Fast aerodynamics prediction of laminar airfoils based on deep attention network

Physics of Fluids, Год журнала: 2023, Номер 35(3)

Опубликована: Март 1, 2023

The traditional method for obtaining aerodynamic parameters of airfoils by solving Navier–Stokes equations is a time-consuming computing task. In this article, novel data-driven deep attention network (DAN) proposed reconstruction incompressible steady flow fields around airfoils. To extract the geometric representation input airfoils, grayscale image airfoil divided into set patches, and these are transformer encoder embedding. extracted from encoder, together with Reynolds number, angle attack, field coordinates, distance field, multilayer perceptron to predict airfoil. Through analysis large number qualitative quantitative experimental results, it concluded that DAN can improve interpretability model while good prediction accuracy generalization capability different flow-field states.

Язык: Английский

---

Reinforcement learning for wind-farm flow control: Current state and future actions

Theoretical and Applied Mechanics Letters, Год журнала: 2023, Номер 13(6), С. 100475 - 100475

Опубликована: Окт. 20, 2023

Wind-farm flow control stands at the forefront of grand challenges in wind-energy science. The central issue is that current algorithms are based on simplified models and, thus, fall short capturing complex physics wind farms associated with high-dimensional nature turbulence and multiscale wind-farm-atmosphere interactions. Reinforcement learning (RL), as a subset machine learning, has demonstrated its effectiveness solving problems various domains, studies performed last decade prove it can be exploited development next generation for wind-farm control. This review two main objectives. Firstly, aims to provide an up-to-date overview works focusing schemes utilizing RL methods. By examining latest research this area, seeks offer comprehensive understanding advancements made through application techniques. Secondly, shed light obstacles researchers face when implementing RL. highlighting these challenges, identify areas requiring further exploration potential opportunities future research.

Язык: Английский

---

Constrained re-calibration of two-equation Reynolds-averaged Navier–Stokes models

Theoretical and Applied Mechanics Letters, Год журнала: 2024, Номер 14(2), С. 100503 - 100503

Опубликована: Фев. 7, 2024

Machine-learned augmentations to turbulence models can be advantageous for flows within the training dataset but often cause harm outside. This lack of generalizability arises because constants (as well as functions) in a Reynolds-averaged Navier–Stokes (RANS) model are coupled, and un-constrained re-calibration these (and disrupt calibrations baseline model, preservation which is critical model's generalizability. To safeguard behaviors beyond dataset, machine learning must constrained such that basic like law wall kept intact. letter aims identify constraints two-equation RANS so future work performed without violating constraints. We demonstrate identified not limiting. Furthermore, they help preserve model.

Язык: Английский

---

Physics-Guided Machine Learning for Wind-Farm Power Prediction: Toward Interpretability and Generalizability

PRX Energy, Год журнала: 2023, Номер 2(1)

Опубликована: Март 30, 2023

A machine-learning model is developed and used to predict the performance of individual wind turbines in farms; strategy leads an accurate, lightweight, generalizable data-driven for wind-farm power prediction.

Язык: Английский

---

Log-law recovery through reinforcement-learning wall model for large eddy simulation

Physics of Fluids, Год журнала: 2023, Номер 35(5)

Опубликована: Май 1, 2023

This paper focuses on the use of reinforcement learning (RL) as a machine-learning (ML) modeling tool for near-wall turbulence. RL has demonstrated its effectiveness in solving high-dimensional problems, especially domains such games. Despite potential, is still not widely used turbulence and primarily flow control optimization purposes. A new wall model (WM) called VYBA23 developed this work, which uses agents dispersed near wall. The trained single Reynolds number (Reτ=104) does rely high-fidelity data, backpropagation process based reward rather than an output error. states RLWM, are representation environment by agents, normalized to remove dependence number. tested compared another RLWM (BK22) equilibrium model, half-channel at eleven different numbers {Reτ∈[180;1010]}. effects varying agents' parameters, actions range, time step, spacing, also studied. results promising, showing little effect average field but some wall-shear stress fluctuations velocity fluctuations. work offers positive prospects developing RLWMs that can recover physical laws extending type ML models more complex flows future.

Язык: Английский

---

Frozen propagation of Reynolds force vector from high-fidelity data into Reynolds-averaged simulations of secondary flows

Physics of Fluids, Год журнала: 2022, Номер 34(11)

Опубликована: Окт. 11, 2022

Successful propagation of information from high-fidelity sources (i.e., direct numerical simulations and large-eddy simulations) into Reynolds-averaged Navier–Stokes (RANS) equations plays an important role in the emerging field data-driven RANS modeling. Small errors carried data can propagate amplified mean flow field, higher Reynolds numbers worsen error propagation. In this study, we compare a series methods for two cases Prandtl's secondary flows second kind: square-duct at low number roughness-induced very high number. We show that frozen treatments result less than implicit treatment stress tensor (RST), with numbers, explicit are not recommended. Inspired by obtained results, introduce to force vector (RFV), which leads Specifically, both RFV results one order magnitude lower compared RST method, three different eddy-viscosity models used evaluate effect turbulent diffusion on that, regardless baseline model, combined extra correction term kinetic energy RFV, makes our technique capable reproducing velocity fields similar data.

