics. The study revealed the coherence and phase relations of different scale motions, with AM effects and opposite effects observed in different regions of the pipe flow.

EXPERIMENTAL THERMAL AND FLUID SCIENCE (2022)

Editorial Material Engineering, Multidisciplinary

Special issue on uncertainty quantification in particle image velocimetry and Lagrangian particle tracking

Andrea Sciacchitano, Stefano Discetti

MEASUREMENT SCIENCE AND TECHNOLOGY (2022)

Article Thermodynamics

Reducing turbulent convective heat transfer with streamwise plasma vortex generators

Rodrigo Castellanos, Theodoros Michelis, Stefano Discetti, Andrea Ianiro, Marios Kotsonis

Summary: This study experimentally investigates the effect of streamwise plasma vortex generators on the convective heat transfer of a turbulent boundary layer. The results show that the plasma-induced vortices are stationary and confined across the spanwise direction due to the action of the plasma discharge. The flow-field measurements reveal a low-velocity region caused by a mass- and momentum-flux deficit within the boundary layer, leading to a reduction in convective heat transfer. Near the wall, the plasma-induced jets divert the main flow, further decreasing the convective heat transfer.

EXPERIMENTAL THERMAL AND FLUID SCIENCE (2022)

Article Mechanics

Machine-learning flow control with few sensor feedback and measurement noise

R. Castellanos, G. Y. Cornejo Maceda, I de la Fuente, B. R. Noack, A. Ianiro, S. Discetti

Summary: This paper presents a comparative assessment of machine-learning methods for active flow control. The study focuses on drag reduction of a two-dimensional Karman vortex street past a circular cylinder at a low Reynolds number. The results show that both Deep Reinforcement Learning (DRL) and Linear Genetic Programming Control (LGPC) successfully reduce the drag and stabilize the vortex alley. DRL demonstrates higher robustness, while LGPC identifies compact and interpretable control laws using only a subset of sensors.

PHYSICS OF FLUIDS (2022)

Article Thermodynamics

Heat transfer enhancement in turbulent boundary layers with a pulsed slot jet in crossflow

Rodrigo Castellanos, Gianfranco Salih, Marco Raiola, Andrea Ianiro, Stefano Discetti

Summary: The experiment investigates the convective heat transfer enhancement in a turbulent boundary layer using a pulsed, slot jet in crossflow. A parametric study on actuation frequencies and duty cycles is conducted. The results show that both jet penetration and overall Nusselt number increase with increasing duty cycle. The flow topology is significantly altered by the jet pulsation, with a wall-attached jet rising from the slot accompanied by counter-rotating vortices.

APPLIED THERMAL ENGINEERING (2023)

Article Thermodynamics

An end-to-end KNN-based PTV approach for high-resolution measurements and uncertainty quantification

Iacopo Tirelli, Andrea Ianiro, Stefano Discetti

Summary: We present a novel end-to-end approach to enhance the resolution of Particle Image Velocimetry (PIV) measurements. Our method utilizes information from different snapshots to obtain high-resolution flow fields and uncertainty estimations with minimal user intervention.

EXPERIMENTAL THERMAL AND FLUID SCIENCE (2023)

Article Mechanics

From snapshots to manifolds - a tale of shear flows

E. Farzamnik, A. Ianiro, S. Discetti, N. Deng, K. Oberleithner, B. R. Noack, V. Guerrero

Summary: We propose a novel nonlinear manifold learning method using Isomap as encoder and K-nearest neighbours algorithm as decoder, and demonstrate its superiority over POD for shedding-dominated shear flows. The method is applied to numerical and experimental datasets including fluidic pinball, swirling jet and wake behind tandem cylinders, and is able to describe the bifurcation, chaotic regime and shedding phases of the flow. The reconstruction error of the manifold model is small, indicating that the low embedding dimensions contain the coherent structure dynamics.

JOURNAL OF FLUID MECHANICS (2023)

Article Engineering, Multidisciplinary

Machine learning for flow field measurements: a perspective

Stefano Discetti, Yingzheng Liu

Summary: Advancements in machine-learning techniques are driving a paradigm shift in image processing, and optical techniques play an important role in flow diagnostics. This perspective reviews the recent advancements in machine learning methods for flow field measurements and highlights possible routes for further developments.

MEASUREMENT SCIENCE AND TECHNOLOGY (2023)

Article Thermodynamics

Genetically-inspired convective heat transfer enhancement in a turbulent boundary layer

Rodrigo Castellanos, Andrea Ianiro, Stefano Discetti

Summary: The convective heat transfer in a turbulent boundary layer (TBL) on a flat plate is enhanced using an artificial intelligence approach based on linear genetic algorithms control (LGAC). The actuator is a set of six slot jets in crossflow aligned with the freestream. The optimal controller yields a slightly asymmetric flow field and the LGAC algorithm converges to the same frequency and duty cycle for all the actuators. The results pinpoint the potential of machine learning control in unravelling unexplored controllers within the actuation space.

APPLIED THERMAL ENGINEERING (2023)

Article Thermodynamics

A simple trick to improve the accuracy of PIV/PTV data

Iacopo Tirelli, Andrea Ianiro, Stefano Discetti

Summary: Particle Image Velocimetry (PIV) estimates velocities through particle image correlations, which leads to a modulation effect. To exploit the scattered data from Particle Tracking Velocimetry (PTV), interpolation on a structured grid is necessary, causing spatial modulation bias. A technique called Ensemble Particle Tracking Velocimetry (EPTV) is introduced to reduce this systematic error by merging different instantaneous realizations and obtaining high-resolution mean flow. The methodology is validated against various datasets with increasing complexity, using PTV and PIV analysis.

EXPERIMENTAL THERMAL AND FLUID SCIENCE (2023)

Article Computer Science, Artificial Intelligence

Super-resolution generative adversarial networks of randomly-seeded fields

Alejandro Guemes, Carlos Sanmiguel Vila, Stefano Discetti

Summary: In this paper, a super-resolution generative adversarial network framework is proposed to estimate field quantities from random sparse sensors. The algorithm utilizes random sampling to provide incomplete views of the high-resolution underlying distributions and has been tested on synthetic databases, showing excellent performance.

NATURE MACHINE INTELLIGENCE (2022)

No Data Available