arXiv subjects

Antonio Attili

Publications and source records attributed to Antonio Attili.

11 recordsLinked to original sources

arXiv2026

Effect of transverse gust on free-falling plates

The effects of transverse gusts on free-falling plates are investigated using two-way coupled fluid-structure interaction simulations for a Galilei number (Ga) between 10 and 50 and a density ratio (rho) between 5 and 50. We consider gust ratios (GR) of up to 5, where GR is the ratio of the free-stream velocity change to an estimate of the terminal velocity. We demonstrate that the plate experiences the gust as a transient horizontal force, which displaces it horizontally. This results in a transient reduction in the angle of attack, an increase in absolute velocity and the generation of circulation. The vertical component of the latter increases the upward aerodynamic force, slowing down the vertical descent of the plate. Furthermore, the plate's horizontal displacement with respect to its original wake results in a further transient increase in the upward aerodynamic force. The altitude gained by the plate in response to the gust is maximum for rho=15, and increases non-monotonically with Ga and GR. The non-monotonic trend is due to plate pitch: if the maximum pitch of the plate in response to the gust is close to vertical, the plate temporarily falls faster, losing some of the altitude it has gained. The present findings reveal an energy-harvesting mechanism that free-falling bodies can exploit to increase their time afloat.

physics.flu-dyn↗

arXiv2026

Effect of a magnetostatic field on laminar premixed hydrogen-air flames

Magnetic fields have shown potential to affect flame characteristics; however, the mechanisms of interaction are not fully understood. This paper investigates the effect of magnetic fields on premixed hydrogen-air flames that are prone to intrinsic instabilities, with a focus on the role of magnetic forces on the flame behaviour. The study is conducted using direct numerical simulations. Two flame conditions, both with an equivalence ratio of 0.5, are studied, one with the reactants at atmospheric conditions and the other at high pressure and high temperature. Different configurations of the magnetic field are investigated, each characterised by a different gradient of the square of the magnitude of the magnetic field, oriented in the direction opposite to the velocity of the incoming reactants. Results show that the investigated configurations of the magnetic field can reduce the flame consumption speed, an effect that is substantial in the lower pressure case, while it becomes negligible at high pressure. The effect of the magnetic forces increases with increasing gradient of the magnetic field and is mainly due to the reduction of the flame area. Results also show that the effects of magnetic fields on the reactivity of the flame and on the small cell structures developed along the flame front are negligible. Analysis of the force contributions demonstrates that the change in the flame area is caused by the rotational component of the magnetic forces, which alter the vorticity of the flow such that the finger-like structures formed by hydrodynamic instabilities tend to close. These forces are significant at low pressure, while they become negligible compared to the pressure gradient at high pressure. Ultimately, the results of this work indicate that magnetic forces have the potential to change the flame behaviour, a mechanism that could be used for active control of flames.

physics.flu-dyn↗

arXiv2026

Turbulent hydrogen premixed flames at high pressure and high temperature

The combined influence of elevated pressure and temperature, representative of gas-turbine operating conditions, on lean premixed hydrogen flames is investigated using Direct Numerical Simulations (DNS) of a turbulent jet. Three cases are considered: 1 atm/298 K, 5 atm/472 K, and 20 atm/700 K, scaled to maintain the same jet Reynolds number and nominal Karlovitz number in the unburnt mixture, enabling a direct comparison of flame-turbulence interactions. Although the combined effects are moderate overall due to compensating influences, measurable differences arise in flame structure and turbulence-flame coupling. They are driven by reduced turbulence dissipation within the flame at high pressure and temperature, which enhances the interaction between turbulence and thermodiffusive effects. Finally, the tangential strain rate exhibits the same universal Kolmogorov scaling observed in homogeneous-isotropic turbulence and in methane flames, confirming its robustness for modelling turbulence

physics.flu-dyn↗

arXiv2025

Dynamic mixed turbulence modeling using a super-resolution generative adversarial approach

