kth.sePublications KTH
Change search
Link to record
Permanent link

Direct link
D'Afiero, Francesco MarioORCID iD iconorcid.org/0009-0008-8155-0392
Publications (8 of 8) Show all publications
D'Afiero, F. M. (2026). A modified entropy functional for improved high order shock capturing methods. Computers & Fluids, 320, Article ID 107293.
Open this publication in new window or tab >>A modified entropy functional for improved high order shock capturing methods
2026 (English)In: Computers & Fluids, ISSN 0045-7930, E-ISSN 1879-0747, Vol. 320, article id 107293Article in journal (Refereed) Published
Abstract [en]

This study proposes a new piecewise entropy functional aimed at improving entropy based troubled element detection in high order shock capturing methods under demanding near vacuum conditions. In such regimes, the classical thermodynamic entropy log⁡(𝑝/𝜌𝛾) becomes singular and highly sensitive to small perturbations of the isentropic pressure, which can compromise the reliability of entropy thresholding strategies. The proposed functional is designed to reduce this stiffness while retaining the classical logarithmic entropy behavior away from the low isentropic pressure regime. Building upon the generalized entropy framework introduced by Harten, the proposed functional is analyzed in terms of compatibility and convexity. The analysis shows that it behaves as a valid convex entropy flux pair above a computable threshold in the isentropic pressure. Below this threshold, where strict convexity is lost, the functional is not interpreted as the basis for a global entropy stability proof, but rather as a regularized entropy indicator for troubled element detection. The functional is constructed to be piecewise continuously differentiable and is specifically tailored for use within modal filtering based shock capturing strategies for Discontinuous Galerkin methods. The only modification to the baseline filtering framework is the replacement of the original thermodynamic entropy used in the detection criterion with the proposed piecewise functional. Numerical experiments demonstrate that the modified entropy improves the reliability of troubled element detection in challenging near vacuum and high Mach number configurations, while preserving the robustness of the underlying shock capturing procedure.

Place, publisher, year, edition, pages
Elsevier BV, 2026
Keywords
Compressible flows, Discontinuous Galerkin, Entropy, Shock capturing, Shock waves, Transonic flows
National Category
Fluid Mechanics
Identifiers
urn:nbn:se:kth:diva-389829 (URN)10.1016/j.compfluid.2026.107293 (DOI)001878581000001 ()2-s2.0-105050396319 (Scopus ID)
Note

QC 20261001

Available from: 2026-10-01 Created: 2026-10-01 Last updated: 2026-10-01Bibliographically approved
D'Afiero, F. M. (2026). Embedded strong stability preserving Runge-Kutta methods with adaptive time stepping for shock-dominated flows. Computers & Fluids, 305, Article ID 106916.
Open this publication in new window or tab >>Embedded strong stability preserving Runge-Kutta methods with adaptive time stepping for shock-dominated flows
2026 (English)In: Computers & Fluids, ISSN 0045-7930, E-ISSN 1879-0747, Vol. 305, article id 106916Article in journal (Refereed) Published
Abstract [en]

Accurate time integration of hyperbolic-parabolic systems, particularly in the presence of shocks and steep gradients, remains a central challenge in computational fluid dynamics. In this work, we propose a robust, adaptive time integration framework for discontinuous Galerkin discretizations that combines an embedded third-order Strong Stability Preserving Runge-Kutta method with physics-based shock capturing and novel error control strategies. The proposed method is based on total variation diminishing properties while leveraging a proportional-integral controller for adaptive step-size selection, eliminating the need for empirical CFL tuning. A key innovation lies in the introduction of an entropy-based filtering mechanism that modulates element-wise error estimates, effectively dampening spurious spikes induced by discontinuities. Additionally, the integral term of the PI controller is stabilized using a moving median over a sliding window, enhancing reliability in shock-dominated regimes. The overall methodology requires no parameter tuning beyond a user-defined error tolerance (as it is common in any ordinary differential equation) and is demonstrated to be stable and accurate across a broad range of canonical test cases. Compared to conventional CFL stable solution obtained for the same numerical setups in a previous work, the proposed approach consistently delivers improved accuracy and robustness for high-fidelity simulations in complex compressible flows.

