MEMPHIS

MEMPHIS - 2025

2025‌Activity reportProject-TeamMEMPHIS‌‌

RNSR: 201521153G

Research center Inria Centre at the‌ University of Bordeaux
In‌ partnership with:Université de‌‌ Bordeaux
Team name: Modeling Enablers for Multi-PHysics and‌ InteractionS
In collaboration with:‌Institut de Mathématiques de‌‌ Bordeaux (IMB)

Creation of the Project-Team: 2016 October‌ 01

Each year, Inria‌ research teams publish an‌‌ Activity Report presenting their work and results over‌ the reporting period. These‌ reports follow a common‌‌ structure, with some optional sections depending on the‌ specific team. They typically‌ begin by outlining the‌‌ overall objectives and research programme, including the main‌ research themes, goals, and‌ methodological approaches. They also‌‌ describe the application domains‌ targeted by the team, highlighting the scientific or‌ societal contexts in which their work is situated.‌

The reports then present the highlights of the‌ year, covering major scientific achievements, software developments, or‌ teaching contributions. When relevant, they include sections on‌ software, platforms, and open data, detailing the tools‌ developed and how they are shared. A substantial‌ part is dedicated to new results, where scientific‌ contributions are described in detail, often with subsections‌ specifying participants and associated keywords.

Finally, the Activity‌ Report addresses funding, contracts, partnerships, and collaborations at‌ various levels, from industrial agreements to international cooperations.‌ It also covers dissemination and teaching activities, such‌ as participation in scientific events, outreach, and supervision.‌ The document concludes with a presentation of scientific‌ production, including major publications and those produced during‌ the year.

Keywords

Computer Science and Digital Science‌

A6. Modeling, simulation and control
A6.1.1. Continuous Modeling‌ (PDE, ODE)
A6.1.5. Multiphysics modeling
A6.2.1. Numerical analysis‌ of PDE and ODE
A6.3.1. Inverse problems
A6.3.2.‌ Data assimilation
A6.3.4. Model reduction
A6.5.1. Solid mechanics‌
A6.5.2. Fluid mechanics
A9.2. Machine learning

Other Research‌ Topics and Application Domains

B2.2.1. Cardiovascular and respiratory‌ diseases
B4.2. Nuclear Energy Production
B4.3.2. Hydro-energy
B4.3.3.‌ Wind energy
B5.2.3. Aviation
B5.2.4. Aerospace
B5.5. Materials‌

1 Team members, visitors, external collaborators

Research Scientists‌

Michel Bergmann [INRIA, Senior Researcher,‌ until Oct 2025, HDR]
Michele Giuliano‌ Carlino [ONERA, until Oct 2025]‌
Alessia Del Grosso [INRIA, ISFP,‌ until Oct 2025]
Tommaso Taddei [INRIA‌, Researcher, until Aug 2025]

Faculty‌ Members

Angelo Iollo [Team leader, UNIV‌ BORDEAUX, Professor Delegation, until Oct 2025‌, HDR]
Angelo Iollo [Team leader‌, UNIV BORDEAUX, Professor, from Nov‌ 2025, HDR]
Afaf Bouharguane [UNIV‌ BORDEAUX, Associate Professor, HDR]

Post-Doctoral‌ Fellows

Joyce Ghantous [INRIA, Post-Doctoral Fellow‌, until Aug 2025]
Ivan Kharsansky Atallah‌ [ONERA, Post-Doctoral Fellow, from Feb‌ 2025 until Oct 2025]

PhD Students

Maxime‌ Chapron [ONERA, until Sep 2025]‌
Elise Declerck [ONERA, until Oct 2025‌]
Jon Labatut [ONERA, until Sep‌ 2025]
Karl Maroun [Université de Poitiers‌, until Jan 2025]
Abdessamad Moussaddak [‌EDF, CIFRE, until Oct 2025]‌
Marc-Olivier Potin [ONERA, until Oct 2025‌]
Alexis Tardieu [UNIV BORDEAUX, ATER‌, until Mar 2025]
Mathias Truel [‌INGELIANCE, CIFRE, until Oct 2025]‌
Alexis Valls [INRIA]

Interns and Apprentices‌

Sofia Curto [INRIA, Intern, from‌ Mar 2025 until Sep 2025]
Giorgia Lanciotti‌ [ONERA, Intern, from Mar 2025‌ until Aug 2025]
Samuel Oyhanto [INRIA‌, Intern, from Aug 2025 until Sep‌ 2025]
Samuel Oyhanto [INRIA, Intern‌, from Jun 2025 until Jul 2025]
Giovanni Polizzi [INRIA‌, Intern, from‌ Feb 2025 until Jul‌‌ 2025]

Administrative Assistant

Anne-Laure Gautier [INRIA‌]

External Collaborators

Philippe‌ Depouilly [CNRS]‌‌
Tommaso Taddei [UNIV SAPIENZA , from Sep‌ 2025 until Oct 2025‌]

2 Overall objectives‌‌

2.1 Multi-physics numerical modeling

2.1.1 Reduced-order models: convergence‌ between PDE models and‌ data

Unprecedented opportunities exist‌‌ to directly use already collected computational or experimental‌ data to improve and‌ build predictive models that‌‌ can be used online for the simulation of‌ parametric problems, robust design,‌ and control in science‌‌ and engineering. In this regard, our goal is‌ to combine mechanistic causal‌ models based on partial‌‌ differential equations (PDEs) with large data sets to‌ reduce the marginal cost‌ of predictions.

Reduced-order models‌‌ (ROMs) are our main tool for this purpose.‌ ROMs are parametric mathematical‌ models derived from the‌‌ full set of PDEs using previously computed solutions.‌ In many applications, the‌ solution space turns out‌‌ to be low-dimensional, so one can trade a‌ minimal loss of accuracy‌ for speed and scalability.‌‌ ROMs counteract the curse of dimensionality by significantly‌ reducing computational complexity.

Overall,‌ ROMs have reached a‌‌ certain level of maturity during the last decade,‌ allowing their implementation in‌ large-scale industrial codes, mainly‌‌ in structural mechanics. Nevertheless, some hard points stand.‌ Parametric problems governed by‌ strong advection fields or‌‌ sensibly compact-support solutions such as moving shocks suffer‌ from a limited possibility‌ of dimensional reduction and,‌‌ at the same time, insufficient generalization of the‌ model (out-of-sample solutions). The‌ main reason for this‌‌ is that the solution space is usually approximated‌ by an affine or‌ linear representation, which is‌‌ intrinsically broad band for such problems.

We have‌ worked on the development‌ of model order reduction‌‌ (MOR) techniques for nonlinear, advection-dominated problems, with emphasis‌ on projection-based Galerkin and‌ Petrov-Galerkin ROMs. First, we‌‌ worked on the development of effective sampling strategies‌ to reduce training costs.‌ Second, we developed nonlinear,‌‌ registration-based approximation techniques, to overcome limitations of linear‌ approximation methods (e.g., proper‌ orthogonal decomposition, POD) to‌‌ deal with strong advection fields. Third, we developed‌ hybrid formulations that combine‌ reduced-order and full-order models‌‌ to deal with complex flow features and/or complex‌ parameterizations.

2.1.2 Schemes for‌ Hierarchical and Chimera meshes,‌‌ multi-physics and asymptotic limits

The schemes we have‌ developed aim at simulating‌ complex multiphysics phenomena through‌‌ appropriate PDE modeling, automatic implicit geometry representation (level‌ sets), hierarchical Cartesian schemes‌ (quad-octrees), parallel simulations, and‌‌ accurate treatment of boundaries. Discretization schemes on hierarchical‌ meshes allow multiscale solution‌ of PDEs on non-body-fitted‌‌ meshes with a drastic reduction in computational setup‌ overhead. The key idea‌ is to use an‌‌ octree mesh to approximate the solution fields, while‌ the geometry is captured‌ by level set functions.‌‌ The boundary conditions are determined by appropriate interpolation‌ methods to achieve sufficient‌ accuracy. This approach eliminates‌‌ the need for boundary conforming meshes, which require‌ time-consuming and error-prone mesh‌ generation procedures, and opens‌‌ the door to easy‌ parallel simulation of very complex geometries.

One of‌ the limitations of this approach is that a‌ mesh with a fixed aspect ratio is not‌ optimal for very anisotropic fields such as boundary‌ layers. For such cases, we explored the idea‌ of using a body-fitted grid near the immersed‌ obstacles and a hierarchical mesh in the background.‌ Essentially, we use the techniques we have developed‌ to impose boundary conditions on non-body-matched meshes further‌ from the boundary, where the solution is smoother‌ and more isotropic. Our current investigations build on‌ discontinuous Galerkin (DG) methods / ADER approaches to‌ combine efficient interpolation strategies at the grid interfaces‌ and compact reconstruction of the data at the‌ grid level.

A differnt approach we have investigated‌ was a space–time FEM/FVM scheme on moving Chimera‌ (overset) grids for linear and nonlinear advection–diffusion problems,‌ with compact and accurate coupling across overlapping regions‌ via space–time interpolation polynomials and a FEM-predictor/FVM-corrector formulation‌ with a new space–time stabilization. The method is‌ extended to incompressible Navier–Stokes equations on evolving domains‌ using a fractional-step approach and a hybrid gradient‌ discretization that automatically handles grid overlap, achieving uniform‌ second-order accuracy in space and time without spurious‌ artifacts.

Part of our activity has been dedicated‌ to improve schemes for all Mach number flows‌ in both fluid dynamics and continuum mechanics. Phenomena‌ of interest involve fluid flows and elastic materials‌ whose deformations are investigated within a monolithic Eulerian‌ framework. With this approach any material (gas, liquid‌ or solid) can be described with the same‌ system of conservation equations and a suitable general‌ formulation of the constitutive law.

These schemes are‌ accurate in computing steady state solutions as well‌ as in approximating material wave propagation in various‌ Mach regimes and different materials. We studied methods‌ to overcome the need to solve for auxiliary‌ relaxation variables while preserving the properties of the‌ linearly implicit schemes. To achieve this, we split‌ the stiff relaxation source terms from the fluxes‌ and then reformulate the homogeneous part in an‌ elliptic form. Further research on obtaining all Mach‌ methods by including multi-dimensional knowledge in the numerical‌ scheme is also being conducted. More specifically, we‌ envisage to exploit a nodal pressure that depends‌ on all the cells around the given node‌ and naturally encompasses a consistent discretization of the‌ divergence of the velocity vector.

3 Research program‌

Coherently with our investigation approach, we start from‌ real-world applications to identify key methodological problems, then,‌ we study those problems and develop new methods‌ to address them; finally, we implement these methods‌ for representative test cases to demonstrate their practical‌ relevance.

MEMPHIS has already evolved in a new‌ project-team, MONHADE, in collaboration with Onera. The objective‌ of the MONHADE team is to develop hybrid‌ numerical and physical modelling methodologies that combine partial‌ differential equation–based models with high-fidelity data, ensuring explainability,‌ accuracy, and robustness, while remaining computationally efficient. These‌ methods target performance prediction, parametric optimisation, control, and data assimilation for fluid‌ systems interacting with structures.‌

3.1 Numerical models

We‌‌ aim to further develop automated model-order reduction (MOR)‌ procedures for large-scale systems‌ in computational mechanics —‌‌ here, automated refers to the ability to complete‌ the analysis with minimal‌ user intervention. First, we‌‌ wish to combine nonlinear MOR with mesh adaptation‌ to simultaneously learn rapid‌ and reliable ROMs and‌‌ effective high-fidelity discretizations over a range of parameters.‌ Second, we wish to‌ develop component-based MOR procedures‌‌ to build inter-operable components for steady and unsteady‌ nonlinear PDEs: towards this‌ end, we should develop‌‌ efficient localized training procedures to build local ROMs‌ for each archetype component,‌ and also domain decomposition‌‌ techniques to glue together the local models for‌ prediction. We also wish‌ to develop and analyze‌‌ hybrid approaches that combine and merge first-principle models‌ with data-fit models, and‌ also full-order and reduced-order‌‌ models for prediction of global engineering quantities of‌ interest.

