Extreme CFD workshop - User contributions [en]

Ecfd:ecfd 9th edition

2026-02-06T07:26:46Z

Luis.carbajal-carrasco: /* TP6 - Comparison of JICF models */

{{DISPLAYTITLE: ECFD workshop, 9th edition, 2026}}

== Description ==

* Event from '''19th of January to 30th of January 2026'''
* Location: [https://www.sport-normandie.fr/le-centre/le-site-de-houlgate Centre Sportif de Normandie], Houlgate, near Caen (14)
* Two types of sessions:
** common technical presentations: roadmaps, specific points
** mini-workshops. Potential workshops are listed below
* Free of charge
* Participants from academics, HPC center/experts and industry are welcome
* The number of participants is limited to 80.


* Organizers
** Guillaume Balarac (LEGI), Simon Mendez (IMAG), Pierre Bénard, Vincent Moureau, Léa Voivenel (CORIA).

[[File:Logo_ECFD9.png|center|frameless|900px|link=https://ecfd.coria-cfd.fr/index.php/Ecfd:ecfd_9th_edition]]



== News ==

* 22/09/2025: First announcement of the '''9th Extreme CFD Workshop & Hackathon''' !
* 15/11/2025: Deadline to submit your project

== Thematics / Mini-workshops ==

To be announced...

== Projects ==

=== Hackathon GENCI - P. Begou (LEGI), V. Moureau (CORIA) ===
This ECFD9 GENCI Hackathon was a rich event, involving 3 differents CFD codes (AVBP, SONICS and YALES2) using various paradigms (C++/cuda/hip, Fortran/OpenMP/OpenACC) with several SDKs (AMD, Cray/HPE, Nvidia, Gnu) on a large range of GPU architectures (Nvidia A100, GH100, AMD instinct Mi210, Mi250, Mi300). This two-week event benefited from a high level support from three HPC mentors, two on-site from AMD (J. Noudohouenou and A. Tsetoglou).

==== H3 - Hackathon SONICS - A. de Brauer (ONERA), B. Michel (ONERA), B. Berthoul (ONERA) & G. Staffelbach (ONERA) ====
CPU code generation for multispecies simulations – Code generation for multispecies simulations is currently being developed in the SoNICS code. The work carried out at ECFD9 focused on the vectorization of the generated code by code transformation : unrolling the species loops, rewriting if statements, and inverting do/while loops (arising from Newton type algorithms) used in the computation of thermodynamic quantities. The loop-unrolling and if statement rewriting have been profiled and show a speed-up of 2x for the vectorized generated code when computing the HLLC flux, compared with the hand-written implementation. The switch of do/while loops was prototyped on a test code and will be integrated into SoNICS. Code generation on GPU has been tested and validated, but a thorough performance profiling of the GPU version is still required.

Porting reactive multi-species terms to GPU – In 2025 multi-species reactive capabilities were introduced in SoNICS and tested on the Preccinsta case on CPU. Recently the multi-species components were ported to the GPU, so this activity concentrated on porting the reactive source terms. Tests on a 0D reactor show identical results on GPU and CPU. Work has also resumed on porting SoNICS to AMD GPU on the ADASTRA system from CINES/GENCI, where the hipGraphs implementation (AMD’s counterpart of cudaGraph) exhibited some issues. Our participation in ECFD9 allowed us to contact the AMD hipGraphs development team, opening the way to a collaboration. With their council, we updated the code to use rocm7.1.1 providing the first successful non reactive results on AMD GPU. Further work on reactive flows is ongoing.

Improving GPU residual calculations – Recent investigations show that SoNICS’s residual calculation on GPUs was about 100× slower than other GPU operations. The bottleneck was traced to the combination of cudaGraph and thrust::reduce, which prevented parallel execution. Replacing this with a hand-written hierarchical reduction kernel that works efficiently within the cudaGraph restores good scalability; the residual computation is now negligible compared with the other operations, as is the case on CPU. Additional timers were added to the cudaGraph kernels to quantify each operator’s cost relative to the CPU.

=== Numerics & User Interface - M. Bernard (LEGI), G. Lartigue (CORIA) & S. Mendez (IMAG) ===

==== N1 - Improving ICS robustness and accuracy - M. Bernard (LEGI), G. Lartigue (CORIA), G. Balarac (LEGI) & T. Berthelon (LEGI) ====
Bad quality meshes generally lead to larger numerical errors when solving partial differential equations.
This project focused on improving the accuracy and robustness of the incompressible Navier-Stokes solver (ICS).
We investigated the sources of discrepancy introduced at each step of the algorithm, with particular attention to the consequences of the coexistence of two discrete velocity representations: (i) the convective flux <math>\vec{u}\cdot\vec{n}\,dS</math> and (ii) the transported nodal velocity field <math>u^n</math>.
Although these quantities are equivalent at the continuous level, this equivalence no longer holds in the discrete setting.
In particular, only the convective velocity strictly satisfies the divergence-free constraint after solving the Poisson problem for the pressure field.
During this two-week workshop, we developed a new correction strategy for the nodal velocity field in order to enforce consistency with the convective velocity and improve the overall behavior of the solver.

==== N3 - Shock & discontinuity treatment for Lattice-Boltzmann solvers - I. Tsetoglou (M2P2), W. Bessem (M2P2), H. Merley (M2P2) & S. Zhao (M2P2) ====
Lattice—Boltzmann methods (LBM) have traditionally been applied to weakly compressible flows; however, recent developments have extended their applicability to fully compressible regimes. In such flow configurations, shock waves and contact discontinuities naturally arise. To properly capture these features in a discretized framework, artificial diffusion mechanisms are commonly introduced to smooth discontinuities over a limited number of grid points.
In this project, the hybrid LBM solver ProLB was employed. In this framework, the mass and momentum equations are solved using an LBM formulation, while the total energy equation is discretized using a finite-volume (FV) approach with consistent spatial and temporal discretization. The primary objective of the work was to develop and implement an artificial diffusion strategy suitable for hybrid LBM/FV solvers.
Shock waves were detected with a Jameson-type pressure-based sensor, while contact discontinuities were identified with a temperature-based sensor. The pressure-based sensor was scaled to obtain a kinematic viscosity contribution, which was incorporated into the LBM collision relaxation time. Similarly, the temperature-based sensor was scaled to define an artificial thermal conductivity, which was added to the FV discretization of the total energy equation.
A set of validation cases—including the Sod shock tube at various pressure ratios, a 2D Riemann problem, and the interaction of a shock wave with a helium bubble in air—was performed. The results demonstrate that the hybrid LBM approach is capable of accurately capturing shocks and contact discontinuities, even on relatively coarse meshes, while avoiding spurious Gibbs oscillations.

==== N5 - Dorothy: Toward Fully Distributed Implementation - A. Vergnaud (LOMC), M. Roperch (LOMC) & G. Pinon (LOMC) ====

Dorothy is a Vortex Particle Method CFD code for turbine wakes. Its parallel performance needs to be improved when large number of particles is used (e.g. multi-turbines farm cases or far-wake studies). Several limitations are observed due to lacks in terms of memory, structure of data, parallel implementation, etc… To overcome these problems, the possibility of another code structure/architecture (fully parallel and scalable), even for large number of particles, needs to be investigated. The aim of this project is to explore the use of the library AMReX (https://amrex-codes.github.io/amrex/overview.html) which provides a large toolbox to manage massively parallel block-structured AMR applications (mesh data structure, particle data structure, load balancing, processors communications, etc...).

Some tests have been performed to study AMReX performances. In particular, a scalability test has been performed over a tutorial particle method case (Particles In Cells tutorial code), upgraded up to 134 millions of particles (which, for now, is much higher than the number of particles used with Dorothy). A good scalability has been measured, better than with Dorothy: over 75% on 800 cores (on CRIANN). These results are encouraging and suggest good performance when the AMReX library will be used to implement the Vortex Particle Method.

==== N6 - Relaxation of the IBM stability constraint - PL. Martin (IMAG) & S. Mendez (IMAG) ====
Many simulations done in the YALES2BIO framework involve fluid-structure interactions handled with the Immersed Boundary Method (IBM).
This model allows for the fluid/solid coupling, with the forces from the solid acting as a source term in the Navier-Stokes equations.
In some cases for red blood cells simulations, and for most cases for von Willebrand Factor simulations, the governing time step is the force time step. When this is the case, we also notice artifacts in the fluid velocity and pressure fields.
The robustness of our IBM implementation was improved for embedded surfaces by shifting our regularization/interpolation kernels away from the wall in case we work with an embedded solid.
Since these simulations are done at low Reynolds and CFL number (0.01 - 0.001), the stability constraint was relaxed by doing substeps without:
1. advancing the convective velocity, 2. correcting the velocity to make it divergence-free.
The artifacts showing when solids are a lot stiffer than the fluid viscous forces were reduced by projecting the regularized solid forces into a divergence-free space.

==== U1 - Yales2 Trame Editor, toward a fully featured graphical user interface for YALES2 - L. Korzeczek (GDTech), T-P. Luu (GDTech), S. Meynet (GDTech), M. Cailler (Safran), R. Letournel (Safran), G. Lartigue (CORIA)====

Yales2 features an initial version of a graphical interface. This version enables users to execute a series of processes on a local machine, covering data preparation, computation, and post-processing for basic aerodynamic and hydrodynamic calculations.

To facilitate industrialization and support advanced users in applying it to complex projects, it is essential to extend this interface to a broader range of physical applications. This includes enabling the implementation of coupled or chained calculations and allowing communication with remote servers.

The work conducted during this ECFD have significantly strengthened the current architecture, enhancing performance, modularity, and the capacity to accommodate complex scenarios. Additionally, new widgets have been developed, and an initial draft for connecting to a remote server has been initiated.

=== Turbulence - L. Voivenel (CORIA), P. Bénard, (CORIA) & T. Berthelon (LEGI) ===

==== T1 - Concurrent Precursor-Successor with Successor automated mesh convergence - P. Launay (CORIA), L. Voivenel (CORIA) & P. Benard (CORIA) ====

Using a periodic precursor simulation remains the more accurate method for generating realistic fully developed atmospheric turbulence for a successor simulation. However, it is also the most expensive one. Only the sequential method was implemented in YALES2, involving 2 separate simulation running one after the other, and relying on a lookup table as a link between the two. This project proposed to reduce the cost of the method by implementing a concurrent version where both simulations run in the mean time.

This was achieved using existing CWIPI developments. Another issue arising in such periodic precursors is the creation of spanwise inhomogeneities namely "streaks". This issue has been addressed using CWIPI by replacing the streamwise periodic boundary conditions by an internal coupling between an internal plane of the precursor and its inlet where it is being recycled. A spanwise shift of the velocity field is applied at the inlet preventing the generation of "streaks". A flow rate correction is also applied for preventing bulk velocity drift as the recycling procedure induces a 1 iteration delay. Note that this method is more efficient and more accurate than the Recycling method already existing in YALES2 and relying on particles. Finally, the method has been furthermore improved using Traction free outlet boundary conditions in both precursor and successor domains allowing the reduction of domain length.

Overall the cost of the whole workflow has been greatly reduced and the formation of streaks has been prevented.
The nature of the turbulent structures before and after this modification needs further investigation, as well as the use of other streamwise boundary conditions (INLET/INLET, ...), and are the subject of current work.

==== T2 - Discharge movement model for breakdown prediction - S. Wang (EM2C), T. Kabir (EM2C), E. Roger (EM2C), C. Laux (EM2C), B. Fiorina (EM2C), Y. Bechane (CORIA) & V. Moureau (CORIA) ====

A well-established approach for performing 3-D simulations of plasma-assisted combustion at reduced computational cost is the use of phenomenological models for Nanosecond Repetitively Pulsed (NRP) plasma discharges. So far, these models have assumed a static cylindrical shape for the discharge energy deposition region. However, the breakdown location is governed by the flow velocity, electron density, and reduced electric field, which are neither static nor uniform. As a result, the discharge may exhibit elongation, translation, or rotation. This project aimed to implement a simplified physics-based discharge movement model using the reduced electric field, electron mobility, and an electron density-like variable.
Most of the model was successfully implemented, except for the final step, in which the field line corresponding to the maximum restrike probability must be constructed to determine the new plasma restrike zone.

==== T3 - Vorticity model for discharge-induced flow dynamics - S. Wang (EM2C), T. Kabir (EM2C), E. Roger (EM2C), C. Laux (EM2C), B. Fiorina (EM2C), Y. Bechane (CORIA) & V. Moureau (CORIA) ====

A well-established approach for performing 3-D simulations of plasma-assisted combustion at reduced computational cost is the use of phenomenological models for Nanosecond Repetitively Pulsed (NRP) plasma discharges. These models can be implemented within a low-Mach number framework to further reduce the cost. However, doing so removes the acoustic necessary to resolve discharge-induced flow dynamics. Recently, Roger et al. (2025) proposed a model using physics-based vorticity patches to recover these flow dynamics. This project aimed to implement this model in the low-Mach number framework of YALES2 (YALES2-VDS). The model, formulated as an external forcing term in the momentum balance equation, was successfully implemented. However, simulations performed with YALES2-VDS without the vorticity model exhibit the formation of vortices in regions where none are expected. A possible source of error may be related to the treatment of the hydrodynamic pressure gradient and the associated baroclinic torque term in the vorticity equation. The behavior of this term requires further investigation before the viability of the vorticity model within a low-Mach number framework can be properly assessed.

==== T4 - Wind field reconstruction based on LiDAR measurements - T. Cousin (LMI), P. Benard (CORIA), G. Lartigue (CORIA) & JB. Lagaert (LMO) ====

Wind turbines experience significant loads due to the wind pressure exerted on their structure. Accurate prediction of wind turbine behavior is essential for effective management. Simulations use wind data as input, and their realism can be improved by incorporating wind profiles derived from on-site LiDAR measurements.
The scope of this project is to provide a suitable mathematical framework phrased as a minimization problem under incompressibility constraint to reconstruct the wind field from the LiDAR dataset. The entire framework has been developed using the YALES2 scalar solver, with the objective of extending it to the NS solver under the low-Mach number and constant-density approximation.