A dynamic mixed super-resolution model (DMSRM) for large-eddy simulations (LESs) is proposed, which combines the traditional dynamic mixed model (DMM) formulation with the generation of super-resolved velocity fields from which the subfilter-scale (SFS) stress tensor can be computed. A data-driven super-resolution generative adversarial network (SR-GAN) is employed to upsample the grid-filtered velocity fields by a factor of two, enabling the evaluation of both scale-similarity and the dynamic Smagorinsky contributions. A priori analyses of forced homogeneous isotropic turbulence show that the SR-GAN accurately reconstructs fine-scale flow features and generalizes well across different filter sizes and higher Reynolds number flow configurations, even for unseen input fields. The DMSRM reproduces SFS stresses and energy dissipation more accurately than the traditional DMM. A posteriori LES calculations further confirm that DMSRM predicts the energy spectrum and intermittency more accurately than DMM, even for different LES grid-scale resolutions and for higher Reynolds numbers than those used for training. Unlike DMM, DMSRM yields realistic backscatter and physically consistent SFS energy dissipation. These improvements arise from the physically accurate super-resolved fields generated by the SR-GAN, from which SFS stresses are directly computed. The result is a closure that accurately reproduces stress magnitudes and dissipation while reducing reliance on additional dissipation from the dynamic term. The DMSRM formulation achieves a balance of physical fidelity, robustness, and computational efficiency, offering a promising alternative to traditional DMMs for turbulence LES modeling.

physics.flu-dyn↗

arXiv2025

ZipGAN: Super-Resolution-based Generative Adversarial Network Framework for Data Compression of Direct Numerical Simulations

The advancement of high-performance computing has enabled the generation of large direct numerical simulation (DNS) datasets of turbulent flows, driving the need for efficient compression/decompression techniques that reduce storage demands while maintaining fidelity. Traditional methods, such as the discrete wavelet transform (DWT), cannot achieve compression ratios of 8 or higher for complex turbulent flows without introducing significant encoding/decoding errors. On the other hand, a super-resolution-based generative adversarial network (SR-GAN), called ZipGAN, can accurately reconstruct fine-scale features, preserving velocity gradients and structural details, even at a compression ratio of 512, thanks to the more efficient representation of the data in a compact latent space. Additional benefits are ascribed to adversarial training. The high GAN training time is significantly reduced with a progressive transfer learning approach and, once trained, they can be applied independently of the Reynolds number. It is demonstrated that ZipGAN can enhance dataset temporal resolution without additional simulation overhead by generating high-quality intermediate fields from compressed snapshots. The ZipGAN discriminator can reliably evaluate the quality of decoded fields, ensuring fidelity even in the absence of original DNS fields. Hence, ZipGAN compression/decompression method presents a highly efficient and scalable alternative for large-scale DNS storage and transfer, offering substantial advantages over the DWT methods in terms of compression efficiency, reconstruction fidelity, and temporal resolution enhancement.

physics.flu-dyn↗

arXiv2024

Influence of adversarial training on super-resolution turbulence reconstruction

Supervised super-resolution deep convolutional neural networks (CNNs) have gained significant attention for their potential in reconstructing velocity and scalar fields in turbulent flows. Despite their popularity, CNNs currently lack the ability to accurately produce high-frequency and small-scale features, and tests of their generalizability to out-of-sample flows are not widespread. Generative adversarial networks (GANs), which consist of two distinct neural networks (NNs), a generator and discriminator, are a promising alternative, allowing for both semi-supervised and unsupervised training. The difference in the flow fields produced by these two NN architectures has not been thoroughly investigated, and a comprehensive understanding of the discriminator's role has yet to be developed. This study assesses the effectiveness of the unsupervised adversarial training in GANs for turbulence reconstruction in forced homogeneous isotropic turbulence. GAN-based architectures are found to outperform supervised CNNs for turbulent flow reconstruction for in-sample cases. The reconstruction accuracy of both architectures diminishes for out-of-sample cases, though the GAN's discriminator network significantly improves the generator's out-of-sample robustness using either an additional unsupervised training step with large eddy simulation input fields and a dynamic selection of the most suitable upsampling factor. These enhance the generator's ability to reconstruct small-scale gradients, turbulence intermittency, and velocity-gradient probability density functions. The extrapolation capability of the GAN-based model is demonstrated for out-of-sample flows at higher Reynolds numbers. Based on these findings, incorporating discriminator-based training is recommended to enhance the reconstruction capability of super-resolution CNNs.