We envision that‌ several methods that are‌‌ currently developed in the team can be complemented‌ by available tools from‌ machine learning: representative examples‌‌ include — but are not limited to —‌ solution clustering, optimal sampling,‌ classification. In this respect,‌‌ a leap forward in industrial applications that we‌ will pursue is without‌ doubts the possibility of‌‌ capitalizing on previous experience drawn from already acquired‌ simulations to build non-intrusive‌ models that combine non-linear‌‌ interpolations and non-linear regression. New perspectives in this‌ direction are offered by‌ the Chair Onera-Nouvelle Aquitaine‌‌ (cf. Regional initiatives).

As regards the work on‌ numerical discretization of PDEs,‌ compared to the previous‌‌ evaluation, we focus on the representation of the‌ solution in each computational‌ cell by adopting a‌‌ DG/ADER approach to improve the resolution of solution's‌ discontinuities. This approach is‌ complemented with a Chimera‌‌ grid at the boundaries in order to improve‌ accuracy by a body‌ fitted mesh avoiding grid‌‌ generation complexity for a general, possibly varying, geometrical‌ topology. The thesis of‌ Alexis Tardieu, which started‌‌ in October 2021 and is funded by the‌ University of Bordeaux, studies‌ this approach.

In parallel,‌‌ we continue our exploration of schemes in asymptotic‌ regimes such as low-‌ and high Mach numbers‌‌ for multi-material flows. We aim for schemes that‌ circumvent the problem of‌ accuracy and time stepping‌‌ in such regimes: the ultimate goal is to‌ devise asymptotic-preserving schemes that‌ are able to capture‌‌ phenomena at the time scale of the fast‌ waves and of the‌ material waves with the‌‌ same accuracy. For such a purpose, a new‌ path based on numerical‌ schemes with multi-dimensional knowledge‌‌ is also being explored.

3.2 Applications

For energy‌ applications, we will continue‌ our investigations on wave‌‌ energy converters and windturbines. Relative to wave energy‌ converters, we are developing‌ multifidelty models that couple‌‌ the incompressible Navier-Stoke equations (NSE) around the floater‌ with a Proper Orthogonal‌ (POD) ROM or a‌‌ simplified-physics model elsewhere.

We are also collaborating with‌ EDF to devise effective‌ ROMs for parametric studies.‌‌ In this collaboration, we‌ emphasize the implementation of projection-based ROMs for real-world‌ applications exploiting industrial codes.

In December 2023, Abdessamad‌ Moussaddak started his PhD thesis on model reduction‌ for river and coastal hydraulics.

In the framework‌ of RedLUM ANR project we develop and apply‌ mathematical and computational tools for real-time estimation and‌ short-term prediction of three-dimensional fluid flows using limited‌ computational resources, by coupling data, numerical simulations, and‌ sparse flow measurements. To achieve these goals, the‌ problem dimensionality are significantly reduced through data-driven and‌ reduced-order models, while the errors induced by dimensionality‌ reduction will be quantified using a stochastic, physics-informed,‌ multiscale parametrization.

The software development will be continued.‌ We will pursue the development of the NEOS‌ library: NEOS will be distributed in open source‌ LGPL-3.0. The HIWIND software will be rewritten based‌ on NEOS library.

4 Application domains

4.1 Energy‌ conversion

We apply the methods developed in our‌ team to the domain of wind engineering and‌ sea-wave converters. In Figure 1, we show‌ results of a numerical model for a sea-wave‌ energy converter. We here rely on a monolithic‌ model to describe the interaction between the rigid‌ floater, air and water; material properties such as‌ densities, viscosities and rigidity vary across the domain.‌ The appropriate boundary conditions are imposed at interfaces‌ that arbitrarily cross the grid using adapted schemes‌ built thanks to geometrical information computed via level‌ set functions 62. The background method for‌ fluid-structure interface is the volume penalization method 48‌ where the level set functions is used to‌ improve the degree of accuracy of the method‌ 5 and also to follow the object. The‌ underlined mathematical model is unsteady, and three dimensional;‌ numerical simulations based on a grid with $𝒪‌ (10^{8})$ degrees of freedom are‌ executed in parallel using 512 CPUs.

Figure 1: numerical modeling of a sea-wave‌ converter by a monolithic model and Cartesian meshes.‌

Figure 2.a — Figure 2: Aerogust project. Left:‌ met mast after its installation. Right: flow around‌ a NREL wind turbine rotor (as predicted by‌ NaSCar).

Figure 2.b — Figure 2: Aerogust project. Left:‌ met mast after its installation. Right: flow around‌ a NREL wind turbine rotor (as predicted by‌ NaSCar).

In the context of the Aerogust (Aeroelastic‌ gust modelling) European project, together with Valorem, we‌ investigated the behavior of wind turbine blades under‌ gust loading. The aim of the project was‌ to optimize the design of wind turbine blades‌ to maximize the power extracted. A meteorological mast‌ (Figure 2(a)) has been installed in March‌ 2017 in Brittany to measure wind on-site: data‌ provided by the mast have been exploited to‌ initialize the mathematical model. Due to the large‌ cost of the full-order mathematical model, we relied‌ on a simplified model 58 to optimize the‌ global twist. Then, we validated the optimal configuration‌ using the full-order Cartesian model based on the‌ NaSCar solver. Figure 2(b) shows the flow‌ around the optimized optimized wind turbine rotor.

4.2‌ Schemes for turbulent flow simulations using Octrees

We‌ have initially developed and tested a 3D first-order Octree code for unsteady‌ incompressible Navier-Stokes equations for‌ full windmill simulations with‌‌ an LES model and wall laws. We have‌ validated this code on‌ Occigen for complex flows‌‌ at increasing Reynolds numbers. This step implied identifying‌ stable and feasible schemes‌ compatible with the parallel‌‌ linear Octree structure. The validation has been conducted‌ with respect to the‌ results of a fully‌‌ Cartesian code (NaSCAR) that we run on Turing‌ (with significantly more degrees‌ of freedom) and with‌‌ respect to experimental results.

Subsequently, we have developed‌ a second-order Octree scheme‌ that has been validated‌‌ on Occigen for a sphere at a moderate‌ Reynolds number ( $Re‌ = 500$ ), see‌‌ Table 1. Then, for a cylinder at‌ ( $Re = 140000‌$ ) (Figures 3(a)‌‌ and 3(b)), close to real applications, we‌ have preliminary validation results‌ for the second-order scheme‌‌ with respect to experimental drag coefficient (Table 2‌). Additional resources will‌ be asked on Occigen‌‌ to complete the study.

Table 1: Flow past‌ a sphere at $Re‌ = 500$ . Results‌‌ in the literature are spread between $C_{D‌} = 0 . 48‌$ and $C_{D} =‌‌ 0.52$ .
Mesh	$Δ x_{min‌}$	number of cells	${C‌}_{D}$ (1st-order‌‌ scheme)	$C_{D}$ (2nd-order scheme)
1‌	$0.094$	$0‌ . 72 \cdot {10‌‌}^{5}$	N.A.	$0 . 526$
2	$0 .‌ 047$	$4 . 9‌ \cdot 10^{5}$	$0‌‌ . 595$	$0 . 522$
3	$0 .‌ 023$	$4 . 7‌ \cdot 10^{6}$	$0‌‌ . 546$	$0 . 492$
4	$0 .‌ 012$	$37 . 6‌ \cdot 10^{6}$	$0‌‌ . 555$	$0 . 496$

Table 2: Flow‌ past a sphere at‌ $Re = 14000$ .‌‌
Case	$C_{D}$
Octree, 1st-order scheme‌	$1.007$
Octree,‌ 2nd-order scheme‌‌	$1.157$
Cartesian	$1.188$
Experimental‌ estimate 54	$1 .‌ 237$

Figure 3.a — Figure‌‌ 3: flow past a cylinder at $Re‌ = 140000$ . Left:‌ vorticity contour lines. Right:‌‌ streamwise velocity section and grid for the second-order‌ Octree scheme.

Figure 3.b — Figure‌‌ 3: flow past a cylinder at $Re‌ = 140000$ . Left:‌ vorticity contour lines. Right:‌‌ streamwise velocity section and grid for the second-order‌ Octree scheme.

4.3 Vascular‌ flows

A new research‌‌ direction pursued by the team is the mathematical‌ modelling of vascular blood‌ flows in arteries. Together‌‌ with the start-up Nurea and the surgeon Eric‌ Ducasse, we aim at‌ developing reliable and automatic‌‌ procedures for aneurysm segmentation and for the prediction‌ of aneurysm rupture risk.‌ Our approach exploits two‌‌ sources of information: (i) numerical simulations of blood‌ flows in complex geometries,‌ based on an octree‌‌ discretization, and (ii) computed tomography angiography (CTA) data.‌ Figure 4 shows the‌ force distribution on the‌‌ walls of the abdominal aorta in presence of‌ an aneurysm; results are‌ obtained using a parallelized‌‌ hierarchical Cartesian scheme based on octrees.

Further information‌ is given in the‌ sections dedicated to the‌‌ new results.

Figure 4: force‌ distribution on the walls‌ of the abdominal aorta‌‌ in presence of an‌ aneurysm.

4.4 Fluid-structure interactions using Eulerian non-linear elasticity‌ models

Mathematical and numerical modeling of continuum systems‌ undergoing extreme regimes is challenging due to the‌ presence of large deformations and displacements of the‌ solid part, and due to the strongly non-linear‌ behavior of the fluid part. At the same‌ time, proper experiments of impact phenomena are particularly‌ dangerous and require expensive facilities, which make them‌ largely impractical. For this reason, there is a‌ growing interest in the development of predictive models‌ for impact phenomena.

In MEMPHIS, we rely on‌ a fully Eulerian approach based on conservation laws,‌ where the different materials are characterized by their‌ specific constitutive laws, to address these tasks. This‌ approach was introduced in 56 and subsequently pursued‌ and extended in 60, 55, 51‌, 53 and 13. In Figure 5‌, we show the results of the numerical‌ simulation of the impact of a copper projectile‌ immersed in air over a copper shield. Results‌ are obtained using a fully parallel monolithic Cartesian‌ method, based on a $4000^{2}$ fixed Cartesian‌ grid. Simulations are performed on a cluster of‌ 512 processors, and benefits from the isomorphism between‌ grid partitioning and processor topology.

Figure 5.a — Figure‌ 5: impact and rebound of a copper‌ projectile on a copper plate. Interface and schlieren‌ at $50 μ s$ , $199 μ s‌$ , $398 μ s$ and $710 μ s‌$ .

Figure 5.b — Figure‌ 5: impact and rebound of a copper‌ projectile on a copper plate. Interface and schlieren‌ at $50 μ s$ , $199 μ s‌$ , $398 μ s$ and $710 μ s‌$ .

In figure 6, we show the‌ results of a three dimensional simulation of a‌ cardiac pump (LVAD, left ventricule assisted device).

Figure 6: 3D numerical modeling‌ of a cardiac pump (LVAD) with an elastic‌ membrane.

5 Social and environmental responsibility

As discussed‌ in the previous section, we are particularly interested‌ in the development of mathematical models and numerical‌ methods to study problems related to renewable energies,‌ and ultimately contribute to next-generation sustainable solutions for‌ energy extraction.

5.1 Impact of research results

We‌ are studying two types of green energy extractors:‌ wave energy converters (WECs) and wind energy.

As‌ regards WECs, we are working with the PoliTO‌ (Torino, Italy) to model the behavior of inertial‌ sea wave energy converters (ISWEC), and we are‌ also working with a Bordeaux-based start-up for another‌ device to extract energy from waves via an‌ Inria-Tech project and a Nouvelle-Aquitaine Regional Project submitted‌ by Memphis in collaboration with the CARDAMOM team.‌

As regards wind energy, we focus on the‌ analysis of wind turbines. In the past, we‌ have supervised two PhD CIFRE theses with VALOREM-Valeol,‌ and are currently working with them in a‌ European RISE ARIA project led by Memphis. We‌ also work with IFPEN on the aeroelastic modeling‌ of large wind turbines and the study and‌ optimization of turbines farms in the framework of‌ the joint laboratory Inria-IFPEN with a thesis funded‌ by IFPEN and a post-doc funded by Inria‌ (which started in October 2021).

In conjunction with these activities, in collaboration‌ with ANDRA (the national‌ agency for storage of‌‌ nuclear waste), we investigated the development of reduced-order‌ models to allow efficient‌ and accurate simulations for‌‌ deep geological storage planning. This activity was the‌ subject of the PhD‌ thesis of Giulia Sambataro‌‌ who successfully defended her PhD thesis in December‌ 2022.