==== T5 – Hybrid RANS/LES based on dual mesh and LES of fluctuations - G. Balarac (LEGI), T. Berthelon (LEGI) & R. Letournel (Safran) ====

This project is devoted to a fully coupled hybrid RANS/LES strategy based on a dual-mesh framework, where the mean flow is solved by RANS on a mesh tailored for the mean field, while only the turbulent fluctuations are resolved by LES on a second mesh. In addition to deterministic drift (relaxation) terms that drive the resolved velocities in each model toward target fields provided by the other one (RANS mean for LES, LES statistics for RANS), a stochastic forcing built from RANS turbulent quantities is introduced in the LES of fluctuations. These combined forcing terms allow a controlled generation of fluctuations at the RANS/LES interface and reduce the sensitivity to interface location. Two-way coupling is achieved by feeding back the Reynolds stresses computed in the LES into the RANS equations in the resolved regions. The approach is demonstrated on turbulent pipe flows, including a fully coupled simulation at high Reynolds number (Re = 44,000), showing that the method enables wall-resolved hybrid simulations at a fraction of the cost of a full LES.

==== T6 - Injection of coherent structures for LES inlet condition - T. Berthelon (LEGI), G. Balarac (LEGI), R. Letournel (Safran), P. Launay (CORIA), L. Voivenel (CORIA) & P. Benard (CORIA) ====

The boundary conditions of an LES calculation play a key role in the predictability of simulations. In particular, the turbulence injected at the inlet can strongly influence the development of turbulence.
The aim of this project was to extend the turbulence injection capabilities of the YALES2 code. On the one hand, the historical strategy of injecting synthetic homogeneous isotropic turbulence calculated from a Passot-Pouquet spectrum model has been enhanced by enabling the generation of richer spectra (Pope and Von-Karman-Pao spectra model).
On the other hand, the Synthetic Eddy Method (SEM), proposed by Jarrin et al (2008), was implemented. This method consists of generating a coherent velocity field that respects a target Reynolds tensor and a characteristic size of the large turbulent scale. To do this, the velocity field is generated by summing the contributions of several eddies whose position is the result of a random process.
First, these new strategies were compared in the case of turbulent flow within a pipe. The SEM and the injection of a richer spectrum show a real gain in terms of the flow establishment length in this case.
Finally, the new SEM method was tested on an urban flow case and in a zonal RANS/LES coupling context.

==== T7 - Integration of a bending blade method with Dorothy - E. Mascrier (LOMC), M. Roperch (LOMC), A. Vergnaud (LOMC) & G. Pinon (LOMC) ====

The size of offshore wind turbine blades has been steadily increasing over the years. Longer blades result in larger structural displacements during operation. Blade deformation has therefore become a key design parameter for large rotors. In this context, the present project focuses on coupling an in-house structural beam solver, based on Timoshenko beam theory, with an in-house Lagrangian vortex particle solver called Dorothy.
The project was initiated during ECFD8, where static blade deformation was implemented. This year, Dorothy has been fully dynamically coupled with the structural solver.
The first results show good agreement with the literature in terms of blade deflection and aerodynamic forces for the NREL 5MW rotor.
This work will be continued after ECFD9, with additional simulations performed to verify the results against other numerical approaches, such as YALES2.

==== T8 - FSI-3D without deformation strategy for internal flows - P. Benez (Safran), H. Lam (LEGI) & P. Benard (CORIA) ====

==== T9 - LES-based aero-servo-elastic simulation of wind turbines - E. Muller (CORIA & SGRE), P. Benard (CORIA), F. Houtin-Mongrolle (SGRE), B. Duboc (SGRE) & H. Hamdani (GDTech) ====

The YALES2 library includes an advanced modular implementation of the Actuator Line Method (ALM). This approach remains state-of-the-art when performing an LES-based analysis of a wind turbine wake. The method also provides an accurate assessment of the aerodynamic loads applied on the turbine as well as the structural deformation when YALES2 is coupled to an external library/code. In the past years two coupling libraries have been developed, one to BHawC (SGRE certification tool) and one to OpenFast (NREL open access/open source tool). To improve the user and developer experience, a generalization and uniformization of the two coupling has been conducted in this project. Extensive tests and validations were performed to guarantee the non-regression.

The ALM and ADM (via ALADIN model) frameworks in the YALES2 code were thus enhanced to benefit from these couplings. Such method allows to take part of the external structural solver and controller in single and multiple turbines configurations. Updates were also initiated directly in the coupling libraries to benefit from the latest developments made in the servo-structural solvers, thus allowing to simulate modern academic wind turbines (with OpenFAST) or industrial flagships (with BHawC) in operation.

Furthermore, works on the Risoe Dynamic stall model, initiated during ECFD6, have been achieved. The implementation and integration of this model has been continued, ported to the parallel-optimized ALM framework, and tested and validated on different configurations.

Miscellaneous tasks related to the ALM code pipeline coverage and documentation have been improved.

==== T10 – Numerical simulation of engine rotors - L. Bricteux (UMONS), G. Balarac (LEGI), Y. Bechane (CORIA) & P. Benard (CORIA) ====
This project investigated the capability of the explicit compressible solver in YALES2 to simulate the fan stage of a turbofan engine. The selected configuration is the CATANA rotor, developed at École Centrale de Lyon, for which experimental data are available.
The mesh of this complex geometry was generated using Gmsh and YALES2 and consists of approximately 220 million tetrahedral elements. The setup of the simulation with a moving mesh framework was carried out during the research stay.
During this work, wall boundary conditions were improved, and it was identified that the near-wall turbulence modeling strategy could be enhanced by introducing a compressible wall model based on the work of Debroeyer et al (JFM 2024). Initial simulations have been performed and have produced promising results.
The next step will be to integrate mesh adaptation and the new compressible wall model, and to compare numerical diagnostics with experimental measurements in order to validate both the modeling approach and further validate the solver.

=== Combustion - Y. Bechane (CORIA), R. Letournel (Safran) & S. Dillon (Safran) ===

==== C3 - LES of the thermal degradation of a composite material - A. Grenouilloux (ONERA), K. Bioche (CORIA), N. Dellinger (ONERA) and R. Letournel (Safran) ====

In order to certify new composite materials for aerospace applications, it is essential to understand their degradation dynamics under severe thermal loads. The ONERA FIRE test bed was designed for this purpose. This burner generates a premixed air propane flame that reproduces a thermal flux consistent with certification standards near the impinging region. During tests, a strong emission of pyrolysis gases and a secondary diffusion flame are observed, and these gases can self ignite in regions not directly exposed to the primary flame. The project aimed to improve the modeling of this burner using Large-Eddy Simulation and reduce the overall computational cost. A reduced kinetic mechanism was derived with the Brookesia library, enabling the modeling of both premixed and diffusion flames to take into account appropriate chemistry at the front face. Used in FIRE simulations, this mechanism achieved a CPU speed-up of a factor of two compared with the previous scheme. A second reduced mechanism was generated to target auto ignition of pyrolysis gas mixtures that can occur at the rear face, and a dedicated test case was designed. Recent developments in the CWIPI interface allow for mesh adaptation during coupling between YALES2 and MoDeTheC solvers.

==== C4 - Flamelet-Progress Variable approach in LBM solvers - U. Chikkabikkodu (M2P2), D. Nouembissi (M2P2), I. Mir (M2P2), H. Meunier (M2P2), I. Tsetoglou (M2P2), S. Zhao (M2P2), P. Boivin (M2P2), J. L. Consalvi (IUSTI), R. Mercier (Safran) & S. Dillon (Safran) ====

This project extended the capabilities of ProLB to support flamelet-based combustion modelling by implementing the flamelet progress variable (FPV) approach together with FTACLES capabilities in our LBM solvers - closing a long-standing gap, since table generation and usage had never been available in ProLB.
During the workshop, transport of a passive scalar was implemented and the SDR was modelled using the passive-scalar gradient, which currently form the two control variables used in the flamelet approach. The implementation was verified through simulations of a 2D laminar methane-air jet diffusion flame.
In parallel, for FTACLES we successfully generated both premixed and non-premixed tables with TECERACT (thanks to Renaud and Samuel), and converted them into a format compatible with our code structure. A progress-variable transport equation was also implemented where the diffusion, source and correction terms were read directly from the tabulation. Validation was performed on a 1D CH4/air premixed flame with 10 sampling points within the filter width, accurately recovering the flame speed and demonstrating successful coupling between the LBM solver and the tabulated chemistry.

==== C5 - NOx prediction with a hybrid FTACLES-Virtual chemistry approach - É. Espada (EM2C), M. Préteseille (EM2C), N. Darabiha (EM2C), B. Fiorina (EM2C) ====

Filtered Tabulated Chemistry is a powerfull yet very cost efficient tool to compute flame structure and its stabilisation. However, it is unable to predict NO concentration wihtout adding additional coordinates in the manifold or by using premixed-flamelet based additional model and tabulation like NOMANI. Virtual chemistry on the other hand is a chemistry reduction method that uses machine learning algorithm to reduce drastically the number of species and reaction. This reduced scheme is then transported like any detailed chemistry mechanism. Although the method is also able to recover flame strucure and pollutants, unlike FTACLES, transported chemistry lacks a turbulent combustion model to be applied on realistics industrial LES mesh grids. This present works aims to couple FTACLES and virtual chemistry in a one way coupling: FTACLES will compute flame structure (density, Temperature, velocity field) thanks to its turbulent combusiton model, and will then feed a virtual mechanism with the "main" grid information in order to compute the pollutant informations.

==== C6 - Modelling laminar & turbulent flames with virtual chemistry - M. Préteseille (EM2C), É. Espada (EM2C), N. Darabiha (EM2C), B. Fiorina (EM2C), S. Dillon (Safran), M. Cailler (Safran) ====

A virtual chemistry framework yielding global-type mechanisms has recently been developed and validated, allowing accurate prediction of flame structures at a substantially reduced computational cost. By coupling virtual chemistry with Adaptive Mesh Refinement (AMR) strategies, this work assesses the ability to dynamically resolve reactive zones while maintaining affordable computational costs in high-fidelity LES of industrial burners. A second objective of ECFD9 was to disseminate the virtual schemes generated using SuperVision, a Python-based automated optimization tool built on Cantera. An optimized hydrogen virtual mechanism was successfully implemented and validated in the Lattice–Boltzmann solver ProLB, demonstrating the ease with which these standardized schemes can be integrated into existing reactive flow solvers, and the spread potential of this new chemistry reduction strategy in the combustion community. Finally, the NOx virtual submechanism for hydrogen combustion was improved to accurately capture both thermal and prompt NO formation in hydrogen flames.

==== C8 - Optimization of chemical source terms stiff integration - G. Lartigue (CORIA), Y. Bechane (CORIA), K. Bioche (CORIA), Q. Cerutti (CORIA), M. El Moatamid (CORIA), M. Laignel (CORIA) ====

Integration of chemical source terms remains computationally expensive in configurations that rely on detailed chemistry approach. This project aimed to reduce that cost by (1) modifying the CVODE integration strategy and (2) applying source-term clustering. A first attempt was to modify CVODE’s internal step-size control strategy but it produced only minor gains as some unnecessary integration steps still occurred, mainly in the unburned gases region. This has finally been addressed by enforcing an initial step based on the CFD time step which reduced the computational cost by a factor 2 in these regions. More importantly, relaxing the relative and absolute tolerances used to determine the accuracy of the method reduced the computational cost by approximately 40% while introducing negligible error in physical properties and flame topology for a 1D premixed flame. These results were confirmed on three methane flame configurations: a 1D premixed flame, a 2D triple flame, and the PRECCINSTA burner. Numerical experiments on the PRECCINSTA burner show a reduction in integration cost by a factor of 2.5 using the adjusted CVODE strategy and by a factor of 4.4 when that strategy is combined with clustering.

==== C11 - Optimization of decoupled approach for heat transfers - T.-P. Luu (GDTech), R. Letournel (Safran), M. Tripiciano (Safran) ====

Many industrial aerothermal applications involve strong interactions between fluid flow and solid thermal response, requiring an accurate representation of fluid–solid coupling to predict wall heat fluxes. Although fully two-way coupled simulations provide high fidelity, their complex numerical setup and high computational cost limit their applicability in industrial design loops. As a result, one-way decoupled approaches based on the estimation of heat transfer coefficients (HTCs) are usually preferred. The classical double-run method, which relies on two simulations with imposed wall temperatures to estimate HTC, remains workflow-intensive and highly sensitive to the choice of reference temperatures. In this project, a single-run methodology is proposed to reduce setup complexity and computational cost. The approach introduces an additional passive scalar representing the variation of the fluid sensible enthalpy induced by a change in imposed wall temperature. The associated transport equation is derived under the assumption that the thermophysical properties of the mixture remain weakly dependent on temperature variations. The method is validated on a canonical three-dimensional heated plate configuration and demonstrates promising results when applied to an industrial burner simulation.

=== Mesh adaptation - A. Grenouilloux (ONERA) & G. Balarac (LEGI) ===

==== M2 – Dynamic of SWBLI in Supersonic Propulsive Nozzle Under Hot Gas Conditions - F.A. Rojas Segovia (CORIA), Y. Bechane (CORIA) & L. Voivenel (CORIA) ====
In this project, a STBLE (Solution Thin Boundary Layer Equations) wall model was implemented in YALES2. The focus was to add and compare this model with the pre-existing wall models in the code, such as the logarithmic law and Duprat, in the context of supersonic nozzles. To achieve this, 2D simulations of supersonic compressible flow over a flat plate were conducted as an initial step and validation. These initial simulations provided good insights for future research on the dynamics of Shock Wave and Boundary Layer Interaction (SWBLI) in supersonic nozzles operating with both cold and hot gas conditions.

==== M3 – Criterion for dynamic mesh adaptation in LES - H.Lam (LEGI), G. Balarac (LEGI), V. Moureau (CORIA), R. Barbera (LEGI), P. Launay (CORIA) & L. Voivenel (CORIA) ====

This project proposes a new criterion for dynamic mesh adaptation in LES, designed to overcome the limitations of static LES mesh convergence (static AMC) strategies based on time-averaged quantities. In both static and dynamic contexts, a cell-based Reynolds number is first used as a DNS criterion to identify regions where all turbulent scales must be resolved. For LES, the DNS constraint is relaxed when the integral scale is sufficiently larger than the local cut-off scale, so that a meaningful GS/SGS separation exists. In static AMC, this condition can be evaluated from statistical quantities. In dynamic mesh adaptation, however, such statistics are not available. To overcome this limitation, the proposed approach relies on the assumption that the instantaneous dissipation is predominantly the turbulent dissipation. The integral scale is then estimated from local instantaneous quantities, allowing a dynamic evaluation of the scale-separation criterion. This provides a continuous transition between DNS-like and LES-like regions during the simulation. The method is complemented by a laminar–turbulent discrimination based on a "sigma-sensor" (inspired by the sigma SGS model), enabling the identification of purely laminar zones. The approach has been assessed on a turbulent jet and on flow around a three-dimensional cylinder. Ongoing work focuses on improving near-wall treatments, in particular through prismatic layers generation on boundaries coupled to mesh adaptation and the introduction of dedicated kernels to stabilize the wall mesh and limit excessive boundary motion.