Place, publisher, year, edition, pages
Elsevier BV, 2026
Keywords
Discontinuous Galerkin, Flux reconstruction, ODE, Shock waves
National Category
Computational Mathematics Computer Sciences
Identifiers
urn:nbn:se:kth:diva-373670 (URN)10.1016/j.compfluid.2025.106916 (DOI)001629707900001 ()2-s2.0-105022612318 (Scopus ID)
Note

QC 20251211

Available from: 2025-12-11 Created: 2025-12-11 Last updated: 2025-12-11Bibliographically approved
D'Afiero, F. M., Mihaescu, M. & Hanifi, A. (2026). Riemann solvers and viscous flux discretizations for turbomachinery flows in Local Discontinuous Galerkin Flux Reconstruction schemes. Computers & Fluids, 315, Article ID 107144.
Open this publication in new window or tab >>Riemann solvers and viscous flux discretizations for turbomachinery flows in Local Discontinuous Galerkin Flux Reconstruction schemes
2026 (English)In: Computers & Fluids, ISSN 0045-7930, E-ISSN 1879-0747, Vol. 315, article id 107144Article in journal (Refereed) Published
Abstract [en]

This work examines how inviscid interface-flux choices and Local Discontinuous Galerkin (LDG) viscous-flux parameters affect solution sensitivity and computational cost in high-order Discontinuous Galerkin Flux Reconstruction (DG-FR) simulations of compressible turbomachinery flows. A transonic low-pressure turbine cascade at M2is=1.22 and Re2is=1.5×105 is considered using a range of inviscid Riemann solvers, including Rusanov, HLLC-type solvers, RoeM and an Exact solver. The LDG discretization is varied through the penalty parameter τ and the directional parameter β. In addition to standard formulations, where the same inviscid flux is used on all interfaces, hybrid HF-R-X formulations are investigated, with Rusanov applied at wall boundaries and solver X applied on interior interfaces. The comparison is based on integrated aerodynamic forces, surface distributions of isentropic Mach number and skin friction, loss coefficients evaluated from inlet and outlet sampling planes, base pressure and normalized wall-clock time. The results show that, for the operating condition and resolution considered, lift and drag are only weakly sensitive to the inviscid Riemann solver, with variations remaining below approximately half a percent across the converged cases. Surface pressure and skin-friction trends are also nearly unchanged among the approximate solvers. Loss-related quantities are more sensitive to the numerical formulation and provide a clearer distinction between inviscid-flux and LDG-parameter choices. From a performance perspective, approximate inviscid solvers introduce only marginal overhead relative to the Rusanov baseline, whereas the Exact solver increases the wall-clock time more noticeably. The hybrid HF-R-E formulation reduces part of this cost penalty by avoiding Exact flux evaluations at wall boundaries. The dominant runtime sensitivity is associated with the LDG directional parameter: centered viscous paths with β=0 produce a substantial cost increase compared with biased paths with β=±0.5. Overall, the results indicate that inexpensive approximate inviscid fluxes are sufficient for integrated load prediction in this case, while loss prediction and computational efficiency are more strongly affected by the LDG viscous-flux construction and by the use of hybrid flux strategies.