5.1.1 Critical Infrastructure‌ & Defense

EDF (Energy):‌‌ A CIFRE thesis (A. Moussaddak) was dedicated to‌ the mechanical state estimation‌ of nuclear containment buildings‌‌ (Vercors mock-up). The work applied data assimilation techniques‌ to identify parameters crucial‌ for the safety monitoring‌‌ of prestressed concrete structures.
Ingeliance (Defense): This partnership‌ (CIFRE thesis, M. Truel)‌ focused on Unsteady Hybrid‌‌ Models for Fluid-Structure Interaction (FSI), aiming to reduce‌ computational costs for complex‌ design simulations.

6 Highlights‌‌ of the year

The team has organized the‌ international workshop "Accurate reduced‌ Order Models". Bidart. .‌‌
A publication highlighting the effectiveness and robustness of‌ the fluid–structure interaction schemes‌ developed by the team‌‌ for interdisciplinary applications: Guillaume Ravel, Théo Mercé, Michel‌ Bergmann , Anja Knoll-Gellida,‌ Afaf Bouharguane, Sara Al‌‌ Kassir, Angelo Iollo, and Patrick J. Babin. Modeling‌ zebrafish electric field pulse-induced‌ escape response reveals neuromuscular‌‌ energetic constraints and efficient body movement adaptation to‌ high-viscosity fluids. iScience 28,‌ 112056, March 21, 2025.‌‌
Building on the competences of the MEMPHIS team‌ in high-fidelity numerical methods,‌ reduced-order modeling, and hybrid‌‌ AI/CFD approaches, the transition of MEMPHIS into the‌ joint Inria–Onera project-team MONHADE‌ constitutes a strategic response‌‌ to the growing demand for certified numerical simulation‌ in critical sectors. With‌ the alliance with the‌‌ French Aerospace Lab (Onera), the team forms a‌ focused task force targeting‌ challenges in hybrid AI/CFD‌‌ modeling for defense and aeronautics.
Afaf Bouharguane, member‌ of the team, has‌ defended her habilitation.

7‌‌ Latest software developments, platforms, open data

7.1 Latest‌ software developments

7.1.1 COCOFLOW‌

Keywords:
3D, Elasticity, MPI,‌‌ Compressible multimaterial flows
Functional Description:
The code is‌ written in fortran 95‌ with a MPI parallelization.‌‌ It solves equations of conservation modeling 3D compressible‌ flows with elastic models‌ as equation of state.‌‌
Contact:
Florian Bernard
Partners:
CNRS, Université Bordeaux 1‌

7.1.2 KOPPA

Name:
Kinetic‌ Octree Parallel PolyAtomic
Keyword:‌‌
Numerical simulations
Functional Description:
KOPPA is a C++/MPI‌ numerical code solving a‌ large range of rarefied‌‌ flows from external to internal flows in 1D,‌ 2D or 3D. Different‌ kind of geometries can‌‌ be treated such as moving geometries coming from‌ CAO files or analytical‌ geometries. The models can‌‌ be solved on Octree grids with dynamic refinement.‌
Contact:
Angelo Iollo
Participant:‌
an anonymous participant

7.1.3‌‌ NaSCar3D

Name:
Navier-Stokes Cartesian 3D
Keywords:
Navier-Stokes, Cartesian‌ grid
Scientific Description:
NaSCar‌ can be used to‌‌ simulate both hydrodynamic bio-locomotion as fish like swimming‌ and aerodynamic flows such‌ wake generated by a‌‌ wind turbine.
Functional Description:

This code is devoted‌ to solve 3D-flows in‌ around moving and deformable‌‌ bodies. The incompressible Navier-Stokes equations are solved on‌ fixed grids, and the‌ bodies are taken into‌‌ account thanks to penalization‌ and/or immersed boundary methods. The interface between the‌ fluid and the bodies is tracked with a‌ level set function or in a Lagrangian way.‌ The numerical code is fully second order (time‌ and space). The numerical method is based on‌ projection schemes of Chorin-Temam's type. The code is‌ written in C language and use Petsc library‌ for the resolution of large linear systems in‌ parallel.

NaSCar can be used to simulate both‌ hydrodynamic bio-locomotion as fish like swimming and aerodynamic‌ flows such wake generated by a wind turbine.‌
URL:
https://gitlab.inria.fr/memphis/nascarElast3D
Contact:
Michel Bergmann
Participant:
an anonymous‌ participant

7.1.4 NS-penal

Name:
Navier-Stokes-penalization
Keywords:
3D, Incompressible‌ flows, 2D
Functional Description:
The software can be‌ used as a black box with the help‌ of a data file if the obstacle is‌ already proposed. For new geometries the user has‌ to define them. It can be used with‌ several boundary conditions (Dirichlet, Neumann, periodic) and for‌ a wide range of Reynolds numbers.
Contact:
Charles-Henri‌ Bruneau
Partner:
Université de Bordeaux

7.1.5 HiWind

Keyword:‌
Simulation
Functional Description:
Hiwind is a software that‌ allows to model in 2D and 3D the‌ effects of air flow on a wind turbine‌ blade (moving solid or elastic structures), and to‌ simulate numerically their interactions. Hiwind also allows to‌ model and characterize the abnormal behavior to warn‌ about a potential weakening of the structure. Hiwind‌ is a "drag and drop" solution (automated meshing‌ phase), massively parallel, and uses adaptive meshing.
Contact:‌
Angelo Iollo
Partner:
Valeol

7.1.6 NEOS

Functional Description:‌
NEOS is a software framework for numerical modeling‌ of multiphysical problems on hierarchical Cartesian meshes (quadtree‌ in 2D and octree in 3D). It is‌ mainly based on the bitpit library (https://optimad.github.io/bitpit/). NEOS‌ provides : - the creation and parallel management‌ of hierarchical Cartesian meshes (2D quadtree or 3D‌ octree) - global or local mesh refinement (based‌ on a distance of levelset or other physical‌ criteria) - the management of several moving geometries‌ in an analytical or explicit form (STL files‌ or others) - the calculation of geometries levelsets‌ at any point of the mesh - 2D/3D‌ differential operators of (gradient, laplacian, hessian, ...) -‌ various 2D/3D interpolators (bilinear, radial basis functions RBF)‌ - an API for solvers (currently with PETSC)‌ - a complete Python3 interface
Contact:
Michel Bergmann‌

8 New results

8.1 Nonlinear approximation methods based‌ on coordinate transformations

Participants: Angelo Iollo, Jon‌ Labatut, Tommaso Taddei, Ishak Tifouti,‌ Mathias Truel, Michel Bergmann.

A significant‌ limitation of MOR techniques that rely on linear‌ approximation spaces is their inability to effectively handle‌ parameter-dependent sharp gradients, which naturally arise in the‌ solutions to advection-dominated problems. This inadequacy of linear‌ approximations has spurred the development of nonlinear methods.‌ Among these, nonlinear approaches based on coordinate transformations‌ have demonstrated their effectiveness across a wide range‌ of problems in computational mechanics. These techniques rely‌ on a registration procedure to find a parametric spatio(-temporal) transformation that improves‌ the linear compressibility of‌ the solution set.

The‌‌ development of registration techniques for MOR has been‌ a major focus of‌ the team's research. We‌‌ refer to the HDR thesis of Tommaso Taddei‌ for a recent review‌ of the ongoing efforts‌‌ on the subject 23. In the past‌ year, three significant advances‌ have been achieved in‌‌ this direction.

Jon Labatut 35 developed a novel‌ registration framework based on‌ parametric vector flows and‌‌ coupled the approach with the convex displacement interpolation‌ method of 14 for‌ two-dimensional and three-dimensional aerodynamic‌‌ problems. Jon Labatut defended his PhD thesis in‌ November 2025 40;‌ we are currently working‌‌ on the submission of two papers from the‌ thesis. Figure 7 shows‌ select numerical results from‌‌ Labatut's thesis for a three-dimensional viscous problem.
Iollo‌ and Taddei proposed a‌ new registration method in‌‌ bounded domains using Fokker Planck equation 15.‌ This paper has led‌ to an ongoing collaboration‌‌ with Giovanni Russo (University of Catania) and Klaas‌ Wilhems (KU Leuven) on‌ the development of specialized‌‌ methods for the Fokker Planck equation in bounded‌ domains.
Tifouti 34 developed‌ a registration-based MOR framework‌‌ with local reduced-order bases that extends the work‌ in 2 to deal‌ with shock-topology changes. Ishak‌‌ Tifouti defended his PhD thesis in November 2025‌ 24; we expect‌ to submit one paper‌‌ based on his work in 2026.
Mathias Truel‌ mainly worked on a‌ non-linear interpolation technique based‌‌ on mappings. A mapping is defined using Free-Form‌ Deformation (FFD) and is‌ parametrized by control points.‌‌ The displacement of control points is tuned via‌ the minimization of an‌ objective function. Implementation using‌‌ the JAX library allows the use of a‌ fast L-BFGS solver. In‌ Figure 8, a‌‌ simple registration case between $u_{0}$ and ${u‌}_{1}$ is demonstrated. The‌ domain is divided into‌‌ two subdomains and two FFD formulations are used.‌ The control points for‌ the first (red circles)‌‌ and second (blue squares) FFD are depicted. The‌ yellow crosses are synchronization‌ points, where the two‌‌ mappings are equal. Overall, four mappings are optimized.‌ Two for the direction‌ $0 \to 1$ and‌‌ two for $1 \to 0$ . After minimizing‌ the objective function, two‌ interpolations are constructed, one‌‌ where $u_{1}$ is deformed into $u_{0‌}$ and the opposite. Displacements‌ of control points and‌‌ synchronization points are also shown. To easily visualize‌ the mappings in the‌ domains, a checkerboard pattern‌‌ is deformed.

Application of the convex displacement interpolation‌ to an ONERA M6‌ wing in the transonic‌‌ viscous regime for varying angle of attack. Left:‌ density over the wing.‌ Right: comparison of density‌‌ residuals convergence with different initialization of the full-order‌ model (CDI=convex displacement interpolation;‌ CI=convex interpolation; P0=piecewise-constant DG‌‌ solution).

Application of the convex displacement interpolation to‌ an ONERA M6 wing‌ in the transonic viscous‌‌ regime for varying angle of attack. Left: density‌ over the wing. Right:‌ comparison of density residuals‌‌ convergence with different initialization‌ of the full-order model (CDI=convex displacement interpolation; CI=convex‌ interpolation; P0=piecewise-constant DG solution).

8.2 Collocation-based‌ reduced order models: analysis and applications

Participants: Michel‌ Bergmann, Michele Giuliano Carlino, Elise Declerck‌, Alessia Del Grosso, Angelo Iollo,‌ Marc-Olivier Potin.

The primary methodological investigation centers‌ on collocation strategies in Model Order Reduction (MOR)‌ 27, designed to enhance the efficiency of‌ the online-offline paradigm. Unlike standard pROMs where solutions‌ are projected onto a reduced basis, this approach‌ solves governing equations at a sparse set of‌ optimal collocation points. The research utilizes the Non-Negative‌ Least Square (NNLS) algorithm to identify a minimal‌ set of collocation points (cells) and associated positive‌ weights during the offline phase. This significantly reduces‌ computational cost as the number of collocation points‌ is much smaller than the total discretization cells.‌

In the most recent developments, not published yet,‌ we analyze two distinct implementations:

Collocated projection-based ROM‌ (cpROM): This method computes the solution approximation in‌ all cells using the reduced ansatz. Theoretically, it‌ is the closest to the classic pROM; if‌ all discretized cells are used as collocation points,‌ cpROM and pROM coincide.
Collocation-based ROM (cROM): This‌ method applies the reduced ansatz only in cells‌ neighboring the collocation points. It differs significantly in‌ theory; if all cells are used, the cROM‌ coincides with the HF discretization.

Current work establishes‌ theoretical foundations for these methods. Specifically, for the‌ cROM case, an equivalence result analogous to the‌ Lax-Richtmyer theorem 59 is derived, where stability and‌ consistency imply convergence.

Structure-preserving cROM.