physics.flu-dyn↗

arXiv2023

Efficient and accurate calculation of dispersion relations for intrinsically unstable premixed flames

Premixed flames are susceptible to hydrodynamic and thermodiffusive instabilities that wrinkle the flame front and lead to complex multiscale patterns. They strongly impact the flame propagation and dynamics, increasing the speed of a laminar flame by several folds, easily as large as a factor of five for lean hydrogen flames at high pressure. The dispersion relation, which represents the growth rate of the different harmonic components of the perturbation of the flame front for different wavelengths, is useful to understand the dynamics during the linear phase of flame instabilities. In this work, an efficient and accurate approach based on a Fourier analysis of flame wrinkling is proposed to calculate the dispersion relation. Differently from the typical approach based on perturbing the flame with a single wavelength, the flame is perturbed with a spectrum of sine waves and their growth is followed with a spectral analysis. With the present method, the full dispersion relation is computed with a single simulation; this is significantly more efficient computationally than running a series of simulations with a single-wavelength perturbation for each point of the dispersion relation. It is shown that the presented approach is accurate and also solves an issue encountered when a single perturbation is imposed to compute the growth rate of large wavelengths. Several numerical and initialisation parameters, including resolution, domain size, and amplitude of the initial perturbation, are studied systematically and assessed.

physics.flu-dyn↗

arXiv2022

Local manifold learning and its link to domain-based physics knowledge

In many reacting flow systems, the thermo-chemical state-space is known or assumed to evolve close to a low-dimensional manifold (LDM). Various approaches are available to obtain those manifolds and subsequently express the original high-dimensional space with fewer parameterizing variables. Principal component analysis (PCA) is one of the dimensionality reduction methods that can be used to obtain LDMs. PCA does not make prior assumptions about the parameterizing variables and retrieves them empirically from the training data. In this paper, we show that PCA applied in local clusters of data (local PCA) is capable of detecting the intrinsic parameterization of the thermo-chemical state-space. We first demonstrate that utilizing three common combustion models of varying complexity: the Burke-Schumann model, the chemical equilibrium model and the homogeneous reactor. Parameterization of these models is known a priori which allows for benchmarking with the local PCA approach. We further extend the application of local PCA to a more challenging case of a turbulent non-premixed $n$-heptane/air jet flame for which the parameterization is no longer obvious. Our results suggest that meaningful parameterization can be obtained also for more complex datasets. We show that local PCA finds variables that can be linked to local stoichiometry, reaction progress and soot formation processes.

physics.flu-dyn↗

arXiv2022

Predictive data-driven model based on generative adversarial network for premixed turbulence-combustion regimes