Place, publisher, year, edition, pages
Elsevier BV, 2026
Keywords
Discontinuous Galerkin, Filtering, Flux reconstruction, Hyperbolic systems, Shock capturing, Spectral element methods
National Category
Computational Mathematics
Identifiers
urn:nbn:se:kth:diva-383044 (URN)10.1016/j.compfluid.2026.107144 (DOI)2-s2.0-105039830761 (Scopus ID)
Note

QC 20260605

Available from: 2026-06-05 Created: 2026-06-05 Last updated: 2026-06-05Bibliographically approved
D'Afiero, F. M. (2025). Explainable sequential unsupervised data driven clustering for feature identification in external flows. Physics of fluids, 37(2), Article ID 027176.
Open this publication in new window or tab >>Explainable sequential unsupervised data driven clustering for feature identification in external flows
2025 (English)In: Physics of fluids, ISSN 1070-6631, E-ISSN 1089-7666, Vol. 37, no 2, article id 027176Article in journal (Refereed) Published
Abstract [en]

This paper introduces a novel unsupervised clustering algorithm tailored for feature identification in external fluid flows within computational fluid dynamics. By employing sequential clustering techniques, the proposed method extracts key flow features, such as shock waves, boundary layers and wakes, vortices, and laminar separation bubbles from compressible and incompressible flows. Unlike traditional approaches, this algorithm relies on primitive flow variables and their inherent physical properties, eliminating preprocessing requirements and enhancing generalization capabilities. Utilizing Gaussian mixture models and incorporating rescaling and binarization operations, the method partitions the flow domain into distinct regions of interest through transformations of primitive variables and gradients of thereof. The parameter-free design ensures ease of implementation and robustness across diverse flow scenarios, including two- and three-dimensional configurations. The algorithm's versatility and accuracy are demonstrated through its application to various cases, showcasing its potential as a powerful and accessible tool to gain deeper insight into complex fluid phenomena.

Place, publisher, year, edition, pages
AIP Publishing, 2025
National Category
Computer Sciences Fluid Mechanics Computational Mathematics
Identifiers
urn:nbn:se:kth:diva-360905 (URN)10.1063/5.0257436 (DOI)001432270300005 ()2-s2.0-85218357117 (Scopus ID)
Note

QC 20250317

Available from: 2025-03-05 Created: 2025-03-05 Last updated: 2025-03-17Bibliographically approved
D'Afiero, F. M. (2025). Feature based analysis of low frequency oscillations in transonic low Reynolds airfoil flows. Physics of fluids, 37(5), Article ID 056112.
Open this publication in new window or tab >>Feature based analysis of low frequency oscillations in transonic low Reynolds airfoil flows
2025 (English)In: Physics of fluids, ISSN 1070-6631, E-ISSN 1089-7666, Vol. 37, no 5, article id 056112Article in journal (Refereed) Published
Abstract [en]

This study investigates the low-frequency oscillations observed in transonic low Reynolds number airfoil flows using a feature-based spectral proper orthogonal decomposition approach. A novel clustering algorithm is employed to selectively identify the shear layer dynamics and analyze its role in driving buffet-type oscillations. Through a combination of space-time analysis and spectral decomposition, it is demonstrated that the periodic thickening and thinning of the shear layer is driven by an intrinsic instability mechanism, rather than solely by a shock-induced feedback loop. The results reveal that the shear layer undergoes cyclic expansion and contraction, influenced by Kelvin-Helmholtz instabilities and interactions with upstream and downstream traveling waves. The spectral proper orthogonal decomposition analysis further highlights the presence of coherent low-frequency structures associated with these oscillations, confirming their fundamental role in the buffet cycle. This study provides novel insights into the governing mechanisms of transonic buffet at low Reynolds numbers, highlighting the dominant role of shear layer instabilities over conventional shock-wave-based interpretations. These findings offer a more refined understanding of buffet onset and prove the effectiveness of feature-based analysis in investigating complex nonlinear flow phenomena.