A critical challenge‌ in MOR is ensuring that reduced models retain‌ the essential physical properties of the HF source‌ model. This line of research aims to develop‌ structure-preserving ROMs that strictly satisfy physical constraints. We‌ are currently investigating the shallow water system with‌ wet-dry fronts. In particular, Giorgio Musso, during his‌ master thesis, implemented a novel transformation approach to‌ strictly preserve the non-negativity of water height during‌ simulations, a property often violated by classical projection‌ methods. Giorgio will continue working on these topics‌ during his PhD that has recently started (supervised‌ by A. Del Grosso in collaboration with C.‌ Fiorini of CNAM).

Implicit-explicit (IMEX) schemes for cROM.‌

We extended the collocation-based methodology to IMEX time‌ integrators. The explicit contribution is treated via projection/prolongation‌ of the reduced solution onto cells within the‌ collocation stencils. The implicit stage involves solving a‌ system $A u = b$ of size $N‌$ . By pre-multiplying by a Boolean masking matrix‌ $M \in ℝ^{{N}_{c} \times N}$ and‌ post-multiplying by the prolongation operator $P \in {ℝ‌}^{N \times N_{c}}$ , we obtain a‌ reduced implicit system:

M​​﻿﻿ A P u =​​​‌ b_{c}

This work is conducted in collaboration‌ with CNRS researcher Walter Boscheri and Beatrice Battisti‌ (Université Savoie Mont Blanc). To handle physical models‌ with non-linear diffusion coefficients, we are currently investigating hyper-reduction strategies (specifically gappy‌ POD) to approximate algebraic‌ operators without full domain‌‌ assembly.

Neural Network Correction for Explicit Schemes.

To‌ enhance the accuracy of‌ cROMs, we introduced a‌‌ hybrid correction strategy 50. A neural network‌ is trained to learn‌ a nonlinear decoder that‌‌ accounts for solution features lying in the orthogonal‌ complement of the Reduced‌ Basis. At each time‌‌ step, the standard cROM prediction is enriched by‌ a nonlinear correction term.‌ The network uses the‌‌ orthogonal residual—obtained by projecting high-fidelity snapshots onto the‌ complement of the reduced‌ space—as an input feature.‌‌ This approach preserves the non-intrusiveness of the collocation‌ framework while capturing complex‌ nonlinear behaviors that escape‌‌ linear subspace approximations. Preliminary results indicate significant accuracy‌ improvements in multiscale and‌ strongly nonlinear regimes. This‌‌ work is ongoing.

CFD-based reduced-order modeling of a‌ turbine stage for parametric‌ optimization.

In order to‌‌ approach the problem, we first modelize the air‌ flow using 1D Euler‌ equations in a tube.‌‌ These equations are solved numerically in python using‌ finite volume schemes.

Then we built two types of reduced‌ order model (ROM) of‌ this high fidelity model‌‌ : Petrov-Galerkin projection-based reduced order model (pROM) and‌ collocation-based reduced order model‌ (cROM) The equations are‌‌ solved in a subspace determined with proper orthogonal‌ decomposition (POD). In the‌ pROM, the non-linear term‌‌ is hyper-reduced using the NNLS algorithm to prevent‌ a computational bottleneck. The‌ cROM uses NNLS and‌‌ the POD to determine a subset of points‌ where the high-fidelity volume‌ scheme is solved. Domain‌‌ decomposition consists in dividing the domain into subdomains.‌ Each subdomain has its‌ own ROM, allowing multifidelity‌‌ simulation. We use such simulations to train a‌ model starting from a‌ coarse grid then using‌‌ a finer grid and full-order model on each‌ subdomain with a MOR‌ on the other. Repeating‌‌ these steps, we build a component-based ROM on‌ the domain for the‌ finer grid (see Figure‌‌ 9).

Optimal Domain Decomposition in Multifidelity Modeling‌ for Uncertain Geometries in‌ Fluid-Structure Interaction.

In the‌‌ PhD thesis of Marc-Olivier Potin, we develop a‌ numerical hybridization architecture based‌ on the multi-fidelity paradigm.‌‌ This approach aims to exploit the plurality of‌ industrial flows: while critical‌ areas (where strong nonlinearities‌‌ and transient phenomena occur) require high-fidelity resolution (HFM),‌ large portions of the‌ fluid domain exhibit more‌‌ regular dynamics that can be effectively described by‌ reduced-order models (ROMs). Figure‌ 10 illustrates part of‌‌ the results of the hybrid multi-fidelity scheme applied‌ to 1D Euler equations‌ in a Laval nozzle‌‌ configuration (overset mesh). This test case, representative of‌ transonic compressible flows with‌ shock wave, highlights the‌‌ behavior of the hyper-reduction algorithm (cMOR) on the‌ background grid. The results‌ demonstrate a direct correlation‌‌ between the selection tolerance $ϵ$ and the fidelity‌ of the solution. Analysis‌ of the cost/accuracy trade-off‌‌ reveals that it is‌ possible to achieve a relative space-time error of‌ less than 2%, while reducing computation time by‌ up to 55%. This validation on 1D cases‌ provides the proof of concept needed to extend‌ this procedures to complex 2D industrial geometries.

8.2.1 Component-based model reduction‌ of complex systems

Participants: Tommaso Taddei, Lei‌ Zang.

“Classical” (monolithic) MOR methods rely on‌ three fundamental assumptions: first, the solution is defined‌ over a parameter-independent spatial domain; second, it is‌ computationally feasible to solve the full-order model (FOM)‌ for several parameter values during the offline stage;‌ third, parameter variations induce changes in the solution‌ field that can be captured by a global‌ (in space and in parameter) low-dimensional approximation. Classical‌ MOR methods are hence ill-suited to deal with‌ problems with parameter-induced topology changes.

Component-based (or localized)‌ model order reduction (CB-MOR) techniques combine approaches from‌ model reduction and domain decomposition (DD) to overcome‌ or significantly mitigate limitations of monolithic strategies. CB-MOR‌ techniques are based on the introduction of a‌ library of archetype components and are heavily inspired‌ by the extensive literature on component mode synthesis‌ (CMS) and dynamic substructuring. Research on CB-MOR has‌ reached a mature stage for linear problems; however,‌ extending these techniques to nonlinear and time-dependent regimes‌ remains an extremely challenging task.

In 2025, Tommaso‌ Taddei and collaborators worked on the development of‌ component-based model reduction techniques for incompressible flows. In‌ more detail, in 65, they extended the‌ method of 63 to fluid structure interaction problems‌ that feature laminar incompressible flows with hyper-elastic structures.‌ More recently, the same authors have considered a‌ different coupling strategy that is provably energy stable‌ at the semi-discrete level and have shown that‌ the use of high-order implicit Runge Kutta methods‌ leads to dramatic improvements in the long-term stability‌ of the reduced-order model 64.

Figures 11‌ and 12 show select results from 64 for‌ the Turek benchmark (FSI3, 67).

Figure 11‌ shows three snapshots of the horizontal velocity field‌ at three time instants $t = 32,‌ 35, 38$ s for a representative parameter‌ value: the plots illustrate the formation of a‌ periodic vortex street downstream of the cylinder and‌ the associated oscillatory motion of the beam.

Figures 12(a) and (b) show‌ the spectra of drag and lifted forces obtained‌ by ROMs for different POD tolerances at two‌ representative test points: we observe that the CB-ROM‌ is effective to approximate the spectra of the‌ force on the structure for out-of-sample configurations. Figure‌ 12(c) shows the number of retained POD‌ modes for velocity ( $n_{u}$ ), pressure‌ ( $n_{p}$ ), and displacement ( ${n‌}_{s}$ ) — the method also includes control‌ variables associated with the displacement at the interface $n_{c}$ . We‌ notice that the number‌ of modes that are‌‌ required to reach an acceptable accuracy is extremely‌ large: this result clearly‌ shows the need for‌‌ nonlinear approximation methods.

8.3 Reduced-Order‌ Modeling with Active Subspace‌ Closure

Participants: Angelo Iollo‌‌, Tommaso Taddei, Alexis Valls.

This‌ activity focuses on the‌ development of data-driven closure‌‌ strategies for reduced-order models (ROMs) of nonlinear, advection-dominated‌ dynamical systems. The work‌ is illustrated on the‌‌ Kuramoto–Sivashinsky equation, a canonical model sharing structural similarities‌ with the Navier–Stokes equations‌ while remaining computationally affordable‌‌ for extensive high-fidelity simulations.

Starting from a Proper‌ Orthogonal Decomposition (POD)–Galerkin projection,‌ the system dynamics are‌‌ decomposed into resolved large-scale modes and unresolved small-scale‌ modes. Standard Galerkin ROMs‌ fail to accurately reproduce‌‌ the long-term dynamics due to the neglected influence‌ of unresolved scales, which‌ motivates the introduction of‌‌ closure terms.

A first approach considers a Markovian,‌ memoryless closure, where the‌ effect of unresolved modes‌‌ is approximated by parametric correction terms depending only‌ on the resolved variables.‌ These parameters are identified‌‌ from data using least-squares regression on high-fidelity snapshots.‌ While this approach improves‌ short-term accuracy, it requires‌‌ a large number of learned parameters and offers‌ limited interpretability.

To address‌ these limitations, a novel‌‌ closure strategy based on the concept of active‌ subspaces is introduced. The‌ key idea is to‌‌ identify low-dimensional directions in the unresolved space that‌ have the strongest influence‌ on the closure term.‌‌ This is achieved by analyzing the sensitivity of‌ the closure with respect‌ to unresolved variables and‌‌ computing the dominant eigenvectors of an associated averaged‌ gradient covariance operator. The‌ resulting active modes define‌‌ a reduced basis for the unresolved dynamics, leading‌ to a compact and‌ physically interpretable closure model.‌‌

Preliminary results indicate that only a small number‌ of active directions are‌ sufficient to capture the‌‌ dominant interactions between resolved and unresolved scales, supporting‌ the relevance of the‌ proposed approach. Ongoing work‌‌ focuses on coupling the reduced active variables with‌ the resolved dynamics through‌ consistent evolution equations, paving‌‌ the way toward efficient and robust non-intrusive ROM‌ closures for complex fluid‌ systems.

8.4 Output reduction:‌‌ Clustered active subspaces

Participants: Michel Bergmann, Maxime‌ Chapron.

Aerodynamic shape‌ optimization often involves navigating‌‌ extremely high-dimensional design spaces and relying on computationally‌ expensive simulations, making standard‌ surrogate-based optimization methods difficult‌‌ to apply because they suffer from the curse‌ of dimensionality. A widely‌ used strategy to overcome‌‌ this challenge is dimension reduction, which seeks to‌ identify a smaller set‌ of directions that most‌‌ strongly influence the objective function. One prominent technique,‌ known as Active Subspaces‌ (AS), uses gradient information‌‌ to uncover linear combinations of input parameters that‌ explain the dominant variations‌ in system performance, allowing‌‌ surrogate models such as‌ Kriging to be trained within a far lower-dimensional‌ space while maintaining accuracy.

Yet, global dimension reduction‌ can be insufficient when dealing with complex, highly‌ nonlinear, or multimodal response surfaces, since no single‌ reduced representation can capture all relevant behaviors across‌ the entire design domain. To address this limitation,‌ the Clustered Active Subspaces (CAS) framework adopts a‌ divide-and-conquer methodology:

The design space is first partitioned‌ using clustering algorithms such as Gaussian Mixture Models,‌ which identify regions of similar functional behavior rather‌ than grouping points simply by geometric distance.
Each‌ cluster is then assigned its own local active‌ subspace and its corresponding local surrogate model.
These‌ local predictions are combined through a Mixture-of-Experts formulation‌ to construct a coherent global model.

This approach‌ improves robustness, scalability, and predictive accuracy, particularly in‌ complex aerodynamic settings. Within Bayesian Optimization, performing the‌ search in the reduced space greatly simplifies the‌ maximization of acquisition functions. Two additional steps are‌ essential for this integration: accurately defining the boundaries‌ of the reduced domain (which takes the form‌ of a zonotope) to avoid unreliable extrapolation; and‌ mapping optimized points from the reduced space back‌ into the full-dimensional design space.

This work has‌ been carried out with ONERA during the thesis‌ of Maxime Chapron, also presented at several conferences‌ 12, 11.

8.5 Local reduction: Data-Driven‌ Wall Modeling for Industrial RANS Simulations

Participants: Michel‌ Bergmann, Michele Romanelli.