==== M4 – Improve mesh adaptation tools - B. Andrieu (ONERA), C. Benazet (ONERA), N. Dellinger (ONERA), G. Janodet (ONERA) & B. Maugars (ONERA) ====

Building upon the foundations established during ECFD7 and ECFD8 — which focused on periodic CAD-based mesh generation in EGADS and periodic parallel metric gradation — our latest developments for ECFD9 mark a significant step toward a fully automated, CAD-based periodic remeshing algorithm.
First, the parallel hierarchical remeshing algorithm prototype was improved by using a more elaborate ownership system in ParaDiGM to drive the mechanism that merges/dissociates the periodic interface mesh before/after the remeshing pass.
Second, the ability of the refine library (developed at NASA) to remesh non-manifold 3D configurations was investigated. Changes have been made to refine's operators to unlock remeshing near the merged periodic interface in 3D, which yielded promising results, but more work is needed to achieve industrial robustness. To enable CAD-based projections on both sides of the merged periodic interface, an algorithm for building a coherent periodized CAD model was implemented in the EGADS library.
This CAD-based periodic remeshing algorithm was validated in serial through a numerical simulation of the 2D LS89 turbine blade using the SoNICS solver. The results demonstrate that the mesh effectively adapts to capture the strongly anisotropic flow features while strictly respecting the periodic constraints and the geometric support.
Non-manifold mesh adaptation was applied to the ablation of a plate up to burnthrough, first in 2D and then in 3D. The burnthrough detection workflow was improved by developing a Python mini-toolbox for basic geometric queries, allowing the removal of non-physical solid fragments in the middle of the hole after burnthrough. The MMG library was also evaluated for its ability to handle non-manifold meshes, and it appears more suitable than the Refine library for this configuration. The workflow is satisfactory in 2D but needs improvement in 3D to continue the simulation after burnthrough.

==== M5 – Anisotropic mesh adaptation for multiphase flows - Robin Barbera (LEGI), Manuel Bernard (LEGI), Giovanni Ghigliotti (LEGI) & Roxane Letrounel (Safran) ====

This project investigates anisotropic mesh adaptation strategies for multiphase flows, with the objective of reducing computational cost while preserving an accurate representation of fluid interfaces. The approach relies on curvature-based anisotropic remeshing, where mesh anisotropy is locally controlled from interface geometry to ensure a prescribed discretization angle. A key limitation of anisotropic coarsening along interfaces is mass loss induced by interpolation during remeshing, which increases with tangential coarsening and therefore directly conflicts with anisotropic strategies. During ECFD9, this issue was addressed by introducing a high-order interpolation scheme for interface variables, replacing the default linear interpolation. The results show that high-order interpolation significantly reduces mass loss, allowing for much higher mesh anisotropy at the interface, at the cost of a limited computational overhead. In addition, the curvature-based adaptation strategy was extended from mean curvature to the full curvature tensor, enabling the mesh to align with the two principal curvatures of three-dimensional interfaces. The approach was demonstrated on canonical multiphase configurations, including droplet advection and rising bubble cases, showing substantial reductions in mesh size compared to isotropic simulations. Ongoing perspectives include coupling curvature-based adaptation with feature-based anisotropic remeshing to better capture turbulent structures away from the interface.

==== M7 – Increased mesh anisotropy for laminar and RANS applications - R. Barbera (LEGI), J.-B. Lagaert (LMO), T. Berthelon (LEGI), R. Letournel (Safran), M. Bernard (LEGI) & G. Balarac (LEGI) ====

This project addresses the limited level of mesh anisotropy obtained with current feature-based anisotropic remeshing criteria in steady laminar and RANS simulations in YALES2. While recent development of anisotropic mesh adaptation have significantly reduced computational cost, the achieved aspect ratios remain moderate (AR < 50), well below the levels commonly reported in the RANS literature (AR > 100). The objective of the project was to identify the main mechanisms that limit anisotropy in practice, including numerical noise in the resolved quantities, inaccuracies in Hessian computation, the formulation of the criterion itself, ... During ECFD9, the current anisotropic criterion applied to a vectorial quantity of interest (QOI) implemented in YALES2 was reformulated as the minimization of a residual-based error estimator. A Newton optimization strategy was introduced to assess whether the theoretical optimum of the criterion differs from criterion use in practice.The approach was analyzed on the Kovasznay flow, and the optimal solution was shown to be very close to the criterion currently used in YALES2. Comparisons with alternative criteria from the literature and based on scalar QOI further demonstrated similar mesh convergence, highlighting the robustness of the YALES2 approach and its main advantage: a flow-independent, non-dimensional target error. Ongoing investigations focus on quantifying the influence of numerical noise in the resolved quantities and Hessian discretization on the achievable mesh aspect ratios.

=== Two-phase flows - J. Carmona (CORIA), N. Gasnier (Safran) & I. Tsetoglou (M2P2) ===

==== TP1 - Simulation of core shifting during investment casting - Y. Mayi (Safran), M. Cailler (Safran), S. Meynet (GDTech) ====

==== TP2 - Lattice Boltzmann method for free-surface two-phase flow - J. Lu (M2P2), Y. Mediene (M2P2), I. Tsetoglou (M2P2) & S. Zhao (M2P2) ====

The project aims to reproduce and improve a two-equation free-surface Lattice Boltzmann Method (LBM) model for two-phase flows. The original free-surface model features a sharp interface and good numerical stability, but it neglects the gas phase and is therefore limited to two-phase flows in which gas effects are negligible. The recently developed two-equation free-surface LBM model (Liu Y., Sun D., Zhang Z., et al., Physics of Fluids, 2024, 36(3)) incorporates the gas phase, enabling interactions between the two phases. However, this model suffers from a lack of mass conservation and insufficient accuracy in curvature computation.
To overcome these limitations, an auxiliary distribution function is introduced to track mass evolution, thereby decoupling mass conservation from pressure evolution and restoring global mass conservation. In parallel, a pseudo-smoothing step is implemented to achieve more accurate calculations of interface normals and curvature. These improvements are validated through two benchmark test cases. (1) A Laplace test involving both static and advected droplets. It demonstrates exact mass conservation and a significant enhancement in surface tension modeling. (2) A two-phase Poiseuille flow. It shows good agreement with theoretical predictions, validating the viscous coupling between the two fluids.
Future work will focus on improving information exchange across the interface to reduce numerical oscillations and enhance numerical stability, as well as on conducting more complex validation cases.

==== TP3 - Modeling of a gear wheel immersed in an oil bath ====

==== TP4 - Implementation of a granular temperature model - T. Ndereyimana (Université de Sherbrooke), S. Moreau (Université de Sherbrooke), Y. Dufresne (Enerkem) ====

In gas–solid systems such as fluidized beds, clusters of particles naturally appear. These clusters tend to exhibit a Gaussian velocity distribution around an average velocity, with a spread that depends on the local environment of the cluster.
In coarse-grained DEM simulations, real particles are replaced by numerical parcels representing groups of particles in order to reduce the computational cost associated with a large number of particles. In this approach, all particles within a parcel are assumed to move at the same velocity; consequently, no velocity distribution is represented.
This project focuses on comparing two approaches to model the standard deviation of the velocity distribution within a parcel: (1) a local averaging method and (2) a kinetic-theory-of-granular-flow-based methodology. The former computes the standard deviation based on the velocities of surrounding parcels, while the latter relies on two-phase flow theory in which this standard deviation is explicitly modeled.
Both methodologies predict a high standard deviation in the vicinity of gas bubbles in the fluidized bed and lower values in very dense and very dilute regimes. However, the local averaging method tends to increase the computational cost by requiring the detection of neighboring parcels for each parcel.

==== TP5 - Jet-A1 cavitation modeling - P. Benez (Safran), J. Carmona (CORIA) ====

==== TP6 - Comparison of JICF models for turbulent reactive applications, S Puggelli (SAE), L Carbajal Carrasco (SAE), E Charles (CERFACS), L Gicquel (CERFACS) ====
Simulating jet-in-crossflow (JICF) configurations using LES for industrial applications is complex due to the high computational cost associated with fully resolving jets, especially for computations in reactive conditions. Within a purely Lagrangian approach, certain physical phenomena—such as surface or column break-up—are not effectively captured. This project aims to assess the Lucid model (Charles et al., International Conference on Numerical Combustion, 2025), which introduces a liquid column representation into the Lagrangian framework, by comparing its performance against a fully resolved SPS simulation of a JICF case validated by experimental data. Additionally, a pure Lagrangian injection case, namely without any ad-hoc model for the jet, is also analyzed. The results indicate that the Lucid model better captures surface breakup dynamics when compared to the pure Lagrangian approach. Nevertheless, both Lagrangian-based methods overlook aerodynamic blockage effects of the liquid column, potentially influencing the downstream distribution of droplets.

==== TP7 - Validation and extension of PCS solver for cryo tanks ====

==== TP8 - Jet-in-crossflow simulation with the Hybrid SPH-FVM solver ====

==== TP9 - Multi-physics effects modeling in film flows - N. Gasnier (Safran), P. Portais (CORIA/Safran), L. Voivenel (CORIA), E. Bourrel (CORIA), M. Cailler (Safran) ====
The project aimed at improving the multi scale model for parietal film flows implemented in the YALES2 platform. This model, based on the Shallow Water equations, allows to describe the dynamics of a thin liquid layer spreading over dry walls at a reduced computational cost. First, the numerical method designed during ECFD8 to convert impinging Lagrangian droplets into film data has been extended to account for droplet splashing and rebound phenomena. Then, a sensitivity analysis has been initiated to determine the influence of inlet conditions on the properties of the droplets generated by film atomization. Preliminary results showed that an increase in the gas velocity causes a significant increase in the number of droplet generated, and a large decrease in the drops diameter. Then, the film dynamics model has been extended to rotating walls by including inertial and Coriolis forces in the momentum conservation equation. A first validation of the implementation has been conducted by analyzing the spreading of a liquid film generated by impinging droplets over a rotating disk at high angular speed, which gave promising results. Finally, a film temperature equation has been added to include thermal effects in the thin film model, this additional equation describes the temporal evolution of the surface temperature of the liquid, which is primarily affected by the temperatures of the wall and of the surrounding air. The influence of thermal effects on the dynamics of the liquid is taken into account through the temperature-dependency of the surface tension, which is likely to cause the onset of Marangoni currents due to heating discrepancies.

==== TP10 - Solid-Fluid Coupling for Nucleate Boiling Simulations - M. Umair (LEGI), G. Ghigliotti (LPSC), H. Lam (LEGI), M. Bernard (LEGI), R. Barbera (LEGI), G. Balarac (LEGI) ====

Ecfd:ecfd 8th edition

2025-02-11T13:11:18Z

Luis.carbajal-carrasco: /* C6 - Flame stabilitity of flame-holders within reheat conditions */

{{DISPLAYTITLE: ECFD workshop, 8th edition, 2025}}

== Description ==

* Event from '''27th of January to 7th of February 2025'''
* Location: [https://www.sport-normandie.fr/le-centre/le-site-de-houlgate Centre Sportif de Normandie], Houlgate, near Caen (14)
* Two types of sessions:
** common technical presentations: roadmaps, specific points
** mini-workshops. Potential workshops are listed below
* Free of charge
* Participants from academics, HPC center/experts and industry are welcome
* The number of participants is limited to 68.

* Objectives
** Bring together experts in high-performance computing, applied mathematics and multi-physics CFDs
** Identify the technological barriers of exaflopic CFD via numerical experiments
** Identify industrial needs and challenges in high-performance computing
** Propose action plans to add to the development roadmaps of the CFD codes
* Organizers
** Guillaume Balarac (LEGI), Simon Mendez (IMAG), Pierre Bénard, Vincent Moureau, Léa Voivenel (CORIA).

[[File:ecfd8.png|600px|link=https://ecfd.coria-cfd.fr/index.php/Ecfd:ecfd_8th_edition]]

[[File:Acknowledgments_ecfd8.png|text-bottom|600px]]

== News ==

* 23/10/2024: First announcement of the '''8th Extreme CFD Workshop & Hackathon''' !
* 22/11/2024: Deadline to submit your project

== Thematics / Mini-workshops ==

These mini-workshops may change and cover more or less topics. This page will be adapted according to your feedback.

To come...

== Projects ==

=== Hackathon GENCI - P. Begou, LEGI ===
This ECFD8 GENCI Hackathon was a rich event, involving 4 differents CFD codes (AVBP, ParaDIGM, SONICS and YALES2) using various paradigms (C++/cuda/hip, Fortran/OpenMP/OpenACC) with several SDKs (AMD, Cray/HPE, Nvidia, Gnu) on a large range of GPU architectures (Nvidia A100, GH100, AMD instinct Mi210, Mi250, Mi300). This two-week event benefited from a high level support from three HPC mentors, two on-site from AMD (J. Noudohouenou and A. Tsetoglou) and one remote from CINES (M. Boudaoud).

==== H1 - ParaDIGM and SONICS on GPU, B. Maugars, G. Staffelbach, R.Cazalbou and B. Michel (ONERA)====

==== H2 - AVBP GPU offloading based on OpenMP, M.Ghenai, L. Legaux and A. Dauptain (CERFACS) ====

This hackathon provided a valuable opportunity to work on GPU offloading for AVBP. In the past, significant efforts were made to offload the entire AVBP code to GPUs. OpenACC was the primary strategy chosen, mainly due to access to NVIDIA's support, along with the availability of both software and hardware. This approach achieved good scalability performance.
Recently, with the deployment of new supercomputers like ADASTRA at CINES, some issues have emerged when running AVBP on AMD GPUs, including both MI250 and MI300. The closed-source nature of the Cray environment has also prevented CERFACS from deploying AVBP on local MI210 GPUs.
This hackathon was a great opportunity to address these challenges by exploring a new approach using OpenMP. An automatic translation tool was used to convert approximately 2,700 OpenACC directives to OpenMP, with each directive manually verified and fine-tuned afterward. AVBP with OpenMP had already been tested on NVIDIA GPUs, and during this hackathon, the focus was on extending support to AMD GPUs.
Two compilers were used: Cray and the newly released AFAR from AMD. With the support of AMD and CINES, a working environment for compiling AVBP was set up, and performance-related issues were identified. Additionally, two mini-apps were used for benchmarking. One of them achieved a 2.5× speedup when compiled with AFAR compared to Cray.
The next steps involve adapting the code to address necessary modifications, such as fixing issues related to Fortran indirections, and continuing performance evaluations with mini-apps. Further comparisons will be conducted using both compilers against results obtained with NVIDIA’s NVHPC.