Premixed flames exhibit different asymptotic regimes of interaction between heat release and turbulence depending on their respective length scales. At high Karlovitz number, the dilatation caused by heat release does not have any relevant effect on turbulent kinetic energy with respect to non-reacting flow, while at low Karlovitz number, the mean shear is a sink of turbulent kinetic energy, and counter-gradient transport is observed. This latter phenomenon is not well captured by closure models commonly used in Large Eddy Simulations that are based on gradient diffusion. The massive amount of data available from Direct Numerical Simulation (DNS) opens the possibility to develop data-driven models able to represent physical mechanisms and non-linear features present in both these regimes. In this work, the databases are formed by DNSs of two planar hydrogen/air flames at different Karlovitz numbers corresponding to the two asymptotic regimes. In this context, the Generative Adversarial Network (GAN) gives the possibility to successfully recognize and reconstruct both gradient and counter-gradient phenomena if trained with databases where both regimes are included. Two GAN models were first trained each for a specific Karlovitz number and tested using the same dataset in order to verify the capability of the models to learn the features of a single asymptotic regime and assess its accuracy. In both cases, the GAN models were able to reconstruct the Reynolds stress subfilter scales accurately. Later, the GAN was trained with a mixture of both datasets to create a model containing physical knowledge of both combustion regimes. This model was able to reconstruct the subfilter scales for both cases capturing the interaction between heat release and turbulence closely to the DNS as shown from the turbulent kinetic budget and barycentric maps.

physics.flu-dyn↗

arXiv2020

Adjoint-Based Sensitivity Analysis of Steady Char Burnout

Simulations of pulverised coal combustion rely on various models, required in order to correctly approximate the flow, chemical reactions, and behavior of solid particles. These models, in turn, rely on multiple model parameters, which are determined through experiments or small-scale simulations and contain a certain level of uncertainty. The competing effects of transport, particle physics, and chemistry give rise to various scales and disparate dynamics, making it a very challenging problem to analyse. Therefore, the steady combustion process of a single solid particle is considered as a starting point for this study. As an added complication, the large number of parameters present in such simulations makes a purely forward approach to sensitivity analysis very expensive and almost infeasible. Therefore, the use of adjoint-based algorithms, to identify and quantify the underlying sensitivities and uncertainties, is proposed. This adjoint framework bears a great advantage in this case, where a large input space is analysed, since a single forward and backward sweep provides sensitivity information with respect to all parameters of interest. In order to investigate the applicability of such methods, both discrete and continuous adjoints are considered, and compared to the conventional approaches, such as finite differences, and forward sensitivity analysis. Various quantities of interest are considered, and sensitivities with respect to the relevant combustion parameters are reported for two different freestream compositions, describing air and oxy-atmospheres. This study serves as a benchmark for future research, where unsteady and finally turbulent cases will be considered.

physics.chem-ph↗

arXiv2020

Turbulent flame speed and reaction layer thickening in premixed jet flames at constant Karlovitz and increasing Reynolds numbers

A series of Direct Numerical Simulations (DNS) of lean methane/air flames was conducted in order to investigate the enhancement of the turbulent flame speed and modifications to the reaction layer structure associated with the systematic increase of the integral scale of turbulence $l$ while the Karlovitz number and the Kolmogorov scale are kept constant. Four turbulent slot jet flames are simulated at increasing Reynolds number and up to $Re \approx 22000$, defined based on the bulk velocity, slot width, and the reactants' properties. The turbulent flame speed $S_T$ is evaluated locally at select streamwise locations and it is observed to increase both in the streamwise direction for each flame and across flames for increasing Reynolds number, in line with a corresponding increase of the turbulent integral scale. In particular, the turbulent flame speed $S_T$ increases exponentially with the integral scale for $l$ up to about 6 laminar flame thicknesses, while the scaling becomes a power-law for larger values of $l$. These trends cannot be ascribed completely to the increase in the flame surface, since the turbulent flame speed looses its proportionality to the flame area as the integral scale increases; in particular, it is found that the ratio of turbulent flame speed to area attains a power-law scaling $l^{0.2}$. This is caused by an overall broadening of the reaction layer for increasing integral scale, which is not associated with a corresponding decrease of the reaction rate, causing a net enhancement of the overall burning rate. This observation is significant since it suggests that a continuous increase in the size of the largest scales of turbulence might be responsible for progressively stronger modifications of the flame's inner layers even if the smallest scales, i.e., the Karlovitz number, are kept constant.

physics.flu-dyn↗