By enabling accurate‌ predictions of wall shear stress without resolving the‌ near-wall viscous sublayer, data-driven wall models provide a‌ powerful form of local model reduction that allows‌ the use of significantly coarser meshes in the‌ vicinity of the wall while maintaining high fidelity‌ in RANS simulations. During the PhD thesis of‌ Michele Romanelli 22, two works 20,‌ 21 advance the development of accurate and computationally‌ efficient data-driven wall models for turbulent boundary layers‌ subjected to arbitrary pressure gradients.

Implicit Framework: The‌ first study 20 introduces a general machine-learning framework‌ that replaces traditional analytical wall laws with a‌ non-parametric model constructed from high-fidelity data. By reformulating‌ the wall boundary condition as a Dirichlet-to-Neumann operator‌ applied at a fixed distance from the wall‌ (typically $y^{+} \approx 30$ –50), a neural‌ network predicts instantaneous wall shear stress. While significantly‌ outperforming classical formulations in regions of strong acceleration‌ or separation, this approach relies on an iterative‌ recovery of skin-friction, adding computational overhead.
Explicit Framework:‌ The second contribution 21 resolves this bottleneck by‌ proposing a fully explicit data-driven wall model. A‌ dedicated neural network directly outputs the magnitude and‌ direction of the wall shear stress from an‌ enriched set of non-dimensional local inputs (including velocity-profile‌ samples and pressure-gradient components). This explicit formulation removes‌ the need for any iterative solver, yielding one‌ to two orders of magnitude speedup while preserving‌ essentially identical predictive accuracy.

Together, these developments establish‌ a coherent framework for data-driven wall modeling, demonstrating‌ that such models are mature enough for deployment in aerodynamic design and‌ uncertainty-quantification workflows.

8.6 Numerical‌ Schemes for Multiphysics

Participants:‌‌ Alessia Del Grosso, Angelo Iollo.

Non-conservative‌ and Relaxation Schemes for‌ Monolithic Elastic Modeling.

We‌‌ investigated the numerical treatment of the evolution equation‌ for gradient deformation in‌ solid mechanics. Traditional conservative‌‌ approaches represent this variable using the gradient of‌ backward characteristics. While this‌ ensures conservation and captures‌‌ discontinuities, it significantly increases the system size (four‌ equations in 2D and‌ nine in 3D, compared‌‌ to two and three, respectively) and complicates the‌ preservation of the irrotational‌ property. To address this,‌‌ we proposed a non-conservative formulation by directly integrating‌ the transport equation for‌ the backward characteristics. This‌‌ approach reduces the system size and automatically preserves‌ the involutive constraint to‌ machine precision, whereas standard‌‌ Lax-Friedrichs (LF) methods typically exhibit errors of order‌ $10^{- 2}$ .‌ However, simulating non-conservative models‌‌ with discontinuous solutions presents stability challenges, specifically regarding‌ path-dependence 61. To‌ ensure consistency with weak‌‌ solutions, we developed a simplified approach based on‌ the Jin-Xin relaxation strategy‌ 52. As illustrated‌‌ in Figure 13, the proposed relaxation method‌ yields results analogous to‌ the reference conservative LF‌‌ method while maintaining a reduced computational footprint.

Density‌ solution for the 2D‌ Riemann problem in a‌‌ copper domain (20 contour lines). Top Left: Reference‌ conservative Lax-Friedrichs (LF) solution.‌ Top Right: Explicit non-conservative‌‌ relaxation scheme. Bottom Left: Non-conservative IMEX scheme with‌ explicit time step. Bottom‌ Right: IMEX scheme with‌‌ a time step $1000 \times$ larger than the‌ explicit limit.

Implicit-explicit strategies‌ for multi-scale flows.

A‌‌ significant challenge in fluid-structure interaction is the disparity‌ in wave speeds, particularly‌ in the subsonic regime‌‌ ( $M < 1$ ) where solid waves‌ are significantly slower than‌ fluid waves. The classic‌‌ Courant-Friedrichs-Lewy (CFL) condition is driven by the fast‌ waves, leading to excessive‌ diffusion in the approximation‌‌ of slow (solid) waves.

To mitigate this, we‌ applied IMEX strategies inspired‌ by 66, 49‌‌. We treat the fast waves implicitly to‌ bypass the acoustic CFL‌ restriction, while keeping other‌‌ terms explicit.

Lagrange-Projection methods: During the master internship‌ of Sofia Curto, we‌ exploited implicit-explicit Lagrange-Projection (LP)‌‌ methods 57. This approach allows for a‌ natural decomposition of wave‌ speeds and was successfully‌‌ applied to the monolithic hyper-elastic model.
Implicit all-Mach‌ relaxation: We developed an‌ implicit relaxation scheme for‌‌ the simulation of compressible flows across all Mach‌ number regimes based on‌ the Jin-Xin relaxation approach.‌‌ The scheme is characterized by its simplicity and‌ effectiveness: thanks to the‌ linearity of the flux‌‌ in the relaxation system, the time-semi discrete scheme‌ can be reformulated into‌ linear decoupled elliptic equations,‌‌ resulting in the same number of unknowns as‌ the original system. To‌ ensure correct numerical diffusion‌‌ in all regimes, a convex combination of upwind‌ and centered fluxes is‌ applied. This method was‌‌ validated on non-linear elasticity models, as well as‌ gas and fluid flow‌ simulations, demonstrating accuracy in‌‌ approximating material waves across‌ different Mach regimes.
Performance: As shown in Figure‌ 13 (bottom right), the IMEX approach correctly captures‌ material waves even when using a time step‌ 1000 times larger than the explicit stability limit.‌

8.6.1 High-order ADER-DG for Advection-Diffusion and Navier-Stokes

Participants:‌ Afaf Bouharguane, Angelo Iollo, Alexis Tardieu‌.

In parallel to solid mechanics, we investigated‌ high-order schemes for fluid flows characterized by strongly‌ varying physical parameters across internal interfaces. Classical body-fitted‌ meshes, while precise, are computationally expensive to generate‌ for moving interfaces and complicate parallel partitioning.

To‌ address this, our research has focused on non-conforming‌ hierarchical meshes (quadtrees or octrees), which offer efficient‌ adaptation and parallelization. Previous work in the team‌ (Raeli, Taymans, Fondanèche) utilized Finite Difference and Finite‌ Volume schemes on such grids; however, the necessity‌ of wide stencils for polynomial reconstruction limited both‌ parallel efficiency and the achievable order of accuracy.‌ Conversely, recent ADER strategies on Chimera grids (Carlino)‌ achieved high-order temporal accuracy but suffered from precision‌ loss in mesh overlap zones.

Consequently, we developed‌ a new Discontinuous Galerkin (DG) scheme combined with‌ an ADER time integration approach. By leveraging a‌ compact stencil and an arbitrary high-order polynomial representation,‌ this method overcomes the limitations of previous hierarchical‌ solvers.

Methodology: We established the optimal performance compromise‌ by analyzing both penalization and relaxation approaches.
Validation‌ and Extension: The scheme was validated on the‌ non-linear advection-diffusion equation and successfully extended to the‌ simulation of incompressible flows via the Navier-Stokes equations.‌
Perspectives: This work lays the foundation for compact,‌ high-precision solvers on hybrid Chimera/quadtree grids, targeting the‌ realistic simulation of boundary layers in aerodynamics, such‌ as flows around airfoils and turbine blades.

Alexis‌ Tardieu has defended his PhD in December 2025‌ 43 and he published two papers.

8.7 Biological‌ Fluid Dynamics and Locomotion

Participants: Michel Bergmann,‌ Angelo Iollo, Karl Maroun.

This research‌ axis leverages high-performance computational fluid dynamics (CFD) to‌ model flow around immersed, mobile, and deformable bodies.‌ The common methodological denominator is the use of‌ penalty methods and Cartesian grid solvers to handle‌ complex fluid-structure interactions (FSI).

Data-Driven Simulation of Zebrafish‌ Locomotion.

The zebrafish is a critical model for‌ studying locomotor diseases. We developed a numerical model‌ driven by experimental data to simulate the escape‌ swim of the zebrafish eleuthero-embryo 19. The‌ solver couples the incompressible Navier–Stokes equations with Newton's‌ laws, using a level-set function to represent the‌ body implicitly. To ensure high fidelity, deformation kinematics‌ were estimated directly from experimental videos using Procrustes‌ analysis, and the morphology was reconstructed in 3D‌ by tracking Lagrangian markers. This framework allowed for‌ the accurate in silico reproduction of the stereotyped‌ escape response (C-bend, counter-bend, and cyclic swimming) and‌ enabled a study on the Cost of Transport‌ (CoT). Results indicated a linear response in transport‌ cost associated with constant energy expenditure, independent of‌ fluid viscosity. Numerical results are obtained with the‌ in-house software NaSCar.

Multiphysics Modeling of Aquatic Maneuvers.

We extended the framework‌ to multiphysics scenarios, such‌ as the self-propelled dolphin‌‌ jump (Figure 14). The model utilizes a‌ fictitious domain approach with‌ Volume Penalization (VP) for‌‌ the solid body and the Continuous Surface Force‌ (CSF) method for the‌ air-water interface 8.‌‌

The geometry, based on Lagenorhynchus obliquidens,‌ follows thunniform deformation laws‌ while global motions are‌‌ computed dynamically. This approach successfully captures the complex‌ interaction between the swimmer‌ and the free surface.‌‌ Numerical results are obtained with the in-house software‌ NaSCar.

Optimization and Control‌ of Undulatory Swimming.

In‌‌ the context of the ANR project DRAGON2, we‌ developed a framework for‌ the optimal control of‌‌ swimming 17. To mitigate the computational cost‌ of high-fidelity simulations, Sparse‌ Identification of Nonlinear Dynamics‌‌ (SINDy) was employed to generate surrogate models. These‌ models facilitated the solution‌ of control problems, including‌‌ velocity tracking via Model Predictive Control (MPC) and‌ the minimization of the‌ Cost of Transport (CoT).‌‌ The optimization yielded a "burst-and-coast" strategy, validating the‌ energy efficiency of intermittent‌ swimming behaviors. Numerical results‌‌ are obtained with the in-house software NaSCar.

Collective‌ Dynamics of Microswimmers.

In‌ collaboration with LOMA, we‌‌ investigated the behavior of artificial swimmers in confined‌ environments (Figure 15).‌ The numerical framework integrated‌‌ penalization methods with short-range contact forces (lubrication and‌ collision) to accurately simulate‌ collective dynamics 28.‌‌

The simulations successfully reproduced experimental observations, such‌ as wall accumulation and‌ the reduction of mean‌‌ velocity with increasing group size, elucidating the role‌ of boundary properties in‌ shaping wake structures. Numerical‌‌ results are obtained with the in-house software NaSCar.‌

8.8 Multi-fidelity multi-scale numerical‌ modeling of wave energy‌‌ converter farms

Participants: Beatrice Battisti, Michel Bergmann‌.

High-fidelity three-dimensional models‌ provide accurate results for‌‌ wave energy converters (WECs) simulations, but are too‌ computationally expensive. Projection-based model‌ order reduction techniques based‌‌ on Proper Orthogonal Decomposition (POD) have demonstrated effectiveness‌ in simulating single-phase flows,‌ but encounter stability challenges‌‌ with multiphase flows. Tests presented in the thesis‌ indicate that classical methods‌ bypassing full-order models are‌‌ not viable in this context. A multi-fidelity, Galerkin-free‌ model is thus proposed,‌ combining CFD for accurately‌‌ describing the WEC near-field, with POD for efficient‌ far-field wave propagation modeling.‌ The two systems are‌‌ coupled using a domain decomposition-inspired strategy (Figure 16‌), ensuring bidirectional information‌ exchange for precise flow‌‌ reconstruction and accurate representation of the floater dynamics.‌ This approach has potential‌ broader applications, serving as‌‌ an advanced far-field boundary condition in problems involving‌ wave propagation.

Numerical results on different wave conditions‌ (Figure 17) prove the versatility of the‌ coupling technique, both for in-sample reconstruction (IS, where‌ the POD basis is computed with a database‌ from the baseline condition $W 0$ ), and‌ an out-of-sample prediction (OOS, where the POD basis‌ is computed with a database from conditions $W‌ 1$ and $W 2$ ).