Язык: Английский

---

Validating the design optimisation of ultrasonic flow meters using computational fluid dynamics and surrogate modelling

International Journal of Heat and Fluid Flow, Год журнала: 2023, Номер 100, С. 109112 - 109112

Опубликована: Янв. 23, 2023

Domestic ultrasonic flow meters with an intrusive two-stand configuration present a complex behaviour due to their unique geometry, which offers interesting case evaluate optimisation methods in wall-bounded turbulent flows. In this study, the design and analysis of computer models by computational fluid dynamics is used predict perform robust meter. The accomplished surrogate modelling based on Kriging, Latin hypercube sampling, Bayesian strategies ensure high-quality space-filled response surface. A novel function quantify meter measurement uncertainty defined evaluated together pressure drop order define multi-objective problem. Pareto front shown compared numerically experimentally against laser Doppler velocimetry experiments, displaying performance gains geometrical changes 3D space. From various improved designs sampled experimentally, 4.9% reduction 37.4% have been analysed baseline case. applied methodology provides efficient framework changes, improving internal-flow problems similar features.

Язык: Английский

---

Data-Enabled Recalibration of the Spalart–Allmaras Model

AIAA Journal, Год журнала: 2023, Номер 61(11), С. 4852 - 4863

Опубликована: Июль 13, 2023

We use experimental and simulation data to recalibrate the standard Spalart–Allmaras model. Free-shear flow, buffer layer, log flows with adverse pressure gradients are targeted. In this process, recalibration does not affect untargeted flows. Our approach uses Bayesian optimization feedforward neural networks. The recalibrated model is implemented in two codes tested 11 flows: majority of which outside training dataset have geometries that distinctly different from those dataset. show data-enabled offers improvements while preserving model’s existing good behavior. particular, our improves behavior separated its behaviors flat-plate boundary-layer channel Further analysis indicates flow mainly due [Formula: see text] function resulting, more precise representation “slingshot” effect.

Язык: Английский

---

Fluid classification through well logging is conducted using the extreme gradient boosting model based on the adaptive piecewise flatness-based fast transform feature extraction algorithm

Physics of Fluids, Год журнала: 2024, Номер 36(1)

Опубликована: Янв. 1, 2024

In recent years, fluid prediction through well logging has assumed a pivotal role in the realm of oil and gas exploration. Seeking to enhance accuracy, this paper introduces an adaptive piecewise flatness-based fast transform (APFFT) algorithm conjunction with XGBoost (extreme gradient boosting) method for prediction. Initially, APFFT technology is employed extract frequency-domain features from data. This dynamically determines optimal frequency interval, transforming raw curves into domain process enhances preservation information reflective characteristics, simultaneously minimizing impact noise non-fluid compositions. Subsequently, acquired are utilized as inputs construct model To validate efficacy proposed approach, real data were collected, extensive experimental evaluation was conducted. The findings underscore substantial advantages APFFT-XGBoost over traditional machine learning models such XGBoost, random forest, K-nearest neighbor algorithm, support vector machine, backpropagation neural network demonstrates ability accurately capture features, leading improved accuracy stability.

Язык: Английский

---

Tidal turbine blade design optimization based on coupled deep learning and blade element momentum theory

Physics of Fluids, Год журнала: 2024, Номер 36(5)

Опубликована: Май 1, 2024

The practical design optimization of blade structures is crucial for enhancing the power capture capability tidal turbines. However, significant computational costs required directly optimizing turbine blades through numerical simulations limit application structure optimization. This paper proposes a framework based on deep learning (DL) and element momentum (BEM). employs control points to parameterize three-dimensional geometric shape blades, uses convolutional neural networks predict hydrodynamic performance each hydrofoil section, couples BEM forecast blades. multi-objective non-dominated sorting genetic algorithm II employed optimize parameters maximize coefficient minimize thrust coefficient, aiming obtain optimal trade-off solution. results indicate that prediction DL-BEM model agrees well with experimental data, significantly improving efficiency. optimized exhibit excellent coefficients reduced coefficients, achieving more balanced structural proposed DL accurately rapidly predicts turbines, facilitating high-performance

Язык: Английский

---