==== H3 - YALES2 GPU from OpenACC to OpenMP, P. Bégou (LEGI), V. Moureau, G. Lartigue (CORIA) and R. Dubois (IMAG) ====
This Hackathon focuses on running Yales2 code on AMD Instinct Mi250 and Mi300 GPUs of the Adastra supercomputer (CINES).
Previously, a first solver in the Yales2 CFD code was successfully ported on the GPU accelerators of the Jean-Zay supercomputer (IDRIS) using Nvidia SDK but difficulties remain on Adastra AMD GPUs, mainly related to the available development tools. High compilation time and the impossibility to use debug flags at compile time as soon as OpenACC is enabled are a real challenge when tracking errors. The current project is to evaluate a freshly deployed version (at the begining of the workshop) of the AMD Fortran compiler. This requires moving to OpenMP paradigm, starting from scratch since the OpenACC branch has largely diverged from the master one while tracking spurious remaining bugs.
If the AMD compiler is able to build the cpu version of Yales2 "out of the box" (wich is not the case for Cray Fortran), the compilation time for each file is significantly higher. However, setting up a 2 stages dynamic compilation process allows for high parallelism that is not possible with Cray Fortran 18 and the library build time drops from nearly 2 hours (Cray Fortran 18) to 17 minutes (Amd Fortran compiler).
Large kernels have been ported from OpenACC to OpenMP, raising some difficulties when offloading intrinsics functions or using strutures attributes in kernels loops. These limitations were also known in the previous OpenACC work. The goal was mainly to check the correctness of the results. The offloading of the complex data structure of Yales2 code was then investigated. Here again some limitations of the "young" compiler were discovered and workarounds were implemented. Several reproducers were built during this ECFD8 and provided to developpers by the 2 on-site AMD engineers.
Preliminary tests on micro-applications show good performances of the generated binaries proving that this compiler could be a serious alternative on AMD GPUs and the goal is now to focus on this SDK in an OpenMP strategy while checking the portablility of this new implementation in Nvidia, Cray/HPE (and Gnu ?) environments.

=== Mesh adaptation - A. Grenouilloux, ONERA & G. Balarac, LEGI ===

=== Numerics - M. Bernard, LEGI & G. Lartigue, CORIA ===

==== N1 - Traction open boundary condition ====

==== N2 - Treatment of Inlet conditions in High-Order solver. M. Bernard (LEGI), Ghislain Lartigue (CORIA), Guillaume Balarac (LEGI) ====
In the context of node-centered Finite Volumes Method, spacial accuracy of a numerical scheme depends on ability to evaluate accurately fluxes through interface of each control volume (CV). Such accurate evaluation is not straightforward, especially when dealing with distorted grids. This project follows the work of [1] where fluxes use pointwise quantities, which are reconstructed from integrated quantities advanced in time. During the previous edition of the ECFD, a new data structure has been developed to store data at location of the boundary conditions facelets, with application to wall boundary conditions. During this 8th edition of the ECFD, we used the same data structure, but dedicated to the treatment of inlet conditions.
The inlet condition is then either imposed directly at facelets center, or at nodes position them extrapolated to facelets center by use of Taylor expansion. For this later solution, high-order treatment requires the successive derivatives to be computed in the plane of the boundary condition. This is not done yet, leading for the moment to low accuracy results but the framework is ready for upcoming implementation.

[1] ''A framework to perform high-order deconvolution for finite-volume method on simplicial meshes, , Bernard et. al., IJNMF 2020''

==== N3 - Conservative mesh-to-mesh interpolation. M. Bernard (LEGI), Ghislain Lartigue (CORIA), Guillaume Balarac (LEGI) ====

Mesh to mesh interpolations occur quite often in CFD simulations : in the context of adaptative mesh convergence or in the case of dynamic mesh adaptation for for example.
Quality of the solution on the destination grid will depend on the characteristics of the interpolation method.
In this project, we did not focus on accuracy of the interpolation method but rather on conservativity characteristics.
A conservative interpolation ensures that the integral of the data on the source grid is exactly retrieved on the destination grid.
This property is highly interesting when dealing with scalar quantities or phase indicators, whose values should remained bounded.
In the context of nodes centered Finite Volume schemes, the methodology we used consists in (i) reconstructing element quantity from average nodal quantities on source grid.
Then, for a cell of the destination mesh, (ii) computing the geometrical intersection between cells of source and destination meshes to evaluate to evaluate the rate of quantities they.
Eventually, (iii) redistributing the solution from elements to control volumes of the destination mesh.
The overall process is fully conservative as it is based on geometrical intersection of locally integrated quantities.
The methodology as been implemented and tested on a few basic configurations and the conservativity is retrieved.

==== N4 - Determination of timestep in semi-implicit solver. T. Berthelon (LEGI), G. Balarac (LEGI), H. Lam (LEGI), M. El Moatamid (CORIA) ====
In order to reduce the computation time associated with incompressible LES simulations, an implicit time integration, based on BDF schemes, has been developed within the ICS solver. This integration eliminates the stability constraints associated with explicit schemes, and therefore opens up the question of the appropriate choice of time step.
In parallel, recent work has been carried out on meshing criteria in LES. The strategy in place consists of adapting the mesh by distinguishing two zones:
- "DNS" zones, where the meshing criterion is based on an estimate of the adimensioned spatial error.
- "LES" zones, where the meshing criterion is based on Kolmorogov theory.
During this project, the spatial criteria were extended to include temporal criteria. In the "DNS" zones, the time step is chosen using an estimate of the temporal error of the BDF scheme judiciously scaled to match the spatial error. In the "LES" zones, the time step is chosen using a scaling law associated with fully developed turbulence.
The new time step selection strategy has been tested on the case of a turbulent jet and leads to an accuracy equivalent to the explicit case while reducing the simulation return time by a factor of nearly 3.

Another aspect of this project was to integrate certain implicit temporal schemes (C-N and SDIRK) recently developed by Mr. El Moatamid into the incompressible solver.

==== N5 - Local timestep. T. Berthelon (LEGI), M. Bernard (LEGI), G. Balarac (LEGI) ====
RANS modelling has recently been developed within the YALES2 library. With this modeling strategy, the objective is to reach as quick as possible a steady state.
During this project, we investigate the use of a local time step to reduce the time to solution of steady computation in the incompressible solver.
This implies solving a variable-coefficient Poisson equation. Encouraging results were obtained in the simple case of "Couette plan" flow artificially constrained by a mesh variation. In fact, the use of local time-step reduce drastically the time to solution on this configuration. This method needs to be tested on real RANS case.

==== N6 - Distributed version of DOROTHY ====

==== N7 - Implicit time advancement for low-Reynolds number flows with particles. S. Mendez, C. Raveleau (IMAG), M. El Moatamid, V. Moureau (CORIA) ====
IMAG runs numerous simulations of red blood cells under flow. Those simulations are at low Reynolds number (0.001 to 1.0, typically). Splitting of the time advancement is used to treat the diffusion terms implicitly, albeit with an important numerical cost: implicit diffusion is 50 to 60% of the computational cost. Recently, M. El Moatamid implemented a genral framework to deal with implicit time advancement for scalars. In this project, the general method has been transposed to the advancement of the velocity field in the ICS and RBC solvers of YALES2/YALES2BIO. This enables testing various linear solvers (GMRES based). However, such solvers do not decrease the CPU time compared to the existing implementation. However, while working on this, it was identified that residual recycling was not activated in the current implementation of the implicit diffusion. This sped up the implicit diffusion cost by 35%, for a total gain of 20% for the computation. In addition to this achievement, moving to the framework coded by Moncef will have other beneficial side effects: we anticipate simplifying the implementation, with an easier merging between YALES2BIO and YALES2. The method will also be implemented in the electrosatic solver, for which the Poisson problem should benefit from the new GMRES-based solvers. In addition, this project highlights the importance of improving the treatment of stiff source terms in the red blood cells simulations, to be able to overcome the current limitation in time step due to those term and have a chance to benefit from higher-order time schemes, efficient at high Fourier numbers.

==== N8 - Boundary Element Method in YALES2. B. Thibaud, S. Mendez (IMAG), G. Lartigue, P. Benard (CORIA), F. Nicoud (IMAG) ====
In the context of microfluidic systems for diagnosis, the Boundary Element Method alows to solve linear PDE such as electrostatic or Stokes. With well chosen kernel functions and the divergence theorem, this method allows to write on the boundary condition only the initial volumic problem. This project aimed at exploring the feasibility of the BEM in the context of massively parallel unstructured solver like YALES2 by developping a Julia demonstrator. The first step have been to implement and validate the method on simple configurations for the Laplace's equation. Only Neumann problems were considered (Dirichlet boundary conditions imposed). In a second time, the multi-domain approach has been identified to be the most suited in the framework of YALES. The inner domain is partitioned on each processor, each having a part of the physical boundary and interfaces between them. Every processor solve its own boundary problem and a parallel Dirichlet-Dirichlet fixed-point is used to converge the interface problem on the all domain. Applied to the ring case, with one interface, we managed to reproduce the linear convergence of the P-DD method.

=== Turbulence - L. Voivenel, CORIA & P. Bénard, CORIA ===

==== T1 - FSI-1D strategy for internal flows====

Many applications developed at Safran Aerosystem are based on internal turbulent flows coupled to a moving body. 2 cases were studied during this ECFD:

'''Case 1 (Incompressible flow)''': Translation of a piston subjected to a pressure difference in a pipe.

The challenges of this case are twofold: the small gap between the piston and the pipe and the large pressure gradient across the piston (>100bar). During the 1st week of ECFD, the CLIB (Conservative Lagrangian Immersed Boundary) solver was tested on this case. The study showed that the solver was unable to ensure the impermeability of the solid under these pressure conditions. In the rest of the study, a porous medium following Darcy's law will be added to the penalty force of the immersed solid to fully satisfy the impermeability of the piston.

'''Case 2 (Compressible flow)''': Rotation of a butterfly in a discharge vane.

The coupling between the ECS (Explicit Compressible Solver) and ALE (Arbitrary Lagrangian Solver) solvers having recently been implemented, this strategy was tested to model the opening of the valve by rotation of the butterfly. The challenge here lies in the small gap between the bottom of the butterfly and the vane casing. To limit the simulation cost, the gap is meshed with 1 element. In this case, MMG succeeded in adapting the mesh up to a critical angle at which the gap becomes too small (Work In Progress).

==== T2 - Dynamic Smagorinsky in Dorothy ====

==== T3 - Turbulence injection strategy for compressible flows ====

==== T4 - Improve wind farm modeling and simulation workflow ====
The YALES2 library includes an advanced modular implementation of the Actuator Line Method (ALM). This approach remains state-of-the-art when performing an LES-based analysis of a wind turbine wake. The method also provides an accurate assessment of the aerodynamic loads applied on the turbine. Still, applying this method to investigate a wind farm flow can be challenging, both in terms of computational cost and simulation setup. For instance, an inadequate management of the wind turbine individual modeling parts in a HPC context can end up being the main bottleneck of the simulation. From another perspective, a wind farm is usually composed of more than 50 wind turbines. For such a case, setting up all YALES2 required inputs manually can be very tedious and error-prone. This project thus mainly aimed to optimize the YALES2 ALM implementation and the user experience around it. Additionally, a cost-effective alternative to the ALM when modeling wind farm flows, namely the Rotating Actuator Disk Method (ADM-R), has been implemented for further investigations.

'''WP1''': Improve Actuator set rotor modelling
* Parallel processing of the ''actuator sets'' used to model the wind turbines
The implementation of the Actuator Line Method (ALM) into the YALES2 library leads to poor performances when many wind turbine rotors are set. Indeed, each rotor object is a derived type treated sequentially by all the processors participating to the computation. With 30 turbines in a computation, the return time is increased by 70% while the arithmetic intensity appears to be low. The objective of this sub-project is to improve the computation performances of the ALM already identified during the 5th iteration of the ECFD: (i) Assign one MPI communicator by rotor object gathering the processors close to the turbine and set-up a master/slave processus by communicator. This will allow the simultaneous rotors computation and reduce the number of MPI exchanges. (ii) Work on the domain decomposition to limit the number of processors attributed to each turbine. This would reduce or even eliminate MPI communications. During this workshop edition, the work focused on the re-synchronisation of the algorithm steps by allowing some packing an unpacking of the object to allow the transfert of the object inbetween workers. This is of major importance to enable load balancing and mesh adaptation during the temporal loop. This work required the refactoring of the involved oject structures.

* Rotating Actuator Disk Method (ADM-R):
According to the usual guidelines, the mesh requirements of the ALM, to profit entirely from its reachable accuracy, can be difficult to achieve or even unaffordable when simulating a wind farm flow, especially from the industrial point of view. Alternatives are available in the literature for this kind of application. Likely, the methods from the Actuator Disk family are the most prominent ones. Several kinds of implementation exist, which mostly differ by their capability to include the wake rotation. During the workshop, a new method from the Rotating Actuator Disk kind has been implemented and underwent an early validation on a single turbine setup. Applications to wind farm flows will follow.

'''WP2''': Improve tools User Experience

Three Python tools have been developed or improved :
*The first tool is the wind farm previsualisation tool, 'y2_wind_previsualization', which is used before the calculation run. This provides an interactive HTML interface for viewing global data for each turbine on the farm (position, hub height, yaw angle, etc.). The tool traces all of these via the parsing of the input file.
* The second tool is for duplicating rotor templates for a wind farm (`y2_wind_duplication`). This tool was developed in the previous ECFD, but this time it has been refactored and incorporated into the y2tools package.
* The third and final tool is a post-processing tool for the temporal processing of global wind turbine simulation metrics (Thrust, Power, etc.), `y2_post_wind`. This tool generates an interactive HTML plot of time-dependent global quantities.