In the‌ OOS case, the L2 relative errors are slightly‌ higher than those in the IS case but‌ remain comparable, with all errors on the order‌ of $𝒪 ({10}^{- 4})$ .‌ The time evolution of the vertical translation of‌ the floating body and the vertical force acting‌ on it are shown in Figure 18.‌

With only slight dissipation at the‌ peaks and troughs, the IS case's temporal evolution‌ of the force and body position nearly matches‌ the one in the reference high-fidelity (HF) solution.‌ Although the overall behavior is still accurate, there‌ is a slightly larger difference between the reference‌ and the predicted solution in the OOS case.‌ In this case, there is an observable phase‌ shift and a higher degree of dissipation, which‌ may be impacted by the time step selected.‌ The phase shift is likely due to the‌ training dataset containing wave periods different from the‌ target period. In addition to strong CPU savings,‌ the computational time of the simulations passes from‌ approximately 10 hours and 30 minutes on 48‌ processors, to less than 4 hours, using only‌ 6 processors.

9 Bilateral contracts and grants with‌ industry

9.1 Bilateral contracts with industry

Participants: Michel‌ Bergmann, Angelo Iollo, Tommaso Taddei.‌

Contrat accompagnement Cifre Ingeliance (M. Truel) - du‌ 01/05/2023 au 30/04/2026
Contrat accompagnement EDF (A. Moussaddak)‌ - du 01/12/2023 au 30/11/2026

10 Partnerships and‌ cooperations

10.1 International initiatives

10.1.1 Associate Teams in‌ the framework of an Inria International Lab or‌ in the framework of an Inria International Program‌

Participants: Michel Bergmann, Alessia Del Grosso,‌ Angelo Iollo, Tommaso Taddei.

ROHM

Title:‌
Reduced Order Hybrid Models
Duration:
2023 -> 2025‌
Coordinator:
Charbel Farhat (cfarhat@stanford.edu)
Partners:
- Stanford University Stanford‌ (États-Unis)
Inria contact:
Angelo Iollo
Summary:
The ROHM‌ project is devoted to mathematical models that combine‌ partial differential equations and prior solutions of such‌ models in order to reduce the size of the problem. In particular,‌ building on the collaboration‌ previously undertaken in the‌‌ previous associated MARE team, it is intended to‌ explore two complementary aspects‌ of reduced modeling. On‌‌ the one hand, it is intended to combine‌ the notion of solution‌ mapping learned in the‌‌ solution space developed at Bordeaux with the projection‌ space partitioning technique developed‌ at Stanford. On the‌‌ other hand, it is intended to combine low-fidelity‌ models and high-fidelity models‌ in physical space or‌‌ time in order to be able to deal‌ with strongly multiscale problems.‌

SPADES

Title:
Structure-Preserving Approximations‌‌ of Dynamical systems in Engineering and Science
Duration:‌
2024 ->
Coordinator:
Benjamin‌ Sanderse (b.sanderse@cwi.nl)
Partners:
- CWI‌‌ Amsterdam (the Netherlands)
Inria contact:
Tommaso Taddei and‌ Alessia Del Grosso
Summary:‌
Model order reduction (MOR)‌‌ of parametric PDEs is a well-established field in‌ scientific computing that aims‌ to reduce the marginal‌‌ cost associated with the solution to parametric systems:‌ MOR is motivated by‌ many-query (optimization, parameter sweeps)‌‌ and real-time (interactive design, monitoring) applications, which naturally‌ arise in the field‌ of continuum mechanics. Despite‌‌ the numerous examples of applications of MOR to‌ large-scale industrial problems, the‌ practical deployment of MOR‌‌ techniques remains limited in computational fluid dynamics (CFD).‌ To address the current‌ limitations of MOR methods,‌‌ several authors have proposed structure-preserving projection techniques and‌ nonlinear data compression methods:‌ the former refer to‌‌ a class of methods that aim to preserve‌ notable properties (e.g., positivity,‌ entropy conservation) of the‌‌ solution to the underlying PDE, which are not‌ necessarily guaranteed at the‌ reduced-order level; the latter‌‌ refer to a class of methods that exploit‌ a nonlinear ansatz to‌ estimate the state field.‌‌ The objective of the Associate Team SPADES between‌ Inria Team MEMPHIS (PI:‌ Tommaso Taddei) and CWI‌‌ (PI: Benjamin Sanderse) is to devise effective structure-preserving‌ nonlinear model reduction techniques‌ for unsteady nonlinear PDEs‌‌ that arise in computational fluid dynamics (CFD). The‌ project benefits from the‌ very complementary expertise in‌‌ nonlinear approximation methods and structure-preserving reduced-order formulations of‌ the two partners, and‌ has the potential to‌‌ address the grand challenges of model reduction techniques‌ for a broad range‌ of applications in CFD.‌‌

10.2 National initiatives

ANR (national agency for research‌ funding) DRAGON2.
- Principal‌ investigators: Michel Bergmann
- Partners:‌‌ CNRS/Université de Poitiers/Inria. 27 k€+ 1 PhD.
- Summary:‌ The goal is study‌ the aquatic swimming a‌‌ several snakes using biomimetism and bioinspiration. In this‌ project, we have experimental‌ data for snake swimming,‌‌ and we are building a numerical twin to‌ compute integral quantities like‌ the efficiency. Reinforcement learning‌‌ is also considered to optimize the snake swimming.‌
ANR (national agency for‌ research funding) RedLum.‌‌
- Principal investigators:Tommaso Taddei .
- Partners: Team ACTA‌ at INRAE Rennes.
- Summary:‌ REDLUM is a joint‌‌ project between the team ACTA at INRAE Rennes‌ and the team MEMPHIS‌ at Inria Bordeaux; the‌‌ project will fund a PhD thesis in October‌ 2024: the selected student‌ will join team MEMPHIS‌‌ in Bordeaux and will‌ work closely with the other partners of the‌ project. The objective of REDLUM is to develop‌ a model reduction procedure for turbulent flows with‌ unknown boundary conditions; the ultimate goal is to‌ devise a rapid and reliable simulation tool to‌ tackle real-time data assimilation tasks in agricultural sciences.‌ The distinctive methodological feature of the approach that‌ we wish to develop is a stochastic closure‌ model to adequately approximate the dynamics of the‌ low-dimensional coherent structures of the system.
Inria Exploratory‌ Action: AM2OR (Adaptive meshes for model order‌ reduction).
- Principal investigators: Nicolas Barral (Inria team:‌ Cardamom), Tommaso Taddei. 14 k€+ 1 PhD +‌ 1 PostDoc.
- Summary: Mesh adaptation and model order‌ reduction both aim at reducing significantly the computational‌ cost of numerical simulations by taking advantage of‌ the solution's features. Model order reduction is a‌ method that builds lighter surrogate models of a‌ system's response over a range of parameters, which‌ is particularly useful in the solution of design‌ and optimization inverse problems. Reduced-order models rely on‌ a high-fidelity (e.g., finite element) approximation that should‌ be sufficiently accurate over the whole range of‌ parameters considered: in presence of structures such as‌ shocks and boundary layers, standard mesh refinement techniques‌ would lead to high-fidelity models of intractable size.‌ In this project, we propose a novel adaptive‌ procedure to simultaneously construct a high-fidelity mesh (and‌ associated discretisation) and a reduced-order model for a‌ range of parameters, with particular emphasis on inverse‌ problems in computational fluid dynamics.

11 Dissemination

11.1‌ Promoting scientific activities

11.1.1 Scientific events: organisation

Michel‌ Bergmann , Alessia Del Grosso , Angelo Iollo‌ and Tommaso taddei have organbized the ARIA Workshop‌ 2025 in Birdart, Spetember 2025

link.

Reviewer‌ - reviewing activities

All team members are reviewers‌ for the most infuencial journals in our community‌ (at least one per month for permanent reseachers):‌ Journal of Computional Physics, SIAM Journal of SCientific‌ Computing, Physics of Fluids, ...

11.1.2 Invited talks‌

Angelo Iollo : 2025 décembre. Conférencier invité à‌ l’Institute for Mathematical & Statistical Innovation, University of‌ Chicago. link
Angelo Iollo : 2025 novembre. Conférencier‌ invité au workshop Numerical Approaches for PDEs, ETNA,‌ Catane. link
Angelo Iollo : 2025 février 22.‌ Sousse. Conférencier invité à Mathematics, AI and Applications‌ 2025. link
Angelo Iollo : 2025 février 18-22‌ Hammamet, Tunisia. Cimpa Research School Control, Optimization and‌ Model Reduction In Machine Learning. 6h of short‌ course on model reduction. link
Tommaso Taddei :‌ Invited speaker at the workshop on “Scientific Machine‌ Learning: error control and analysis”, January 2025. Besançon‌ (France), Registration in bounded domains for model‌ reduction of parametric conservation laws.link
Tommaso Taddei‌ : Invited speaker at the conference Shark-FV, May‌ 2025. Minho (Portugal)Registration in bounded domains for‌ model reduction of parametric conservation laws.link
Michel‌ Bergmann : Invited speaker at the Institut Henri‌ Poincaré for the CEA-SMAI/GAMNI Mécanique de Fluides Numériques,‌ January 2025. Paris (France), From Linear to Nonlinear Interpolation: Reduced Basis‌ and Optimal Transport for‌ Data Recycling in Incompressible‌‌ Flow Simulations. link
Michel Bergmann : Invited speaker‌ at the Conférence Climath‌ : Coastal flows, extreme‌‌ waves and wave-structure interaction, November 2025. Bordeaux (France)‌, From CFD approach‌ to wave-structure interactions. link‌‌
Alessia Del Grosso : Invited speaker at the‌ workshop ETNA November 2025.‌ Catania (Italy), A‌‌ non-conservative scheme for hyperelastic materials circumventing the involution‌ constraint link
Alessia Del‌ Grosso : Invited speaker‌‌ at the workshop HyPNuT November 2025. Amiens (France)‌, A non-conservative scheme‌ for hyperelastic materials circumventing‌‌ the involution constraint link
Alessia Del Grosso :‌ Invited speaker at the‌ workshop ARIA 2025. September‌‌ 2025. Bidart (France), Collocation-based ROMs: analysis and‌ applications link
Alessia Del‌ Grosso : Invited speaker‌‌ at the workshop 3C: Challenges in Computational methods‌ for Complex environmental applications.‌ May 2025. Le Bourget-du-Lac‌‌ (France), From supersonic to low Mach number‌ flows using cell-centered finite‌ volume multi-point schemes link‌‌

11.1.3 Leadership within the scientific community

Michel Bergmann‌ is Délégué Scientifique Adjoint‌ (DSA) of the centre‌‌ Inria de l'université de Bordeaux (since July 2024).‌ He is a member‌ of inria evaluation committee,‌‌ the scientific committee of the institute of mathematics‌ in Bordeaux (IMB), is‌ member of the scientific‌‌ committee of méso-centre aquitain MCIA, and is member‌ of the CDT (commission‌ de développements technologiques) at‌‌ Inria.
Angelo Iollo is PI with Denis Sipp‌ of Onera of the‌ Chair PROVE, endowed by‌‌ a grant of Region Nouvelle Aquitaine and Onera.‌ The chair is endowed‌ with PhD grants, PostDocs‌‌ and funding for research animation. He is also‌ the PI of the‌ newly created common team‌‌ Inria-ONERA MONHADE (since 1st Novemebre 2025).

11.1.4 Scientific‌ expertise

Angelo Iollo was‌ expert for the IdEX‌‌ initiative of the University of Nice as well‌ as expert for the‌ University of Modena program‌‌ FAR.

11.2 Teaching - Supervision - Juries -‌ Educational and pedagogical outreach‌

Two members of the‌‌ team are Professor (Angelo Iollo) or Assistant Professor‌ (Afaf Bouharguane) at Université‌ de Bordeaux and have‌‌ teaching duties, which consist in courses and practical‌ exercises in numerical analysis‌ and scientific computing. Michel‌‌ Bergmann (DR) teaches around 64 hours per year‌ (courses on renewable energies‌ and practical exercises in‌‌ programming for scientific computing). Tommaso Taddei (CR) teaches‌ around 50 hours per‌ year (practical exercises in‌‌ numerical analysis and scientific computing). Alessia Del Grosso‌ (ISFP) teaches around 30‌ hours per year (mathematics‌‌ for engineering).

11.2.1 Supervision

Michel Bergmann supervises or‌ co-supervises the PhD theses‌ of Maxime Chapron, Mathias‌‌ Truel and Karl Maroun.

