Papers by Philippe Parnaudeau
HAL (Le Centre pour la Communication Scientifique Directe), Sep 28, 2022
HAL (Le Centre pour la Communication Scientifique Directe), Aug 29, 2022
Journal of Computational Physics, 2019
Computation Fluid Dynamics (CFD) simulation has become a routine design tool for i) predicting ac... more Computation Fluid Dynamics (CFD) simulation has become a routine design tool for i) predicting accurately the thermal performances of electronics set ups and devices such as cooling system and ii) optimizing configurations. Although CFD simulations using discretization methods such as finite volume or finite element can be performed at different scales, from component/board levels to larger system, these classical discretization techniques can prove to be too costly and time consuming, especially in the case of optimization purposes where similar systems, with different design parameters have to be solved sequentially. The design parameters can be of geometric nature or related to the boundary conditions. This motivates our interest on model reduction and particularly on reduced basis methods. As is well documented in the literature, the offline/online implementation of the standard RB method (a Galerkin approach within the reduced basis space) requires to modify the original CFD calculation code, which for a commercial one may be problematic even impossible. For this reason, we have proposed in a previous paper, with an application to a simple scalar convection diffusion problem, an alternative non-intrusive reduced basis approach (NIRB) based on a two-grid finite element discretization. Here also the process is two stages: offline, the construction of the reduced basis is performed on a fine mesh; online a new configuration is simulated using a coarse mesh. While such a coarse solution, can be computed quickly enough to be used in a rapid decision process, it is generally not accurate enough for practical use. In order to retrieve accuracy, we first project every such coarse solution into the reduced space, and then further improve them via a rectification technique. The purpose of this paper is to generalize the approach to a CFD configuration.
HAL (Le Centre pour la Communication Scientifique Directe), May 11, 2021
HAL (Le Centre pour la Communication Scientifique Directe), Sep 23, 2019
We study the numerical solution to two-phase problems using the reduced Kapila model, i.e. four-e... more We study the numerical solution to two-phase problems using the reduced Kapila model, i.e. four-equations model. In particular we focus the present study on HLLC numerical flux approximation and its extensions. These numerical methods are tested on expansion tube cases with a final goal of constructing higher order hybrid numerical tool for solving the bubble collapsing under the shock wave problem.
HAL (Le Centre pour la Communication Scientifique Directe), Aug 1, 2016
International audienceOptimizing solvers for linear systems is a major challenge in scientific co... more International audienceOptimizing solvers for linear systems is a major challenge in scientific computation.Indeed computational scientifics would like to have the more accurate result in a shorter time.To achieve this, a well-knowledge of the computer science is necessary, not only in term of programmation paradigms, but also in order to follow the evolution of hardware. With the advent of new computers, we have to rewrite the numerical algorithms or create new ones to fully use the parallel architecture of the computers.Parallel space discretisations of Partial Differential Equations (PDE), like Domain Decomposition Methods (DDM) are efficient and well documented methods. At a first glance, parallelisation seems to be inconsistent with the time evolution which is inherently sequential. In fact, parallelisation is not limited to the space direction. The time direction is also a candidate for parallelisation. In this paper we present a new and simple method for time parallelization. This solver is based on partial fraction decomposition of the inverse of some special matrices. We discuss its application for PDE as well as its limitations. Numerical experiments are given
HAL (Le Centre pour la Communication Scientifique Directe), Dec 14, 2004
Des sillages complexes sont etudies par Simulations Numeriques Directes et aux Grandes Echelles. ... more Des sillages complexes sont etudies par Simulations Numeriques Directes et aux Grandes Echelles. Afin de tenir compte de l'obstacle, une methode de forcage a ete implantee au sein d'un code de calcul qui utilise des methodes numeriques de haute precision sur un maillage cartesien. Pour modeliser le plus fidelement possible l'ecoulement autour d'un cylindre, la methode de forcage a subi des modifications. Ainsi a l'aide d'un ecoulement miroir, defini au sein de l'obstacle, nous avons ameliore cette methode. Les SND autour d'un cylindre mettent en evidence ces ameliorations. En particulier, nous nous sommes attaches a demontrer l'importance de l'utilisation de schemas de discretisations spatiales d'ordre elevees dans le cadre de notre etude. L'etude d'un sillage de cylindre pour un nombre de Reynolds de 3900 a permis de valider notre approche dans le cadre des SGE. Enfin une etude preliminaire d'une couche de melange en interaction avec un sillage est proposee.
HAL (Le Centre pour la Communication Scientifique Directe), Sep 14, 2007
HAL (Le Centre pour la Communication Scientifique Directe), Jun 10, 2019
Computer Physics Communications, 2021
This paper is concerned with the numerical investigation of Quantum Turbulence (QT) described by ... more This paper is concerned with the numerical investigation of Quantum Turbulence (QT) described by the Gross-Pitaevskii (GP) equation. Numerical simulations are performed using a parallel (MPI-OpenMP) code based on a pseudo-spectral spatial discretization and second order splitting for the time integration. We start by revisiting (in the framework of high-performance/high-accuracy computations) well-known GP-QT settings, based on the analogy with classical vortical flows: Taylor-Green (TG) vortices and Arnold-Beltrami-Childress (ABC) flow. Two new settings are suggested to build the initial condition for the QT simulation. They are based on the direct manipulation of the wave function by generating a smoothed random phase (SRP) field, or seeding random vortex rings (RVR) pairs. The new initial conditions have the advantage to be simpler to implement than the TG and ABC approaches, while generating statistically equivalent QT fields. Each of these four GP-QT settings is described in detail by defining corresponding benchmarks that could be used to validate/calibrate new GP codes. We offer a comprehensive description of the numerical and physical parameters of each benchmark. We analyze the results in detail and present values, spectra and structure functions of main quantities of interest (energy, helicity, etc.) that are useful to describe the turbulent flow. Some general features of QT are identified, despite the variety of initial states.