==== T5 - Improve atmospheric inflow turbulence ====
Atmospheric inflow turbulence is generated using the precursor database method. A half-channel flow driven by a pressure gradient is used to obtain the inflow which is used as inlet boundary condition for the wind turbine simulation domain. This project aimed to improve the whole methodology, from generation to injection.

* WP1: Improve inflow generation
Anand: pressure controller

* WP2: Improve injection methodology (method A)
The previous workflow used plane probes in the ASCII format to sample the flow. The COWIT2 toolbox was used to convert the file into turbulence box (.man format). While functioning, this methodology had two major flaws. First the probe files are heavy ~O(10Go). Second, the method requires a lot of human effort, allowing numerous sources of errors.
During this workshop, a new methodology has been developed. First, the probes are generated using the HDF5 format (now available for all probe types), leading to lighter file ~O(1Go). Second, Y2_tools is used to read HDF5 format (working for probes and temporals). HDF5 file is then converted into a Look-up Table. Finally, the Look-up Table is read directly by YALES2 as a boundary conditions.

* WP3: Improve injection methodology (method B)
Even though improvements achieved in WP2 prove to be very handy while removing many potential human errors, injecting a turbulent inflow through wind boxes ('offline' precursor approach) can sometimes remain cumbersome for several reasons: (1) no periodicity is enforced in the streamwise direction of those boxes, (2) potential high memory consumption, and (3) the boxes need to be moved to other cores whenever a mesh adaptation occurs. An alternative consists in co-simulating the precursor flow and the flow of interest (refered as the 'successor' simulation) at the same time ('online' precursor approach). The inlet boundary condition for the successor flow is then obtained by mapping the outflow of the precursor domain. During the workshop, some work has been initiated to implement this kind of coupling using the CWIPI library, for which YALES2 provides already an interface.

==== T6 - FSI model in Dorothy ====

=== Two Phase Flow - J. Leparoux, SAFRAN & J. Carmona, CORIA ===

==== TP1 - Towards very small contact angles in Nucleate boiling ====

Participants: Henri Lam (LEGI), Mohammad Umair (LEGI), Manuel Bernard (LEGI), Robin Barbera (LEGI) and Giovanni Ghigliotti (LPSC)

The boiling solver (BOI) was not able to accurately impose a contact angle (angle formed by the two-phase interface on the wall) at values lower than 30°. This angle is needed when simulating nucleate boiling. A similar limitation in contact angle value was applying to the spray (SPS) solver. A modified version of the level set reinitialization has been implemented during ECFD8, based on a different normal vector in the blind spot region around the contact line, vector now chosen to be a zero-order extension from outside the blind spot. This modification, that implied other modifications to the level set reinitialization in the blind spot, has been tested successfully on the spray solver (where no phase change occurs). Then, this new reinitialisation has been tested in the boiling solver for nucleate boiling, with great improvements. Now simulations of nucleate boiling at very small contact angle (10°) can be accurately performed.
In the meanwhile, the level set reinitialization algorithm has been streamlined and the computational cost greatly reduced, resolving a computational cost issue that appeared when using the contact angle imposition both in the spray and boiling solvers, and that hampered its use in industrial configurations.

==== TP2 - Modeling spray-film interactions ====

Participants: Nicolas Gasnier (EM2C-SafranTech), Julien Leparoux (SafranTech), Mehdi Helal (CORIA-SafranTech) and Julien Carmona (CORIA)

The numerical simulation of sheared liquid films over walls with conventional high-fidelity methods requires fine meshes to depict accurately the dynamics of the fluids, as well as the gas-liquid interactions at the phase interface. In the context of thin film flows (with a thickness h ~ 1.0E-4 m), the spatial resolution required to ensure the validity of the simulations can be computationally prohibitive. To tackle this issue, a reduced-order model of thin film dynamics based on the Saint Venant equations has been implemented in the YALES2 platform over the last months. This numerical model is able to reproduce accurately the dynamics of strongly sheared film flows over partially wetted surfaces, and to take into account capillary effects. The objective of the ECFD8 was to couple the thin film framework with the pre-existing multiphase methods of the YALES2 platform: the Eulerian multiphase solver - based on the ACLS -, and the Lagrangian solver. This coupling was performed in three steps: first, a numerical method was designed to convert impinging Lagrangian droplets into source terms for the thin film model. Then, a phenomenological model was implemented to depict the atomization of films at sharp edges under the action of a high speed gas flow. The liquid film is converted into Lagragian droplets, whose dimensions are computed based on the PAMELA model. Finally, a numerical method has been designed to convert Eulerian droplets into source terms for the film model. This method is based on a transfer of the liquid properties from the nodes to the boundaries by using a set of fictitious particles. It ensures conservation of the liquid mass, and it is more robust than the first attempts that had been made during ECFD7.

==== TP3 - High-fidelity two-phase flow simulations of the purge of a fuel feed line ====

Participants: Thomas LAROCHE (Safran HE), Romain JANODET (Safran AE), Julien Leparoux (Safran Tech) and Melody Cailler (Safran Tech)

During the second week of the ECFD8, the fuel feed line purge process has been numerically investigated. In the context of aeronautical engines, the fuel feed line—carrying the fuel from the fuel tank to the injectors within the combustion chamber—needs to be purged at engine shutdown. This is intended to prevent fuel stagnation near hot metal parts, which could lead to coke formation and therefore decrease engine performance. Since this complex phenomenon is mainly driven by two-phase flow physics, the spray solver (SPS) of the YALES2 library has been considered in order to understand the physics of such process. The numerical setup was first converged on a simplified test case: the possibility of driving the flow dynamics with inlet and outlet pressure conditions was tested beforehand on a single-phase, incompressible case, and then on a two-phase flow problem. The setup has then been successfully applied to an industrial configuration: a pressure-swirl injector connected to a reduced portion of the fuel feed line. Due to the large scale of the domain, the interface resolution was set to 50μm, which is intentionally coarse for such problem. This initial computation successfully ran up to 3ms of physical time during the workshop, proving YALES2's capability to model the fuel purge. The computation is to be continued and analyzed further even after the workshop.

==== TP4 - Volume of Fluid solver in YALES2 ====

Participants: Léa Voivenel (CORIA), Julien Carmona (CORIA), Mehdi Helal (CORIA), Pierre Portais (CORIA), Julien Leparoux (Safran Tech), Mélody Cailler (Safran Tech) and Nicolas Gasnier (EM2C / Safran Tech)

==== TP5 - Implement a local operator to distribute the solid volume of a particle over multiple cells ====

Participants: Théo Ndereyimana (Université de Sherbrooke), Stéphane Moreau (Université de Sherbrooke)

In the CFD-DEM, the cell size is required to be larger than the particle size for stability condition and keep feasible solid volume fraction. However, some applications require a cell size smaller than the particle. During this 8th edition of the ECFD, the use of operators to distribute the particle volume over multiple cells, ensuring a feasible solid volume particle has been tested on a fluidized bed configuration. The main operators tested (gather-scatter filter and gaussian filter) showed a tendency to blur the void structure interfaces. The equivalence of the gaussian filter of bandwidth <math>b</math> and the resolution of a diffusion equation over a pseudo-time <math>T=b^2/4</math> has been verified.
One anisotropic diffusion constant has been tested and shows a possibility to adress the sharpness requirements.

Another objective was to develop a post-processing tool to detect and track the void structures (bubbles) in a fluidized bed. Based on previous work from J. Carmona, a tool to track the bubbles has been initiated.

==== TP6 - Complex thermodynamics in sloshing tanks ====

Participants: C. Merlin (AGS), D. Fouquet (CORIA), V. Moureau (CORIA), J. Carmona (CORIA) and G. Lartigue (CORIA)

During this ECFD, the phase change solver was improved. After introduction of conservative transport for enthalpy and energy, a new framework for multi fluid two phase flow was used that relies on a new reinitialization of the conservative level set, the transport of discontinuous scalars like temperature, energy or enthalpy and the transport of weight to ensure consistency. The pressure loop was improved to ensure mass and energy conservation while meeting the low Mach hypothesis. Many 1D test cases were performed with only the gas phase and two phase flow to validate the pressure loop. The new framework for the transport of discontinuous scalar was first derived for temperature with constant properties in each phase. A first step was to extend it to sensible enthalpy with ideal gas behavior for the gas phase. Then, the framework was extended to NIST tabulated gas & liquid with ideal or real gas behavior. After some 2D cases, a tank pressurization test was investigated as well as a sloshing test without phase change (just the pressure drop due to the increase of the thermal exchange between the liquid and the gas phase).

=== Combustion - Y. Bechane, CORIA & S. Dillon, SAFRAN & K. Bioche, CORIA ===

==== C1 - LES of the thermal degradation of a composite material ====
Participants: A. Grenouilloux (ONERA), K. Bioche (CORIA), N. Dellinger (ONERA) and R. Letournel (SafranTech)

The FIRE test bed is an experimental air-propane burner operated by ONERA. It is dedicated to the study of the thermal degradation of composite materials. This project concerned the implementation of a three-solver coupling methodology to simulate the dynamics of the impinging flame. The methodology considered is based on the coupling between the variable density solver (VDS) and the radiative solver (RDS) of the massively parallel library YALES2 and the solver dedicated to the degradation of composite materials, MoDeTheC, developed by ONERA. Given the typical test times of the order of tens of seconds, a methodology based on 2D axisymmetric calculations was considered. Various tests were performed to determine the optimal coupling frequency between solvers. Cases dedicated to the injection of pyrolysis gasses were set up, with the aim of simulating the auto-ignition phenomenon. Comparisons with experimental data are presented.

==== C2 - Flame stabilization by NRP plasma discharge ====

==== C3 - Extending and validating a generalized formalism of virtual chemistry ====

==== C4 - Turbulent combustion model for NOx prediction ====

==== C5 - Towards 3D simulation of detonation combustion ====

==== C6 - Flame stabilitity of flame-holders within reheat conditions ====

Participants : N. Detomaso (Safran AE), R. Janodet (Safran AE), L. Carbajal (Safran AE)

The mechanisms of flame stabilisation in a flame-holder are still not completely understood. This is particularly true for reheat conditions : the hot viciated gases at inlet, highly compressible flows, and a strong liquid/gas coupling make the flame stability hard to predict. During the ECFD8, simplified configurations of flame-holders within reheat conditions were analysed. After some try and error simulations due to simplifications of the flame-holders, cases leading to flame stability (or not) were identified. Post-processing tools were developped in order to recover criteria relevant to the flame stability. This step marks the beginning of a systematic mapping of dimensionless characteristic times ratios, and a comparison with integral quantities.

==== C7 - Thermal radiation in oxyflames ====

==== C8 - A first step toward hybrid CPU / GPU for reactive flow in YALES2 ====

Participants: M. Laignel (CORIA), G. Lartigue (CORIA), K. Bioche (CORIA) and V. Moureau (CORIA)

In numerical simulations of reacting flows, one of the most computationally intensive tasks is the evaluation of source terms resulting from chemical reactions in the species transport equations. This step can account for up to 90% of the total simulation cost , depending on the complexity of the kinetic mechanism involved. To reduce this cost, various techniques such as mechanism reduction, virtual chemistry, etc. have been explored. However, the emergence of GPUs as powerful accelerators offers a promising alternative by providing massive parallelism. Despite their potential, GPUs often require significant adaptation of CPU-based codes. This project aims to address this challenge by taking a first step towards a hybrid CPU/GPU framework for reactive flow simulations. Specifically, the focus is on coupling Y2 with the updated version of the stiff time integration solver (CVODE), which is compatible with GPU (CUDA, HIP, OpenMP). The ultimate goal is to establish a foundation for hybrid computations by implementing and testing the updated solver on simplified test cases.

==== C9 - Soots numerical modeling ====

==== C10 - TECERACT : Tabulated chemistry generator for aeronautical combustion ====

==== C11 - Exploring efficient tabulation strategies for detailed chemistry ====

==== C12 - Dynamic sub-grid-scale modelling of multi-regime flame wrinkling ====

==== C13 - LES of a semi-industrial burner using a non-adiabatic virtual chemical scheme ====

=== User Experience & Data - L. Korzeczek, GDTECH ===

==== U1 - Low-fidelity (RANS) rotor/stator simulations, application to Kaplan Turbine - Y. Lakrifi, G. Balarac (LEGI), R. Mercier (SAFRAN), V. Moureau (CORIA) ====

==== U2 - Coupling PyTorch/YALES2, combustion cartesian look-up tables - J. Leparoux, N. Treleaven, S. Dillon (SAFRAN), K. Bioche, G. Lartigue (CORIA) ====

Participants: Julien Leparoux (Safran Tech), Kévin Bioche (CORIA), Ghislain Lartigue (CORIA), Nicholas Treleaven (Safran Tech)

Neural Networks offer a promising alternative to Cartesian look-up tables for combustion simulations, reducing memory footprint. In this project, we investigated how to integrate an NN model for real-time inference in the YALES2 platform, exploring two approaches: a Python interface and a Fortran Torch binding (using FTorch[https://github.com/Cambridge-ICCS/FTorch]). We validated that the model remains accurate when embedded online and identified improvements for robustness. Inference costs were evaluated on a Mac M3 and the Austral cluster, revealing a strong dependency on data volume. To optimize efficiency, we propose grouping cells at the processor level.

==== U3 - Yales2 Trame Editor, toward a fully featured graphical user interface for YALES2 - L. Korzeczek, S. Meynet (GDTECH), J. Leparoux, M. Cailler (SAFRAN) ====

Ecfd:ecfd 8th edition

2025-02-11T13:07:34Z

Luis.carbajal-carrasco: /* C6 - Flame stabilitity of flame-holders within reheat conditions */

{{DISPLAYTITLE: ECFD workshop, 8th edition, 2025}}

== Description ==

* Event from '''27th of January to 7th of February 2025'''
* Location: [https://www.sport-normandie.fr/le-centre/le-site-de-houlgate Centre Sportif de Normandie], Houlgate, near Caen (14)
* Two types of sessions:
** common technical presentations: roadmaps, specific points
** mini-workshops. Potential workshops are listed below
* Free of charge
* Participants from academics, HPC center/experts and industry are welcome
* The number of participants is limited to 68.