Place, publisher, year, edition, pages
AIP Publishing, 2025
National Category
Fluid Mechanics
Identifiers
urn:nbn:se:kth:diva-366155 (URN)10.1063/5.0266901 (DOI)001493187300031 ()2-s2.0-105006879604 (Scopus ID)
Note

QC 20250728

Available from: 2025-07-28 Created: 2025-07-28 Last updated: 2025-07-28Bibliographically approved
D'Afiero, F. M. (2025). Static deep stall analysis augmented by numerically constrained turbulent flows. Physics of fluids, 37(1), Article ID 015175.
Open this publication in new window or tab >>Static deep stall analysis augmented by numerically constrained turbulent flows
2025 (English)In: Physics of fluids, ISSN 1070-6631, E-ISSN 1089-7666, Vol. 37, no 1, article id 015175Article in journal (Refereed) Published
Abstract [en]

As computational resources have advanced, scale-resolving methods like large eddy simulation (LES) have gained popularity in the aerodynamic design particularly in predicting massively separated flow regions. The limitations of traditional Reynolds-averaged Navier-Stokes (RANS) are evident in this regime, given the importance of the intrinsic unsteadiness and the strong three-dimensionality of the flow. This research focuses on wall-resolved implicit large eddy simulation (ILES) for airfoil flows in deep stall with a focus on its dynamics. Moreover, it is also of interest to understand how the spanwise extent of the computational domain affects the resolved aerodynamic forces and flow dynamics. The simulations in this study analyzed various spanwise extents (lz ranging from 0.10 c to 2c) under specific conditions: a chord Reynolds number R e = 10 5 , an asymptotic Mach number M ∞ = 0.15 , and an angle of attack α = 20 degrees. Significant differences were found in aerodynamic coefficients and flow dynamics when the smallest spanwise extent was used. Specifically, spectral proper orthogonal decomposition showed that a small span failed to capture critical low-frequency dynamics necessary for predicting adequate stall behavior. Additionally, the turbulence field in the small span case remained largely rod-like, unlike the three-dimensional turbulence seen in larger spans. The current work propose the nondimensional spanwise integral length scale ( SILS / l z ) as a metric for determining the sufficiency of spanwise extent, using two-point correlations for the streamwise velocity. Findings suggest that a spanwise extent of l z = 1 c is sufficient, with minimal differences observed between l z = 1 c and l z = 2 c . The results emphasize the importance of adequate spanwise resolution to accurately capture deep-stall flow structures and their unsteady behavior.

Place, publisher, year, edition, pages
AIP Publishing, 2025
National Category
Fluid Mechanics
Identifiers
urn:nbn:se:kth:diva-358895 (URN)10.1063/5.0248856 (DOI)001394069900026 ()2-s2.0-85214658821 (Scopus ID)
Note

QC 20250127

Available from: 2025-01-23 Created: 2025-01-23 Last updated: 2025-01-27Bibliographically approved
D'Afiero, F. M., Mihaescu, M. & Hanifi, A. (2025). Transonic Flow Through a Low-Pressure Turbine Cascade Using a Local Discontinuous Galerkin Flux Reconstruction. In: Proceedings of ASME Turbo Expo 2025: Turbomachinery Technical Conference and Exposition, GT 2025: . Paper presented at 70th ASME Turbo Expo 2025: Turbomachinery Technical Conference and Exposition, GT 2025, Memphis, United States of America, June 16-20, 2025. ASME International, Article ID V011T32A001.
Open this publication in new window or tab >>Transonic Flow Through a Low-Pressure Turbine Cascade Using a Local Discontinuous Galerkin Flux Reconstruction
2025 (English)In: Proceedings of ASME Turbo Expo 2025: Turbomachinery Technical Conference and Exposition, GT 2025, ASME International , 2025, article id V011T32A001Conference paper, Published paper (Refereed)
Abstract [en]