Angelo Iollo supervises or‌ co-supervises the PhD theses‌ of Jon Labatut, Elise‌‌ Declerk, Mathias Truel, Marc-Oliveir Potin and Alexis Valls.‌

Tommaso Taddei supervises or‌ co-supervises the PhD theses‌‌ of Jon Labatut, Abdessamad Moussaddak, Ishak Tifouti and‌ Alexis Valls.

Alessia Del‌ Grosso co-supervises the PhD‌‌ theses of Lucas Brelivet and Giorgio Musso.

11.2.2‌ Juries

Michel Bergmann was‌ the president of the‌‌ jury for the PhD‌ of Lou Guérin (Poitiers University); he further reviewed‌ the PhD thesis of Clément Caron (University of‌ Ile de la Réunion) and Céline Van Landeghem‌ (Strasbourg University). He was the President of Inria‌ selection board (jury d'admissibilité) for the recruitment of‌ CRCN (junior researchers) and ISFP at the Centre‌ Inria de l'université de Rennes, and member of‌ the selection board (jury d'admissibilité) for the recruitment‌ of DR2 (research Director, Senior Researchers).

Angelo Iollo‌ was reviewer of the PhD thesis of Raphael‌ Villiers, Université de Poitiers; reviewer of the PhD‌ thesis of Moaad Khamlich, SISSA; reviewer of the‌ HDR of Lionel Mathelin, Cnrs; reviewer of the‌ PhD theis of Erica Tamellini, Politecnico di Milano.‌

11.2.3 Specific official responsibilities in science outreach structures‌

Michel Bergmann leads the storytelling efforts for the‌ Energy research axis at Centre Inria de l'université‌ de Bordeaux.

Tommaso Taddei was a member of‌ the CUMI-R committee and of the CDT before‌ july. Since July, Alessia Del Grosso is a‌ member of the CUMI-R committee and of the‌ CDT.

11.2.4 Productions (articles, videos, podcasts, serious games,‌ ...)

Michel Bergmann and Angelo Iollo have participed‌ to a presentation article on the new Inria-Onera‌ common team MONHADE link.

Michel Bergmann :‌ video "une minute avec" link

Alessia Del Grosso‌ : video "une minute avec" link

12 Scientific‌ production

12.1 Major publications

1 articleE.E.‌ Abbate, A.A. Iollo and G.G.‌ Puppo. An all-speed relaxation scheme for gases‌ and compressible materials.Journal of Computational Physics‌3512017, 1-24HAL DOI
2 article‌N.Nicolas Barral, T.Tommaso Taddei and‌ I.Ishak Tifouti. Registration-based model reduction of‌ parameterized PDEs with spatio-parameter adaptivity.Journal of‌ Computational Physics499February 2024, 112727HAL‌DOI back to text
3 articleL.Luis‌ Benetti Ramos, O.Olivier Marquet, M.‌Michel Bergmann and A.Angelo Iollo. Fluid--solid‌ Floquet stability analysis of self-propelled heaving foils.‌Journal of Fluid Mechanics9102021, A28‌HAL DOI
4 articleM.M. Bergmann,‌ C.-H.Charles-Henri Bruneau and A.A. Iollo.‌ Enablers for robust POD models.Journal of‌ Computational Physics22822009, 516--538
5‌ articleM.M. Bergmann, J.J. Hovnanian‌ and A.A. Iollo. An accurate cartesian‌ method for incompressible flows with moving boundaries.‌Communications in Computational Physics1552014,‌ 1266--1290back to text
6 articleM.M.‌ Bergmann and A.A. Iollo. Bioinspired swimming‌ simulations.Journal of Computational Physics3232016‌, 310 - 321
7 articleM.M.‌ Bergmann and A.A. Iollo. Modeling and‌ simulation of fish-like swimming.Journal of Computational‌ Physics23022011, 329 - 348‌
8 articleM.Michel Bergmann. Numerical modeling‌ of a self propelled dolphin jump out of‌ water.Bioinspiration and Biomimetics1762022‌, 065010HAL DOIback to text
9 articleF.F. Bernard‌, A.A. Iollo‌ and G.G. Puppo‌‌. Accurate Asymptotic Preserving Boundary Conditions for Kinetic‌ Equations on Cartesian Grids‌.Journal of Scientific‌‌ Computing2015, 34
10 articleA.A.‌ Bouharguane, A.A.‌ Iollo and L.L.‌‌ Weynans. Numerical solution of the Monge--Kantorovich problem‌ by density lift-up continuation‌.ESAIM: Mathematical Modelling‌‌ and Numerical Analysis4961577November 2015‌
11 inproceedingsM.Maxime‌ Chapron, C.Christophe‌‌ Blondeau, I.Itham Salah El Din,‌ D.Denis Sipp and‌ M.Michel Bergmann.‌‌ Clustered Active Subspaces Applied to Aerodynamic Shape Optimisation‌.9th European Congress‌ on Computational Methods in‌‌ Applied Sciences and EngineeringECCOMAS 2024 - 9th‌ European Congress on Computational‌ Methods in Applied Sciences‌‌ and EngineeringLisbonne, PortugalOctober 2024HAL back‌ to text
12 inproceedings‌M.Maxime Chapron,‌‌ C.Christophe Blondeau, I.Itham Salah El‌ Din, D.Denis‌ Sipp and M.Michel‌‌ Bergmann. Clustered Active Subspaces for Aerodynamic Shape‌ Optimization.AIAA SCITECH‌ 2025 ForumAIAA SCITECH‌‌ 2025 ForumOrlando, United StatesAmerican Institute of‌ Aeronautics and AstronauticsJanuary‌ 2025, AIAA 2025-0652‌‌HAL DOI back to text
13 articleA.‌A. De Brauer,‌ A.A. Iollo and‌‌ T.T. Milcent. A Cartesian Scheme for‌ Compressible Multimaterial Models in‌ 3D.Journal of‌‌ Computational Physics3132016, 121-143back to‌ text
14 articleA.‌Angelo Iollo and T.‌‌Tommaso Taddei. Mapping of coherent structures in‌ parameterized flows by learning‌ optimal transportation with Gaussian‌‌ models.Journal of Computational Physics471111671‌2022, 111671HAL‌DOI back to text‌‌
15 articleA.Angelo Iollo and T.Tommaso‌ Taddei. Point-set registration‌ in bounded domains via‌‌ the Fokker–Planck equation.Comptes Rendus. Mathématique363‌G82025, 809-824‌HAL DOI back to‌‌ text
16 articleF.F. Luddens, M.‌M. Bergmann and L.‌L. Weynans. Enablers‌‌ for high-order level set methods in fluid mechanics‌.International Journal for‌ Numerical Methods in Fluids‌‌79December 2015, 654-675
17 articleK.‌Karl Maroun, P.‌Philippe Traoré and M.‌‌Michel Bergmann. Data-driven optimal control of undulatory‌ swimming.Physics of‌ Fluids367September‌‌ 2024HAL DOI back to text
18 article‌T.T. Meuel,‌ Y. L.Y. L.‌‌ Xiong, P.P. Fischer, C.-H.Charles-Henri‌ Bruneau, M.M.‌ Bessafi and H.H.‌‌ Kellay. Intensity of vortices: from soap bubbles‌ to hurricanes.Scientific‌ Reports3December 2013‌‌, 3455 (1-7)
19 articleG.Guillaume Ravel‌, M.Michel Bergmann‌, A.Alain Trubuil‌‌, J.Julien Deschamps, R.Romain Briandet‌ and S.Simon Labarthe‌. Inferring characteristics of‌‌ bacterial swimming in biofilm matrix from time-lapse confocal‌ laser scanning microscopy.‌eLife11June 2022‌‌HAL DOI back to text
20 articleM.‌Michele Romanelli, S.‌Samir Beneddine, I.‌‌Ivan Mary, H.‌Heloise Beaugendre, M.Michel Bergmann and D.‌Denis Sipp. Data-driven wall models for Reynolds‌ Averaged Navier-Stokes simulations.International Journal of Heat‌ and Fluid Flow99January 2023, 109097‌HAL DOI back to text back to text‌
21 articleM.Michele Romanelli, S.Samir‌ Beneddine, I.Ivan Mary, H.Héloïse‌ Beaugendre, M.Michel Bergmann and D.Denis‌ Sipp. Efficient and accurate data-driven wall modelling‌ strategy for Reynolds averaged Navier–Stokes simulations.Journal‌ of Computational Physics538October 2025, 114128‌HAL DOI back to text back to text‌
22 thesisM.Michele Romanelli. Deep Wall‌ Models for Aerodynamic Simulations.Université de Bordeaux‌December 2024HAL back to text
23 thesis‌T.Tommaso Taddei. Some contributions to model‌ reduction of parametric systems in nonlinear mechanics.‌École doctorale Mathématiques et Informatique, Université de Bordeaux‌April 2024HAL back to text
24 thesis‌I.Ishak Tifouti. Construction of reduced-order models‌ with mesh adaptation.Université de BordeauxNovember‌ 2025HAL back to text
25 articleY.‌ L.Y. L. Xiong, C.-H.Charles-Henri Bruneau‌ and H.H. Kellay. A numerical study‌ of two dimensional flows past a bluff body‌ for dilute polymer solutions.Journal of Non-Newtonian‌ Fluid Mechanics1962013, 8-26

12.2 Publications‌ of the year

International journals

26 articleB.‌Beatrice Battisti, G.Giovanni Bracco and M.‌Michel Bergmann. A Multi‐Fidelity Model for Wave‌ Energy Converters.International Journal for Numerical Methods‌ in Fluids9742025, 427-445HAL‌DOI
27 articleM.Michel Bergmann, M.‌ G.Michele Giuliano Carlino and A.Angelo Iollo‌. Model Order Reduction Using a Collocation Scheme‌ on Chimera Meshes: Addressing the Kolmogorov N-Width Barrier‌.SIAM Journal on Scientific Computing474‌August 2025, A2272-A2298HAL DOI back to‌ text
28 articleJ. F.Jean François Boudet‌, M.Michel Bergmann, A.Angello Iollo‌, A.Angelo Iollo and H.Hamid Kellay‌. Effects of boundaries for high Reynolds number‌ artificial swimmers.Scientific Reports1512025‌, 14264HAL DOIback to text
29‌ articleA.Alessia Del Grosso, W.Wasilij‌ Barsukow, R.Raphaël Loubère and P.-H.Pierre-Henri‌ Maire. An all-Mach cell-centered multi-dimensional nite volume‌ numerical scheme for the Euler equations.Computers‌ and Fluids3062026, 106951HAL DOI‌
30 articleJ.Joyce Ghantous. Using curved‌ meshes to derive a priori error estimates for‌ a linear elasticity problem with Robin boundary conditions‌.ESAIM: Mathematical Modelling and Numerical Analysis59‌5September 2025, 2613 - 2637HAL‌
31 articleG.Guillaume Ravel, T.Théo‌ Mercé, M.Michel Bergmann, A.Anja‌ Knoll-Gellida, A.Afaf Bouharguane, S.Sara‌ Al Kassir, A.Angelo Iollo and P.‌Patrick Babin. Modeling zebrafish escape swim reveals‌ maximum neuromuscular power output and efficient body movement adaptation to increased water‌ viscosity.iScience28‌32025, 112056‌‌HAL DOI
32 articleM.Michele Romanelli,‌ S.Samir Beneddine,‌ I.Ivan Mary,‌‌ H.Héloïse Beaugendre, M.Michel Bergmann and‌ D.Denis Sipp.‌ Efficient and accurate data-driven‌‌ wall modelling strategy for Reynolds averaged Navier–Stokes simulations‌.Journal of Computational‌ Physics538October 2025‌‌, 114128HAL DOI
33 articleT.Tommaso‌ Taddei, X.Xuejun‌ Xu and L.Lei‌‌ Zhang. Optimization-based model order reduction of fluid-structure‌ interaction problems.Journal‌ of Computational Physics536‌‌September 2025, 114084HAL DOI

Invited conferences‌

34 inproceedingsI.Ishak‌ Tifouti, N.Nicolas‌‌ Barral and T.Tommaso Taddei. Registration-based model‌ reduction with local reduced‌ order bases.AIAA‌‌ Scitech 2025 -Orlando (FL), United StatesJanuary‌ 2025HAL back to‌ text