* Objectives
** Bring together experts in high-performance computing, applied mathematics and multi-physics CFDs
** Identify the technological barriers of exaflopic CFD via numerical experiments
** Identify industrial needs and challenges in high-performance computing
** Propose action plans to add to the development roadmaps of the CFD codes
* Organizers
** Guillaume Balarac (LEGI), Simon Mendez (IMAG), Pierre Bénard, Vincent Moureau, Léa Voivenel (CORIA).

[[File:ecfd8.png|600px|link=https://ecfd.coria-cfd.fr/index.php/Ecfd:ecfd_8th_edition]]

[[File:Acknowledgments_ecfd8.png|text-bottom|600px]]

== News ==

* 23/10/2024: First announcement of the '''8th Extreme CFD Workshop & Hackathon''' !
* 22/11/2024: Deadline to submit your project

== Thematics / Mini-workshops ==

These mini-workshops may change and cover more or less topics. This page will be adapted according to your feedback.

To come...

== Projects ==

=== Hackathon GENCI - P. Begou, LEGI ===
This ECFD8 GENCI Hackathon was a rich event, involving 4 differents CFD codes (AVBP, ParaDIGM, SONICS and YALES2) using various paradigms (C++/cuda/hip, Fortran/OpenMP/OpenACC) with several SDKs (AMD, Cray/HPE, Nvidia, Gnu) on a large range of GPU architectures (Nvidia A100, GH100, AMD instinct Mi210, Mi250, Mi300). This two-week event benefited from a high level support from three HPC mentors, two on-site from AMD (J. Noudohouenou and A. Tsetoglou) and one remote from CINES (M. Boudaoud).

==== H1 - ParaDIGM and SONICS on GPU, B. Maugars, G. Staffelbach, R.Cazalbou and B. Michel (ONERA)====

==== H2 - AVBP GPU offloading based on OpenMP, M.Ghenai, L. Legaux and A. Dauptain (CERFACS) ====

This hackathon provided a valuable opportunity to work on GPU offloading for AVBP. In the past, significant efforts were made to offload the entire AVBP code to GPUs. OpenACC was the primary strategy chosen, mainly due to access to NVIDIA's support, along with the availability of both software and hardware. This approach achieved good scalability performance.
Recently, with the deployment of new supercomputers like ADASTRA at CINES, some issues have emerged when running AVBP on AMD GPUs, including both MI250 and MI300. The closed-source nature of the Cray environment has also prevented CERFACS from deploying AVBP on local MI210 GPUs.
This hackathon was a great opportunity to address these challenges by exploring a new approach using OpenMP. An automatic translation tool was used to convert approximately 2,700 OpenACC directives to OpenMP, with each directive manually verified and fine-tuned afterward. AVBP with OpenMP had already been tested on NVIDIA GPUs, and during this hackathon, the focus was on extending support to AMD GPUs.
Two compilers were used: Cray and the newly released AFAR from AMD. With the support of AMD and CINES, a working environment for compiling AVBP was set up, and performance-related issues were identified. Additionally, two mini-apps were used for benchmarking. One of them achieved a 2.5× speedup when compiled with AFAR compared to Cray.
The next steps involve adapting the code to address necessary modifications, such as fixing issues related to Fortran indirections, and continuing performance evaluations with mini-apps. Further comparisons will be conducted using both compilers against results obtained with NVIDIA’s NVHPC.

==== H3 - YALES2 GPU from OpenACC to OpenMP, P. Bégou (LEGI), V. Moureau, G. Lartigue (CORIA) and R. Dubois (IMAG) ====
This Hackathon focuses on running Yales2 code on AMD Instinct Mi250 and Mi300 GPUs of the Adastra supercomputer (CINES).
Previously, a first solver in the Yales2 CFD code was successfully ported on the GPU accelerators of the Jean-Zay supercomputer (IDRIS) using Nvidia SDK but difficulties remain on Adastra AMD GPUs, mainly related to the available development tools. High compilation time and the impossibility to use debug flags at compile time as soon as OpenACC is enabled are a real challenge when tracking errors. The current project is to evaluate a freshly deployed version (at the begining of the workshop) of the AMD Fortran compiler. This requires moving to OpenMP paradigm, starting from scratch since the OpenACC branch has largely diverged from the master one while tracking spurious remaining bugs.
If the AMD compiler is able to build the cpu version of Yales2 "out of the box" (wich is not the case for Cray Fortran), the compilation time for each file is significantly higher. However, setting up a 2 stages dynamic compilation process allows for high parallelism that is not possible with Cray Fortran 18 and the library build time drops from nearly 2 hours (Cray Fortran 18) to 17 minutes (Amd Fortran compiler).
Large kernels have been ported from OpenACC to OpenMP, raising some difficulties when offloading intrinsics functions or using strutures attributes in kernels loops. These limitations were also known in the previous OpenACC work. The goal was mainly to check the correctness of the results. The offloading of the complex data structure of Yales2 code was then investigated. Here again some limitations of the "young" compiler were discovered and workarounds were implemented. Several reproducers were built during this ECFD8 and provided to developpers by the 2 on-site AMD engineers.
Preliminary tests on micro-applications show good performances of the generated binaries proving that this compiler could be a serious alternative on AMD GPUs and the goal is now to focus on this SDK in an OpenMP strategy while checking the portablility of this new implementation in Nvidia, Cray/HPE (and Gnu ?) environments.

=== Mesh adaptation - A. Grenouilloux, ONERA & G. Balarac, LEGI ===

=== Numerics - M. Bernard, LEGI & G. Lartigue, CORIA ===

==== N1 - Traction open boundary condition ====

==== N2 - Treatment of Inlet conditions in High-Order solver. M. Bernard (LEGI), Ghislain Lartigue (CORIA), Guillaume Balarac (LEGI) ====
In the context of node-centered Finite Volumes Method, spacial accuracy of a numerical scheme depends on ability to evaluate accurately fluxes through interface of each control volume (CV). Such accurate evaluation is not straightforward, especially when dealing with distorted grids. This project follows the work of [1] where fluxes use pointwise quantities, which are reconstructed from integrated quantities advanced in time. During the previous edition of the ECFD, a new data structure has been developed to store data at location of the boundary conditions facelets, with application to wall boundary conditions. During this 8th edition of the ECFD, we used the same data structure, but dedicated to the treatment of inlet conditions.
The inlet condition is then either imposed directly at facelets center, or at nodes position them extrapolated to facelets center by use of Taylor expansion. For this later solution, high-order treatment requires the successive derivatives to be computed in the plane of the boundary condition. This is not done yet, leading for the moment to low accuracy results but the framework is ready for upcoming implementation.

[1] ''A framework to perform high-order deconvolution for finite-volume method on simplicial meshes, , Bernard et. al., IJNMF 2020''

==== N3 - Conservative mesh-to-mesh interpolation. M. Bernard (LEGI), Ghislain Lartigue (CORIA), Guillaume Balarac (LEGI) ====

Mesh to mesh interpolations occur quite often in CFD simulations : in the context of adaptative mesh convergence or in the case of dynamic mesh adaptation for for example.
Quality of the solution on the destination grid will depend on the characteristics of the interpolation method.
In this project, we did not focus on accuracy of the interpolation method but rather on conservativity characteristics.
A conservative interpolation ensures that the integral of the data on the source grid is exactly retrieved on the destination grid.
This property is highly interesting when dealing with scalar quantities or phase indicators, whose values should remained bounded.
In the context of nodes centered Finite Volume schemes, the methodology we used consists in (i) reconstructing element quantity from average nodal quantities on source grid.
Then, for a cell of the destination mesh, (ii) computing the geometrical intersection between cells of source and destination meshes to evaluate to evaluate the rate of quantities they.
Eventually, (iii) redistributing the solution from elements to control volumes of the destination mesh.
The overall process is fully conservative as it is based on geometrical intersection of locally integrated quantities.
The methodology as been implemented and tested on a few basic configurations and the conservativity is retrieved.

==== N4 - Determination of timestep in semi-implicit solver. T. Berthelon (LEGI), G. Balarac (LEGI), H. Lam (LEGI), M. El Moatamid (CORIA) ====
In order to reduce the computation time associated with incompressible LES simulations, an implicit time integration, based on BDF schemes, has been developed within the ICS solver. This integration eliminates the stability constraints associated with explicit schemes, and therefore opens up the question of the appropriate choice of time step.
In parallel, recent work has been carried out on meshing criteria in LES. The strategy in place consists of adapting the mesh by distinguishing two zones:
- "DNS" zones, where the meshing criterion is based on an estimate of the adimensioned spatial error.
- "LES" zones, where the meshing criterion is based on Kolmorogov theory.
During this project, the spatial criteria were extended to include temporal criteria. In the "DNS" zones, the time step is chosen using an estimate of the temporal error of the BDF scheme judiciously scaled to match the spatial error. In the "LES" zones, the time step is chosen using a scaling law associated with fully developed turbulence.
The new time step selection strategy has been tested on the case of a turbulent jet and leads to an accuracy equivalent to the explicit case while reducing the simulation return time by a factor of nearly 3.

Another aspect of this project was to integrate certain implicit temporal schemes (C-N and SDIRK) recently developed by Mr. El Moatamid into the incompressible solver.

==== N5 - Local timestep. T. Berthelon (LEGI), M. Bernard (LEGI), G. Balarac (LEGI) ====
RANS modelling has recently been developed within the YALES2 library. With this modeling strategy, the objective is to reach as quick as possible a steady state.
During this project, we investigate the use of a local time step to reduce the time to solution of steady computation in the incompressible solver.
This implies solving a variable-coefficient Poisson equation. Encouraging results were obtained in the simple case of "Couette plan" flow artificially constrained by a mesh variation. In fact, the use of local time-step reduce drastically the time to solution on this configuration. This method needs to be tested on real RANS case.

==== N6 - Distributed version of DOROTHY ====

==== N7 - Implicit time advancement for low-Reynolds number flows with particles. S. Mendez, C. Raveleau (IMAG), M. El Moatamid, V. Moureau (CORIA) ====
IMAG runs numerous simulations of red blood cells under flow. Those simulations are at low Reynolds number (0.001 to 1.0, typically). Splitting of the time advancement is used to treat the diffusion terms implicitly, albeit with an important numerical cost: implicit diffusion is 50 to 60% of the computational cost. Recently, M. El Moatamid implemented a genral framework to deal with implicit time advancement for scalars. In this project, the general method has been transposed to the advancement of the velocity field in the ICS and RBC solvers of YALES2/YALES2BIO. This enables testing various linear solvers (GMRES based). However, such solvers do not decrease the CPU time compared to the existing implementation. However, while working on this, it was identified that residual recycling was not activated in the current implementation of the implicit diffusion. This sped up the implicit diffusion cost by 35%, for a total gain of 20% for the computation. In addition to this achievement, moving to the framework coded by Moncef will have other beneficial side effects: we anticipate simplifying the implementation, with an easier merging between YALES2BIO and YALES2. The method will also be implemented in the electrosatic solver, for which the Poisson problem should benefit from the new GMRES-based solvers. In addition, this project highlights the importance of improving the treatment of stiff source terms in the red blood cells simulations, to be able to overcome the current limitation in time step due to those term and have a chance to benefit from higher-order time schemes, efficient at high Fourier numbers.

==== N8 - Boundary Element Method in YALES2. B. Thibaud, S. Mendez (IMAG), G. Lartigue, P. Benard (CORIA), F. Nicoud (IMAG) ====
In the context of microfluidic systems for diagnosis, the Boundary Element Method alows to solve linear PDE such as electrostatic or Stokes. With well chosen kernel functions and the divergence theorem, this method allows to write on the boundary condition only the initial volumic problem. This project aimed at exploring the feasibility of the BEM in the context of massively parallel unstructured solver like YALES2 by developping a Julia demonstrator. The first step have been to implement and validate the method on simple configurations for the Laplace's equation. Only Neumann problems were considered (Dirichlet boundary conditions imposed). In a second time, the multi-domain approach has been identified to be the most suited in the framework of YALES. The inner domain is partitioned on each processor, each having a part of the physical boundary and interfaces between them. Every processor solve its own boundary problem and a parallel Dirichlet-Dirichlet fixed-point is used to converge the interface problem on the all domain. Applied to the ring case, with one interface, we managed to reproduce the linear convergence of the P-DD method.

=== Turbulence - L. Voivenel, CORIA & P. Bénard, CORIA ===

==== T1 - FSI-1D strategy for internal flows====

Many applications developed at Safran Aerosystem are based on internal turbulent flows coupled to a moving body. 2 cases were studied during this ECFD:

'''Case 1 (Incompressible flow)''': Translation of a piston subjected to a pressure difference in a pipe.

The challenges of this case are twofold: the small gap between the piston and the pipe and the large pressure gradient across the piston (>100bar). During the 1st week of ECFD, the CLIB (Conservative Lagrangian Immersed Boundary) solver was tested on this case. The study showed that the solver was unable to ensure the impermeability of the solid under these pressure conditions. In the rest of the study, a porous medium following Darcy's law will be added to the penalty force of the immersed solid to fully satisfy the impermeability of the piston.

'''Case 2 (Compressible flow)''': Rotation of a butterfly in a discharge vane.

The coupling between the ECS (Explicit Compressible Solver) and ALE (Arbitrary Lagrangian Solver) solvers having recently been implemented, this strategy was tested to model the opening of the valve by rotation of the butterfly. The challenge here lies in the small gap between the bottom of the butterfly and the vane casing. To limit the simulation cost, the gap is meshed with 1 element. In this case, MMG succeeded in adapting the mesh up to a critical angle at which the gap becomes too small (Work In Progress).

==== T2 - Dynamic Smagorinsky in Dorothy ====

==== T3 - Turbulence injection strategy for compressible flows ====

==== T4 - Improve wind farm modeling and simulation workflow ====
The YALES2 library includes an advanced modular implementation of the Actuator Line Method (ALM). This approach remains state-of-the-art when performing an LES-based analysis of a wind turbine wake. The method also provides an accurate assessment of the aerodynamic loads applied on the turbine. Still, applying this method to investigate a wind farm flow can be challenging, both in terms of computational cost and simulation setup. For instance, an inadequate management of the wind turbine individual modeling parts in a HPC context can end up being the main bottleneck of the simulation. From another perspective, a wind farm is usually composed of more than 50 wind turbines. For such a case, setting up all YALES2 required inputs manually can be very tedious and error-prone. This project thus mainly aimed to optimize the YALES2 ALM implementation and the user experience around it. Additionally, a cost-effective alternative to the ALM when modeling wind farm flows, namely the Rotating Actuator Disk Method (ADM-R), has been implemented for further investigations.