In this study, we focus on the simulation of transonic flows in the context of turbomachinery applications, employing a high-order Discontinuous Galerkin Flux Reconstruction (DG-FR) methodology. Conducting high-order simulations within the context of transonic turbines is challenging due to the presence of extreme unsteadiness in the flow accompanied by fluid compressibility effects, such as transonic vortex shedding and unsteady shock wave-boundary layer interactions. These phenomena significantly influence the aerodynamic performance of the airfoils playing a major role also in the determination of the losses in the flow. The primary objective of this work is to evaluate the impact of the formulated non-reflective boundary conditions and of a recently proposed shock capturing scheme on crucial aerodynamic predictions, including the isentropic Mach number and the skin friction distribution. The non-reflective boundary condition is specifically designed to leverage modern Graphical Processing Unit (GPU) architectures, aiming to minimize the necessity for blocking communication. The outcomes of this new outlet boundary condition will be thoroughly analyzed in terms of both computational cost and its influence on the resolved flowfield. The developed physics-based shock-capturing scheme is particularly attractive due to its reduced computational cost and ease of implementation and usage. The focus will be on assessing its performance in the context of highly unsteady flows, such as the current low-pressure turbine cascade. The effectiveness of the proposed computational setup will be rigorously examined to ensure its suitability for simulating complex transonic flows in turbomachinery applications.

Place, publisher, year, edition, pages
ASME International, 2025
Keywords
Discontinuous Galerkin, Flux Reconstruction, GPU, Shock Capturing, Turbine Cascade
National Category
Fluid Mechanics Energy Engineering Vehicle and Aerospace Engineering
Identifiers
urn:nbn:se:kth:diva-370453 (URN)10.1115/GT2025-151123 (DOI)001561952100014 ()2-s2.0-105014747609 (Scopus ID)
Conference
70th ASME Turbo Expo 2025: Turbomachinery Technical Conference and Exposition, GT 2025, Memphis, United States of America, June 16-20, 2025
Note

Part of ISBN 9780791888872

QC 20250930

Available from: 2025-09-30 Created: 2025-09-30 Last updated: 2026-01-09Bibliographically approved
D'Afiero, F. M., Mihaescu, M. & Hanifi, A. (2024). Very-low-dissipation, physics-based, and parameter-free shock capturing for discontinuous Galerkin methods. Physics of fluids, 36(12), Article ID 126115.
Open this publication in new window or tab >>Very-low-dissipation, physics-based, and parameter-free shock capturing for discontinuous Galerkin methods
2024 (English)In: Physics of fluids, ISSN 1070-6631, E-ISSN 1089-7666, Vol. 36, no 12, article id 126115Article in journal (Refereed) Published
Abstract [en]

This work addresses the challenge of shock capturing in numerical simulations of hyperbolic conservation laws, focusing on the discontinuous Galerkin (DG) method. It proposes a novel modal filtering approach based on physical criteria to detect elements near discontinuities and tune modal damping in the solutions. The minimum entropy principle combined with prescribed physical criteria is used to detect discontinuities and the Ducros sensor, a contact discontinuity sensor, and the Persson modal sensor are used to appropriately adjust filter intensity. The filter function is designed to apply higher dissipation to high-frequency modes, which often contribute to spurious oscillations. Additionally, temporal blending of the filter intensity is introduced to handle non-stationary discontinuities effectively. The effectiveness of the proposed approach is demonstrated through various test cases involving Euler and Navier-Stokes equations, ranging from smooth compressible flows to complex shock-boundary layer interactions. Results show that the method successfully eliminates spurious oscillations while maintaining high-order accuracy in smooth regions. It proves to be computationally efficient and less dissipative compared to other shock-capturing techniques. The proposed modal filtering approach offers an effective, efficient, and robust solution for shock capturing in the DG framework, capable of handling a wide range of flow scenarios while preserving solution accuracy.

Place, publisher, year, edition, pages
AIP Publishing, 2024
National Category
Computational Mathematics Applied Mechanics
Identifiers
urn:nbn:se:kth:diva-357935 (URN)10.1063/5.0241108 (DOI)001378199100045 ()2-s2.0-85211374821 (Scopus ID)
Note

QC 20241220

Available from: 2024-12-19 Created: 2024-12-19 Last updated: 2025-01-20Bibliographically approved
Organisations
Identifiers
ORCID iD: ORCID iD iconorcid.org/0009-0008-8155-0392

Search in DiVA

Show all publications