HAL (Le Centre pour la Communication Scientifique Directe), Jun 11, 2018
The investigation of the pressure developed by a collapsing cavitation bubble is of primary inter... more The investigation of the pressure developed by a collapsing cavitation bubble is of primary interest for hydraulic turbomachineries, fuel injectors, naval propulsion systems and biomedical technologies. One of the most critical consequences is the shock wave emitted by bubbles collapse and the structural damage resulting when this process takes place close to solid walls. This paper presents a numerical study of the interaction between a planar incident shock wave with a gas bubble. Simulations are performed using an inviscid compressible one-fluid solver composed by three conservation laws for mixture variables, namely mass, momentum and total energy along with a supplementary transport equation for the volume fraction of the gas phase [1-2]. Three-dimensional simulations are compared with the 2D reference case similar to the one presented in [3]. The mesh is composed by 1600x800x800 nodes. The time integration is done using a 3-step Rung-Kutta method and the numerical fluxes are computed with a second-order HLLC scheme.
Physics of Fluids, Nov 1, 2021
This paper presents a numerical study of the strong loads caused by the collapse of an air bubble... more This paper presents a numerical study of the strong loads caused by the collapse of an air bubble immersed in water at the vicinity of a wall and impacted by a normal shock wave. Simulations are performed using an efficient parallel fully compressible two-phase solver based on an homogeneous mixture model. Different configurations are investigated by varying the distance of the initial bubble to the wall. Comparisons are done with exiting results and with two-dimensional simulations highlighting large discrepancies on the computed pressure peaks. The computations show that the stand-off distance has significant effects on the collapse dynamics and the maximum wall pressure leading to potential wall damage. A power-law is proposed for the evolution of the maximum pressure peak as a function of the stand-off distance. Finally, a twin-bubble collapse is computed illustrating collective effects and the amplification of the pressure peak at the wall.
HAL (Le Centre pour la Communication Scientifique Directe), May 17, 2021
HAL (Le Centre pour la Communication Scientifique Directe), Dec 13, 2021
8th European Congress on Computational Methods in Applied Sciences and Engineering, 2022
International Journal of Numerical Methods for Heat & Fluid Flow, Nov 22, 2019
The aim of this work is to quantify the relative importance of the multiphase model for the simul... more The aim of this work is to quantify the relative importance of the multiphase model for the simulation of a gas bubble impacted by a normal shock wave in water. Both the free-field case and the collapse near a wall are investigated. Simulations are performed on both two-dimensional and three-dimensional configurations. The main phenomena involved in the bubble collapse are illustrated. A focus on the maximum pressure reached during the collapse is proposed. Design/methodology/approach Simulations are performed using an inviscid compressible homogeneous solver based on different systems of equations. It consists in solving different mixture or phasic conservation laws and a transport-equation for the gas volume fraction. Three-dimensional configurations are considered for which an efficient massively parallel strategy was developped. The code is based on a Finite Volume discretization for which numerical fluxes are computed with a HLLC scheme. Findings The comparison of three multiphase models is proposed. It is shown that a simple four-equation model is well-suited to simulate such strong shock-bubble interaction. The three-dimensional collapse near a wall is investigated. It is shown that the intensity of pressure peaks on the wall is drastically increased (more than 200%) in comparison with the cylindrical case. Originality Such a comparison of multiphase models in the case of a strong shock-induced bubble collapse is clearly original. Usually models are tested separately leading to a large dispersion of results. Moreover, simulations of a three-dimensional bubble collapse are scarce in the literature using such fine grids.
HAL (Le Centre pour la Communication Scientifique Directe), 2005
∗ Cemagref, 17 av. de Cucillé CS64427, F-35044 Rennes, France ∗∗ Laboratoire d' ´Etudes Aéro... more ∗ Cemagref, 17 av. de Cucillé CS64427, F-35044 Rennes, France ∗∗ Laboratoire d' ´Etudes Aérodynamiques - UMR CNRS 6609 - Université de Poitiers - ENSMA, Bd. Marie et Pierre Curie, F-86962 FUTUROSCOPE Chasseneuil, France [email protected]
Nucleation and Atmospheric Aerosols, 2020
We study the numerical solution to two-phase problems using the reduced Kapila model, i.e. four-e... more We study the numerical solution to two-phase problems using the reduced Kapila model, i.e. four-equations model. In particular we focus the present study on HLLC numerical flux approximation and its extensions. These numerical methods are tested on expansion tube cases with a final goal of constructing higher order hybrid numerical tool for solving the bubble collapsing under the shock wave problem.
Uploads
Papers by Philippe Parnaudeau