'''WP1''': Improve Actuator set rotor modelling
* Parallel processing of the ''actuator sets'' used to model the wind turbines
The implementation of the Actuator Line Method (ALM) into the YALES2 library leads to poor performances when many wind turbine rotors are set. Indeed, each rotor object is a derived type treated sequentially by all the processors participating to the computation. With 30 turbines in a computation, the return time is increased by 70% while the arithmetic intensity appears to be low. The objective of this sub-project is to improve the computation performances of the ALM already identified during the 5th iteration of the ECFD: (i) Assign one MPI communicator by rotor object gathering the processors close to the turbine and set-up a master/slave processus by communicator. This will allow the simultaneous rotors computation and reduce the number of MPI exchanges. (ii) Work on the domain decomposition to limit the number of processors attributed to each turbine. This would reduce or even eliminate MPI communications. During this workshop edition, the work focused on the re-synchronisation of the algorithm steps by allowing some packing an unpacking of the object to allow the transfert of the object inbetween workers. This is of major importance to enable load balancing and mesh adaptation during the temporal loop. This work required the refactoring of the involved oject structures.

* Rotating Actuator Disk Method (ADM-R):
According to the usual guidelines, the mesh requirements of the ALM, to profit entirely from its reachable accuracy, can be difficult to achieve or even unaffordable when simulating a wind farm flow, especially from the industrial point of view. Alternatives are available in the literature for this kind of application. Likely, the methods from the Actuator Disk family are the most prominent ones. Several kinds of implementation exist, which mostly differ by their capability to include the wake rotation. During the workshop, a new method from the Rotating Actuator Disk kind has been implemented and underwent an early validation on a single turbine setup. Applications to wind farm flows will follow.

'''WP2''': Improve tools User Experience

Three Python tools have been developed or improved :
*The first tool is the wind farm previsualisation tool, 'y2_wind_previsualization', which is used before the calculation run. This provides an interactive HTML interface for viewing global data for each turbine on the farm (position, hub height, yaw angle, etc.). The tool traces all of these via the parsing of the input file.
* The second tool is for duplicating rotor templates for a wind farm (`y2_wind_duplication`). This tool was developed in the previous ECFD, but this time it has been refactored and incorporated into the y2tools package.
* The third and final tool is a post-processing tool for the temporal processing of global wind turbine simulation metrics (Thrust, Power, etc.), `y2_post_wind`. This tool generates an interactive HTML plot of time-dependent global quantities.

==== T5 - Improve atmospheric inflow turbulence ====
Atmospheric inflow turbulence is generated using the precursor database method. A half-channel flow driven by a pressure gradient is used to obtain the inflow which is used as inlet boundary condition for the wind turbine simulation domain. This project aimed to improve the whole methodology, from generation to injection.

* WP1: Improve inflow generation
Anand: pressure controller

* WP2: Improve injection methodology (method A)
The previous workflow used plane probes in the ASCII format to sample the flow. The COWIT2 toolbox was used to convert the file into turbulence box (.man format). While functioning, this methodology had two major flaws. First the probe files are heavy ~O(10Go). Second, the method requires a lot of human effort, allowing numerous sources of errors.
During this workshop, a new methodology has been developed. First, the probes are generated using the HDF5 format (now available for all probe types), leading to lighter file ~O(1Go). Second, Y2_tools is used to read HDF5 format (working for probes and temporals). HDF5 file is then converted into a Look-up Table. Finally, the Look-up Table is read directly by YALES2 as a boundary conditions.

* WP3: Improve injection methodology (method B)
Even though improvements achieved in WP2 prove to be very handy while removing many potential human errors, injecting a turbulent inflow through wind boxes ('offline' precursor approach) can sometimes remain cumbersome for several reasons: (1) no periodicity is enforced in the streamwise direction of those boxes, (2) potential high memory consumption, and (3) the boxes need to be moved to other cores whenever a mesh adaptation occurs. An alternative consists in co-simulating the precursor flow and the flow of interest (refered as the 'successor' simulation) at the same time ('online' precursor approach). The inlet boundary condition for the successor flow is then obtained by mapping the outflow of the precursor domain. During the workshop, some work has been initiated to implement this kind of coupling using the CWIPI library, for which YALES2 provides already an interface.

==== T6 - FSI model in Dorothy ====

=== Two Phase Flow - J. Leparoux, SAFRAN & J. Carmona, CORIA ===

==== TP1 - Towards very small contact angles in Nucleate boiling ====

Participants: Henri Lam (LEGI), Mohammad Umair (LEGI), Manuel Bernard (LEGI), Robin Barbera (LEGI) and Giovanni Ghigliotti (LPSC)

The boiling solver (BOI) was not able to accurately impose a contact angle (angle formed by the two-phase interface on the wall) at values lower than 30°. This angle is needed when simulating nucleate boiling. A similar limitation in contact angle value was applying to the spray (SPS) solver. A modified version of the level set reinitialization has been implemented during ECFD8, based on a different normal vector in the blind spot region around the contact line, vector now chosen to be a zero-order extension from outside the blind spot. This modification, that implied other modifications to the level set reinitialization in the blind spot, has been tested successfully on the spray solver (where no phase change occurs). Then, this new reinitialisation has been tested in the boiling solver for nucleate boiling, with great improvements. Now simulations of nucleate boiling at very small contact angle (10°) can be accurately performed.
In the meanwhile, the level set reinitialization algorithm has been streamlined and the computational cost greatly reduced, resolving a computational cost issue that appeared when using the contact angle imposition both in the spray and boiling solvers, and that hampered its use in industrial configurations.

==== TP2 - Modeling spray-film interactions ====

Participants: Nicolas Gasnier (EM2C-SafranTech), Julien Leparoux (SafranTech), Mehdi Helal (CORIA-SafranTech) and Julien Carmona (CORIA)

The numerical simulation of sheared liquid films over walls with conventional high-fidelity methods requires fine meshes to depict accurately the dynamics of the fluids, as well as the gas-liquid interactions at the phase interface. In the context of thin film flows (with a thickness h ~ 1.0E-4 m), the spatial resolution required to ensure the validity of the simulations can be computationally prohibitive. To tackle this issue, a reduced-order model of thin film dynamics based on the Saint Venant equations has been implemented in the YALES2 platform over the last months. This numerical model is able to reproduce accurately the dynamics of strongly sheared film flows over partially wetted surfaces, and to take into account capillary effects. The objective of the ECFD8 was to couple the thin film framework with the pre-existing multiphase methods of the YALES2 platform: the Eulerian multiphase solver - based on the ACLS -, and the Lagrangian solver. This coupling was performed in three steps: first, a numerical method was designed to convert impinging Lagrangian droplets into source terms for the thin film model. Then, a phenomenological model was implemented to depict the atomization of films at sharp edges under the action of a high speed gas flow. The liquid film is converted into Lagragian droplets, whose dimensions are computed based on the PAMELA model. Finally, a numerical method has been designed to convert Eulerian droplets into source terms for the film model. This method is based on a transfer of the liquid properties from the nodes to the boundaries by using a set of fictitious particles. It ensures conservation of the liquid mass, and it is more robust than the first attempts that had been made during ECFD7.

==== TP3 - High-fidelity two-phase flow simulations of the purge of a fuel feed line ====

Participants: Thomas LAROCHE (Safran HE), Romain JANODET (Safran AE), Julien Leparoux (Safran Tech) and Melody Cailler (Safran Tech)

During the second week of the ECFD8, the fuel feed line purge process has been numerically investigated. In the context of aeronautical engines, the fuel feed line—carrying the fuel from the fuel tank to the injectors within the combustion chamber—needs to be purged at engine shutdown. This is intended to prevent fuel stagnation near hot metal parts, which could lead to coke formation and therefore decrease engine performance. Since this complex phenomenon is mainly driven by two-phase flow physics, the spray solver (SPS) of the YALES2 library has been considered in order to understand the physics of such process. The numerical setup was first converged on a simplified test case: the possibility of driving the flow dynamics with inlet and outlet pressure conditions was tested beforehand on a single-phase, incompressible case, and then on a two-phase flow problem. The setup has then been successfully applied to an industrial configuration: a pressure-swirl injector connected to a reduced portion of the fuel feed line. Due to the large scale of the domain, the interface resolution was set to 50μm, which is intentionally coarse for such problem. This initial computation successfully ran up to 3ms of physical time during the workshop, proving YALES2's capability to model the fuel purge. The computation is to be continued and analyzed further even after the workshop.

==== TP4 - Volume of Fluid solver in YALES2 ====

Participants: Léa Voivenel (CORIA), Julien Carmona (CORIA), Mehdi Helal (CORIA), Pierre Portais (CORIA), Julien Leparoux (Safran Tech), Mélody Cailler (Safran Tech) and Nicolas Gasnier (EM2C / Safran Tech)

==== TP5 - Implement a local operator to distribute the solid volume of a particle over multiple cells ====

Participants: Théo Ndereyimana (Université de Sherbrooke), Stéphane Moreau (Université de Sherbrooke)

In the CFD-DEM, the cell size is required to be larger than the particle size for stability condition and keep feasible solid volume fraction. However, some applications require a cell size smaller than the particle. During this 8th edition of the ECFD, the use of operators to distribute the particle volume over multiple cells, ensuring a feasible solid volume particle has been tested on a fluidized bed configuration. The main operators tested (gather-scatter filter and gaussian filter) showed a tendency to blur the void structure interfaces. The equivalence of the gaussian filter of bandwidth <math>b</math> and the resolution of a diffusion equation over a pseudo-time <math>T=b^2/4</math> has been verified.
One anisotropic diffusion constant has been tested and shows a possibility to adress the sharpness requirements.

Another objective was to develop a post-processing tool to detect and track the void structures (bubbles) in a fluidized bed. Based on previous work from J. Carmona, a tool to track the bubbles has been initiated.

==== TP6 - Complex thermodynamics in sloshing tanks ====

Participants: C. Merlin (AGS), D. Fouquet (CORIA), V. Moureau (CORIA), J. Carmona (CORIA) and G. Lartigue (CORIA)

During this ECFD, the phase change solver was improved. After introduction of conservative transport for enthalpy and energy, a new framework for multi fluid two phase flow was used that relies on a new reinitialization of the conservative level set, the transport of discontinuous scalars like temperature, energy or enthalpy and the transport of weight to ensure consistency. The pressure loop was improved to ensure mass and energy conservation while meeting the low Mach hypothesis. Many 1D test cases were performed with only the gas phase and two phase flow to validate the pressure loop. The new framework for the transport of discontinuous scalar was first derived for temperature with constant properties in each phase. A first step was to extend it to sensible enthalpy with ideal gas behavior for the gas phase. Then, the framework was extended to NIST tabulated gas & liquid with ideal or real gas behavior. After some 2D cases, a tank pressurization test was investigated as well as a sloshing test without phase change (just the pressure drop due to the increase of the thermal exchange between the liquid and the gas phase).

=== Combustion - Y. Bechane, CORIA & S. Dillon, SAFRAN & K. Bioche, CORIA ===

==== C1 - LES of the thermal degradation of a composite material ====
Participants: A. Grenouilloux (ONERA), K. Bioche (CORIA), N. Dellinger (ONERA) and R. Letournel (SafranTech)

The FIRE test bed is an experimental air-propane burner operated by ONERA. It is dedicated to the study of the thermal degradation of composite materials. This project concerned the implementation of a three-solver coupling methodology to simulate the dynamics of the impinging flame. The methodology considered is based on the coupling between the variable density solver (VDS) and the radiative solver (RDS) of the massively parallel library YALES2 and the solver dedicated to the degradation of composite materials, MoDeTheC, developed by ONERA. Given the typical test times of the order of tens of seconds, a methodology based on 2D axisymmetric calculations was considered. Various tests were performed to determine the optimal coupling frequency between solvers. Cases dedicated to the injection of pyrolysis gasses were set up, with the aim of simulating the auto-ignition phenomenon. Comparisons with experimental data are presented.

==== C2 - Flame stabilization by NRP plasma discharge ====

==== C3 - Extending and validating a generalized formalism of virtual chemistry ====

==== C4 - Turbulent combustion model for NOx prediction ====

==== C5 - Towards 3D simulation of detonation combustion ====

==== C6 - Flame stabilitity of flame-holders within reheat conditions ====

Participants : N. Detomaso (Safran AE), R. Janodet (Safran AE), L. Carbajal (Safran AE)

The mechanisms of flame stabilisation in a flame-holder are still not completely understood. This is particularly true for reheat conditions : the hot viciated gases at inlet, highly compressible flows, and a strong liquid/gas coupling make the flame stability hard to predict. During the ECFD8, simplified configurations of flame-holders were analysed. This step marks the beginning of a systematic mapping of dimensionless characteristic times ratios.

==== C7 - Thermal radiation in oxyflames ====

==== C8 - A first step toward hybrid CPU / GPU for reactive flow in YALES2 ====

Participants: M. Laignel (CORIA), G. Lartigue (CORIA), K. Bioche (CORIA) and V. Moureau (CORIA)

In numerical simulations of reacting flows, one of the most computationally intensive tasks is the evaluation of source terms resulting from chemical reactions in the species transport equations. This step can account for up to 90% of the total simulation cost , depending on the complexity of the kinetic mechanism involved. To reduce this cost, various techniques such as mechanism reduction, virtual chemistry, etc. have been explored. However, the emergence of GPUs as powerful accelerators offers a promising alternative by providing massive parallelism. Despite their potential, GPUs often require significant adaptation of CPU-based codes. This project aims to address this challenge by taking a first step towards a hybrid CPU/GPU framework for reactive flow simulations. Specifically, the focus is on coupling Y2 with the updated version of the stiff time integration solver (CVODE), which is compatible with GPU (CUDA, HIP, OpenMP). The ultimate goal is to establish a foundation for hybrid computations by implementing and testing the updated solver on simplified test cases.

==== C9 - Soots numerical modeling ====

==== C10 - TECERACT : Tabulated chemistry generator for aeronautical combustion ====

==== C11 - Exploring efficient tabulation strategies for detailed chemistry ====

==== C12 - Dynamic sub-grid-scale modelling of multi-regime flame wrinkling ====

==== C13 - LES of a semi-industrial burner using a non-adiabatic virtual chemical scheme ====

=== User Experience & Data - L. Korzeczek, GDTECH ===

==== U1 - Low-fidelity (RANS) rotor/stator simulations, application to Kaplan Turbine - Y. Lakrifi, G. Balarac (LEGI), R. Mercier (SAFRAN), V. Moureau (CORIA) ====

==== U2 - Coupling PyTorch/YALES2, combustion cartesian look-up tables - J. Leparoux, N. Treleaven, S. Dillon (SAFRAN), K. Bioche, G. Lartigue (CORIA) ====

Participants: Julien Leparoux (Safran Tech), Kévin Bioche (CORIA), Ghislain Lartigue (CORIA), Nicholas Treleaven (Safran Tech)

Neural Networks offer a promising alternative to Cartesian look-up tables for combustion simulations, reducing memory footprint. In this project, we investigated how to integrate an NN model for real-time inference in the YALES2 platform, exploring two approaches: a Python interface and a Fortran Torch binding (using FTorch[https://github.com/Cambridge-ICCS/FTorch]). We validated that the model remains accurate when embedded online and identified improvements for robustness. Inference costs were evaluated on a Mac M3 and the Austral cluster, revealing a strong dependency on data volume. To optimize efficiency, we propose grouping cells at the processor level.

==== U3 - Yales2 Trame Editor, toward a fully featured graphical user interface for YALES2 - L. Korzeczek, S. Meynet (GDTECH), J. Leparoux, M. Cailler (SAFRAN) ====

Ecfd:ecfd 7th edition

2024-02-05T09:40:19Z

Luis.carbajal-carrasco: /* M1: ASMR for reheat chamber applications - Paul Pouech (CERFACS), Thibault Duranton & Luis Carbajal Carrasco (Safran) */

{{DISPLAYTITLE: ECFD workshop, 7th edition, 2024}}

== Description ==

* Event from '''22th of January to 2nd of February 2024'''
* Location: [https://www.hotelclubdelaplage.com Hôtel Club de la Plage], Merville-Franceville, near Caen (14)
* Two types of sessions:
** common technical presentations: roadmaps, specific points
** mini-workshops. Potential workshops are listed below
* Free of charge
* More than 70 participants from academics, HPC center/experts and industry.

* Objectives
** Bring together experts in high-performance computing, applied mathematics and multi-physics CFDs
** Identify the technological barriers of exaflopic CFD via numerical experiments
** Identify industrial needs and challenges in high-performance computing
** Propose action plans to add to the development roadmaps of the CFD codes

[[File:ecfd7.png|600px|link=https://ecfd.coria-cfd.fr/index.php/Ecfd:ecfd_6th_edition]]

[[File:sponsor_ecfd7.png|text-bottom|600px]]



== Agenda ==

[[File:agenda_ecfd7.png|text-bottom|600px]]

== Thematics / Mini-workshops ==

These mini-workshops may change and cover more or less topics. This page will be adapted according to your feedback.

To come...

== Projects ==

=== Hackathon GENCI - P. Begou, LEGI ===
The '''GENCI Hackathon''' will be devoted to porting two CFD codes to the Mi250 GPUs of the Adastra supercomputer deployed by GENCI at CINES.

For the '''YALES2''' code the goal is to obtain a first reference version giving the expected results then, if possible, to start its optimization to gain performance. The approach is OpenACC based with the objective of an implementation as least intrusive as possible in the existing code and which remains portable with the work done on the Nvidia GPUs of the Jean-Zay supercomputer at IDRIS.

The porting of the '''AVBP''' code is more advanced with a prototype already functional on Adastra but "hard-coded". The objective is to rationalize this first implementation, to integrate the latest developments in the code, to centralize memory management (host and device), to work on porting the Lagrangian part of the code and, of course, to improve the global performance.

This Hackathon is supported by GENCI, HPE, AMD and CINES with the presence on site of several development experts on AMD GPUS.

=== Mesh adaptation - R. Letournel, Safran ===

==== M1: ASMR for reheat chamber applications - Paul Pouech (CERFACS), Thibault Duranton & Luis Carbajal Carrasco (Safran) ====

Combustion in reheat chambers feature a wide range of lenght scales. Mesh refinement is thus mandatory to capture the flow characteristics within a reasonnable CPU cost for LES computations using the AVBP code. The purpose of this project is to consolidate mesh refinement criteria and strategy in an academic reference case. The retained workflow is supported by the [https://lemmings.readthedocs.io/en/latest/readme_copy.html Lemmings] code that calls the Tékigô wrapper for the mesh adaptations. During the ECFD7, the convergence time needed to have significant distribution of quantities of interest was analysed. An optimum runtime, based on a characteristic flow time-scale, was thus identified and led to a reduced running time for each adaptation step. As a second step, discussions with the ECFD7 participants led to the identification of interesting refinement criteria, namely the flame sensor or the mach rms for instance. Parametric analysis showed the robustness of the workflow based on a ponderation of different criteria. Finally, in order to facilitate the use of the workflow, efforts were made to improve the user experience by making it more human readable.

=== Numerics - S. Mendez, IMAG & G. Balarac, LEGI ===

==== WIP N1: High-Order schemes for Navier-Stokes - M. Bernard & G. Balarac (LEGI) & G. Lartigue (Total Eneriges) ====

Finite Volumes method.
Distinction between integrate and pointwise quantities.
Spacial accuracy depend on ability to evaluate accurately fluxes, starting from integrated quantities.
Accurate evaluation of fluxes through interface of each control volume (CV).
OK inside the domain, but chalenging in case of boundary conditions : by definition non uniformity of the stencil.
BC on faces instead of nodes.
Enforcement of the BC by modifying the normal component of the wall gradient in order to evaluate accurate diffusif flux.
Treatment of inlet BC by integrating convective flux through each boundary-facelet.

==== N5: Optimization of the RBC solver - F. Rojas & S. Mendez (IMAG) ====

==== N6: Electrodeformation of red blood cells, extension to 3D and improved accuracy at membrane - A. Spadotto & S. Mendez (IMAG), M. Bernard (LEGI) ====
The Leaky Dielectric Model is a popular framework to describe electric stresses over micro-scale membranes. We have adopted it to simulate the effect of a DC electric field on a red blood cell using the YALES2BIO solver. The goal of the project is to reproduce the electric charging process of the membrane, as well as the resulting stresses, which may yield to electrodeformation of the cell. From the point of view of the implementation, the grid is represented by a 2D surface mesh embedded in a 3D eulerian grid. The need to make variables stored on the surface interact with quantities stored on the Eulerian grid calls for a proper bidirectional 2D-membrane/3D-grid dynamic connectivity. The advancement of theis task during this ECFD has led to the first 3D simulation of a charging fixed spherical shell. Moreover, the estimation of grid variables on elements cut by the membrane has been improved thanks to a High-Order extrapolation. The latter has been successfully tested on 2D configurations. The project opens the way for a series of validation tests. In particular, future work will demand treatment of instabilities emerging in symmetrical configurations.

=== Turbulence - P. Benard, CORIA & L. Bricteux, UMONS ===

=== T4: Atmospheric solver ===
Wind turbines, bigger and bigger, are now influenced by atmospheric flows. An atmospheric solver has already been developed in YALES2 to represents some of its effects (Coriolis, veer, thermal stratification). In this continuum, the project has been divided into two work-packages.
- Work-package 1: The use of the Variable density solver (VDS).
Before ECFD7, thermal stratification was taken into account using the Boussinesq buoyancy approximation within the incompressible solver framework. Now, VDS can be used, taking into account all thermal effect. Results are promissing.
- Work-package 2: Wall law velocity filtering.
Wall law are using velocity at the first grid node to compute wall shear stress. Before ECFD7, atmospheric wall law were using the local velocity, leading sometimes to convergence errors. Now a gather-scatter filter can be used to average velocity (and temperature) at first grid node.

=== Two Phase Flow - M. Cailler, Safran Tech & V. Moureau, CORIA ===

=== Combustion - K. Bioche, CORIA & R. Mercier, Safran ===

=== User Experience & Data - L. Korzeczek, GDTech ===

Ecfd:ecfd 7th edition

2024-02-05T08:38:03Z

Luis.carbajal-carrasco: /* Mesh adaptation - R. Letournel, Safran */

{{DISPLAYTITLE: ECFD workshop, 7th edition, 2024}}

== Description ==

* Event from '''22th of January to 2nd of February 2024'''
* Location: [https://www.hotelclubdelaplage.com Hôtel Club de la Plage], Merville-Franceville, near Caen (14)
* Two types of sessions:
** common technical presentations: roadmaps, specific points
** mini-workshops. Potential workshops are listed below
* Free of charge
* More than 70 participants from academics, HPC center/experts and industry.

* Objectives
** Bring together experts in high-performance computing, applied mathematics and multi-physics CFDs
** Identify the technological barriers of exaflopic CFD via numerical experiments
** Identify industrial needs and challenges in high-performance computing
** Propose action plans to add to the development roadmaps of the CFD codes

[[File:ecfd7.png|600px|link=https://ecfd.coria-cfd.fr/index.php/Ecfd:ecfd_6th_edition]]

[[File:sponsor_ecfd7.png|text-bottom|600px]]



== Agenda ==

[[File:agenda_ecfd7.png|text-bottom|600px]]

== Thematics / Mini-workshops ==

These mini-workshops may change and cover more or less topics. This page will be adapted according to your feedback.

To come...

== Projects ==

=== Hackathon GENCI - P. Begou, LEGI ===
The '''GENCI Hackathon''' will be devoted to porting two CFD codes to the Mi250 GPUs of the Adastra supercomputer deployed by GENCI at CINES.

For the '''YALES2''' code the goal is to obtain a first reference version giving the expected results then, if possible, to start its optimization to gain performance. The approach is OpenACC based with the objective of an implementation as least intrusive as possible in the existing code and which remains portable with the work done on the Nvidia GPUs of the Jean-Zay supercomputer at IDRIS.

The porting of the '''AVBP''' code is more advanced with a prototype already functional on Adastra but "hard-coded". The objective is to rationalize this first implementation, to integrate the latest developments in the code, to centralize memory management (host and device), to work on porting the Lagrangian part of the code and, of course, to improve the global performance.

This Hackathon is supported by GENCI, HPE, AMD and CINES with the presence on site of several development experts on AMD GPUS.

=== Mesh adaptation - R. Letournel, Safran ===

==== M1: ASMR for reheat chamber applications - Paul Pouech (CERFACS), Thibault Duranton & Luis Carbajal Carrasco (Safran) ====

Combustion in reheat chambers feature a wide range of lenght scales. Mesh refinement is thus mandatory to capture the flow characteristics within a reasonnable CPU cost for LES computations using the AVBP code. The purpose of this project is to consolidate mesh refinement criteria and strategy in an academic reference case. The retained workflow is supported by the [https://lemmings.readthedocs.io/en/latest/readme_copy.html Lemmings] code that calls the Tékigô wrapper for the mesh adaptations. During the ECFD7, the convergence time needed to have significant distribution of quantities of interest was analysed. An optimum runtime, based on a characteristic flow time scale, was thus identified and led to a reduced running time for each adaptation step. As a second step, discussions with the ECFD7 participants led to the identification of interesting refinement criteria, namely the flame sensor or the mach rms for instance. Parametric analysis showed the robustness of the workflow based on a ponderation of different criteria. Finally, in order to facilitate the use of the workflow, efforts were made to improve the user experience by making it more human readable.

=== Numerics - S. Mendez, IMAG & G. Balarac, LEGI ===

==== WIP N1: High-Order schemes for Navier-Stokes - M. Bernard & G. Balarac (LEGI) & G. Lartigue (Total Eneriges) ====

Finite Volumes method.
Distinction between integrate and pointwise quantities.
Spacial accuracy depend on ability to evaluate accurately fluxes, starting from integrated quantities.
Accurate evaluation of fluxes through interface of each control volume (CV).
OK inside the domain, but chalenging in case of boundary conditions : by definition non uniformity of the stencil.
BC on faces instead of nodes.
Enforcement of the BC by modifying the normal component of the wall gradient in order to evaluate accurate diffusif flux.
Treatment of inlet BC by integrating convective flux through each boundary-facelet.

==== N5: Optimization of the RBC solver - F. Rojas & S. Mendez (IMAG) ====

==== N6: Electrodeformation of red blood cells, extension to 3D and improved accuracy at membrane - A. Spadotto & S. Mendez (IMAG), M. Bernard (LEGI) ====
The Leaky Dielectric Model is a popular framework to describe electric stresses over micro-scale membranes. We have adopted it to simulate the effect of a DC electric field on a red blood cell using the YALES2BIO solver. The goal of the project is to reproduce the electric charging process of the membrane, as well as the resulting stresses, which may yield to electrodeformation of the cell. From the point of view of the implementation, the grid is represented by a 2D surface mesh embedded in a 3D eulerian grid. The need to make variables stored on the surface interact with quantities stored on the Eulerian grid calls for a proper bidirectional 2D-membrane/3D-grid dynamic connectivity. The advancement of theis task during this ECFD has led to the first 3D simulation of a charging fixed spherical shell. Moreover, the estimation of grid variables on elements cut by the membrane has been improved thanks to a High-Order extrapolation. The latter has been successfully tested on 2D configurations. The project opens the way for a series of validation tests. In particular, future work will demand treatment of instabilities emerging in symmetrical configurations.

=== Turbulence - P. Benard, CORIA & L. Bricteux, UMONS ===

=== T4: Atmospheric solver ===
Wind turbines, bigger and bigger, are now influenced by atmospheric flows. An atmospheric solver has already been developed in YALES2 to represents some of its effects (Coriolis, veer, thermal stratification). In this continuum, the project has been divided into two work-packages.
- Work-package 1: The use of the Variable density solver (VDS).
Before ECFD7, thermal stratification was taken into account using the Boussinesq buoyancy approximation within the incompressible solver framework. Now, VDS can be used, taking into account all thermal effect. Results are promissing.
- Work-package 2: Wall law velocity filtering.
Wall law are using velocity at the first grid node to compute wall shear stress. Before ECFD7, atmospheric wall law were using the local velocity, leading sometimes to convergence errors. Now a gather-scatter filter can be used to average velocity (and temperature) at first grid node.

=== Two Phase Flow - M. Cailler, Safran Tech & V. Moureau, CORIA ===

=== Combustion - K. Bioche, CORIA & R. Mercier, Safran ===

=== User Experience & Data - L. Korzeczek, GDTech ===