International peer-reviewed conferences‌‌

35 inproceedingsJ.Jon Labatut, A.Angelo‌ Iollo, T.Tommaso‌ Taddei, J.-B.Jean-Baptiste‌‌ Chapelier, D.Denis Sipp and A.Alberto‌ Remigi. Non Linear‌ CFD Data Interpolation for‌‌ Compressible Parametric Flows Dominated by Convection.AIAA‌ SCITECH 2025 ForumAIAA‌ SCITECH 2025 ForumOrlando,‌‌ United StatesAmerican Institute of Aeronautics and Astronautics‌January 2025HAL DOI‌back to text

Conferences‌‌ without proceedings

36 inproceedingsP.Philippe Barbet,‌ M.Merveille Talla,‌ A.Alexis Valls,‌‌ M.Matthieu Petit, F.Florian Regnault,‌ N.Nicolas Salminieri,‌ L.Laurence Wallian,‌‌ V.Valentin Resseguier and G.Giovanni Stabile.‌ Efficient data assimilation of‌ unsteady 3D turbulent flows‌‌ using intrusive generative models.ADMOS 2025 -‌ XII International Conference on‌ Adaptive Modeling and Simulation‌‌Barcelone, SpainJune 2025HAL
37 inproceedingsM.‌Michel Bergmann, M.‌ G.Michele Giuliano Carlino‌‌ and A.Angelo Iollo. Collocation-based model order‌ reduction on moving chimera‌ grids for parametric problems‌‌.USNCCM18 2025 - U.S. National Congress on‌ Computational MechanicsChicago, United‌ States2025HAL
38‌‌ inproceedingsL.Lucas Brelivet, A.Alessia Del‌ Grosso, R.Raphaël‌ Loubère, R.Riccardo‌‌ Milani and F.Florent Renac. Robust and‌ accurate gradient reconstruction methods‌ for cell-centered finite volume‌‌ schemes on anisotropic meshes.SHARK-FV 2025 -‌ Sharing Higher-order Advanced Research‌ Known-how on Finite Volume‌‌Minho, PortugalJune 2025HAL
39 inproceedingsM.‌Maxime Chapron, C.‌Christophe Blondeau, I.‌‌Itham Salah El Din, D.Denis Sipp‌ and M.Michel Bergmann‌. Clustered Active Subspaces‌‌ for Aerodynamic Shape Optimization.AIAA SCITECH 2025‌ ForumAIAA SCITECH 2025‌ ForumOrlando, United States‌‌American Institute of Aeronautics and AstronauticsJanuary 2025‌, AIAA 2025-0652HAL‌DOI
40 inproceedingsJ.‌‌Jon Labatut, A.Angelo Iollo, T.‌Tommaso Taddei and J.-B.‌Jean-Baptiste Chapelier. A‌‌ Diffeomorphic Mapping Approach for Model Order Reduction in‌ Aerodynamics.SIMAI 2025‌ - Società Italiana di‌‌ Matematica Applicata e IndustrialeTrieste, Italy2025HAL‌back to text
41‌ inproceedingsR.Romain Tiphaigne‌‌, P.Philippe Barbet‌, M.Matthieu Petit, T. M.Talla‌ M. Merveille C., A.Antoine Moneyron,‌ A.Alexis Valls, F.Florian Regnault,‌ G.Giovanni Stabile, L.Laurence Wallian,‌ V.Valentin Resseguier and D.Dominique Heitz.‌ Hyperreduction in a method coupling stochastic reduced models‌ and data assimilation to predict turbulent flows.‌ETMM-15 2025 - 15th International ERCOFTAC Symposium on‌ Engineering Turbulence Modelling and MeasurementsDubrovnik, Croatia2025‌, 1-23HAL
42 inproceedingsR.Romain Tiphaigne‌, P.Philippe Barbet, M.Matthieu Petit‌, T. M.Talla M. Merveille C.,‌ A.Antoine Moneyron, A.Alexis Valls,‌ F.Florian Regnault, G.Giovanni Stabile,‌ L.Laurence Wallian, V.Valentin Resseguier and‌ D.Dominique Heitz. Stochastic Reduced-Order Model of‌ the deterministic Navier-Stokes equation for turbulent flow prediction‌.MORTech 2025 - 7th International Workshop on‌ Model Reduction TechniquesZaragoza, SpainNovember 2025HAL‌

Doctoral dissertations and habilitation theses

43 thesisA.‌Alexis Tardieu. ADER-DG approaches for the nonlinear‌ advection-diffusion : application to the incompressible Navier-Stokes equations‌.Université de BordeauxNovember 2025HAL back‌ to text

Reports & preprints

44 reportS.‌Soraya Arias, M.Michel Bergmann, F.‌Fabien Campillo, M.-A.Marie-Agnès Enard, C.‌Christian Fabre, F.Frédérick Garcia, B.‌Benjamin Guedj, E.Emmanuel Jeannot, G.‌Giovanni Neglia, D.Diane Peurichard, D.‌Daniel Racoceanu, B.Benoît Sagot and G.‌Gaëlle Tworkowski. Réflexions sur l'usage de l'IA‌ générative pour les métiers de la recherche.‌Inria2025, 1-10HAL
45 miscM.‌Michel Bergmann, U.Umberto Bosi and M.‌Martin Parisot. A two-way multi-fidelity coupling approach‌ combining Boussinesq and Navier-Stokes models.December 2025‌HAL
46 miscA.Afaf Bouharguane, A.‌Angelo Iollo and A.Alexis Tardieu. Nonlinear‌ advection-diffusion equation: ADER-DG penalty vs. relaxation schemes.‌September 2025HAL
47 miscC.Celia Caballero‌, I.Irene Gómez-Bueno, A.Alessia Del‌ Grosso, J.Julian Koellermeier and T.Tomás‌ Morales de Luna. A semi-implicit exactly fully‌ well-balanced relaxation scheme for the Shallow Water Linearized‌ Moment Equations.January 2025HAL

12.3 Cited‌ publications

48 articleP.P. Angot, C.-H.‌Charles-Henri Bruneau and P.P. Fabrie. A‌ penalization method to take into account obstacles in‌ a incompressible flow.Numerische Mathematik814‌1999, 497-520back to text
49 article‌S.Stavros Avgerinos, F.Florian Bernard,‌ A.Angelo Iollo and G.Giovanni Russo.‌ Linearly implicit all Mach number shock capturing schemes‌ for the Euler equations.Journal of Computational‌ Physics3932019, 278-312URL: https://www.sciencedirect.com/science/article/pii/S0021999119302530DOI‌back to text
50 articleJ.Joshua Barnett‌ and C.Charbel Farhat. Quadratic approximation manifold‌ for mitigating the Kolmogorov barrier in nonlinear projection-based‌ model order reduction.J. Comput. Phys.464‌2022, 111348back to text
51 articleP.P.T. Barton,‌ D.D. Drikakis,‌ E.E. Romenski and‌‌ V.V.A. Titarev. Exact and approximate solutions‌ of Riemann problems in‌ non-linear elasticity.Journal‌‌ of Computational Physics228182009, 7046-7068‌back to text
52‌ articleS.Stefano Bianchini‌‌. Hyperbolic limit of the Jin‐Xin relaxation model‌.Communications on Pure‌ and Applied Mathematics59‌‌05 2006, 688 - 753DOI back‌ to text
53 article‌A. D.A. De‌‌ Brauer, A.A. Iollo and T.T.‌ Milcent. A Cartesian‌ scheme for compressible multimaterial‌‌ models in 3D.Journal of Computational Physics‌3132016, 121-143‌URL: http://www.sciencedirect.com/science/article/pii/S0021999116000966DOI back‌‌ to text
54 articleB.B Cantwell and‌ D.D Coles.‌ An experimental study of‌‌ entrainment and transport in the turbulent near wake‌ of a circular cylinder‌.Journal of fluid‌‌ mechanics1361983, 321--374back to text‌
55 articleS.S.L.‌ Gavrilyuk, N.N.‌‌ Favrie and R.R. Saurel. Modelling wave‌ dynamics of compressible elastic‌ materials.Journal of‌‌ Computational Physics22752008, 2941-2969back‌ to text
56 book‌S.S.K. Godunov.‌‌ Elements of continuum mechanics.Nauka Moscow1978‌back to text
57‌ articleA. D.A.‌‌ Del Grosso, M. C.M. Castro Díaz‌, C.C. Chalons‌ and T. M.T.‌‌ Morales de Luna. On well-balanced implicit-explicit Lagrange-projection‌ schemes for two-layer shallow‌ water equations.Applied‌‌ Mathematics and Computation4422023, 127702URL:‌ https://www.sciencedirect.com/science/article/pii/S0096300322007706DOI back to‌ text
58 phdthesisX.‌‌X Jin. Construction d'une chaîne d'outils numériques‌ pour la conception aérodynamique‌ de pales d'éoliennes.‌‌Université de Bordeaux2014back to text
59‌ articleP. D.P.‌ D. Lax and R.‌‌ D.R. D. Richtmyer. Survey of the‌ stability of linear finite‌ difference equations.Communications‌‌ on Pure and Applied Mathematics921956‌, 267-293URL: https://onlinelibrary.wiley.com/doi/abs/10.1002/cpa.3160090206‌DOI back to text‌‌
60 articleG.G.H. Miller and P.P.‌ Colella. A Conservative‌ Three-Dimensional Eulerian Method for‌‌ Coupled Solid-Fluid Shock Capturing.Journal of Computational‌ Physics18312002‌, 26-82back to‌‌ text
61 inproceedingsC.C. Parés. Path-Conservative‌ Numerical Schemes for Nonconservative‌ Hyperbolic Systems.Hyperbolic‌‌ Problems: Theory, Numerics, ApplicationsBerlin, HeidelbergSpringer Berlin‌ Heidelberg2008, 817--824‌back to text
62‌‌ bookJ. A.J. A. Sethian. Level‌ Set Methods and Fast‌ Marching Methods.Cambridge‌‌ University Press, Cambridge, UK1999back to text‌
63 articleT.Tommaso‌ Taddei, X.Xuejun‌‌ Xu and L.Lei Zhang. A non-overlapping‌ optimization-based domain decomposition approach‌ to component-based model reduction‌‌ of incompressible flows.Journal of Computational Physics‌5092024, 113038‌back to text
64‌‌ articleT.Tommaso Taddei, X.Xuejun Xu‌ and L.Lei Zhang‌. High-order implicit Runge-Kutta‌‌ time integrators for component-based model reduction of FSI‌ problems.arXiv preprint‌ arXiv:2512.233632025back to‌‌ text back to text‌
65 articleT.Tommaso Taddei, X.Xuejun‌ Xu and L.Lei Zhang. Optimization-based model‌ order reduction of fluid-structure interaction problems.Journal‌ of Computational Physics2025, 114084back to‌ text
66 articleA.Andrea Thomann. Semi-implicit‌ relaxed finite volume schemes for hyperbolic multi-scale systems‌ of conservation laws.Journal of Computational Physics‌5392025, 114263URL: https://www.sciencedirect.com/science/article/pii/S0021999125005467DOI back‌ to text
67 incollectionS.Stefan Turek and‌ J.Jaroslav Hron. Proposal for numerical benchmarking‌ of fluid-structure interaction between an elastic object and‌ laminar incompressible flow.Fluid-structure interaction: modelling, simulation,‌ optimisationSpringer2006, 371--385back to text‌

MEMPHIS - 2025

MEMPHIS - 2025

2025﻿﻿﻿‌Activity reportProject-TeamMEMPHIS﻿‌​‌

Keywords

Computer﻿​﻿﻿ Science and Digital Science​‌﻿﻿

Other Research​​​‌ Topics and Application Domains﻿​﻿﻿

1 Team members, visitors,​​﻿﻿ external collaborators

Research Scientists​​​‌

Faculty​​​‌ Members

Post-Doctoral​​​‌ Fellows

PhD Students

Interns and Apprentices​‌﻿﻿

Administrative Assistant﻿​​﻿

External Collaborators

2 Overall objectives﻿‌​‌

2.1 Multi-physics numerical modeling﻿​​﻿

2.1.1 Reduced-order models: convergence​​​‌ between PDE models and﻿﻿﻿‌ data

2.1.2 Schemes for﻿﻿﻿‌ Hierarchical and Chimera meshes,﻿‌​‌ multi-physics and asymptotic limits﻿​​﻿

3 Research program​​​‌

3.1 Numerical models

3.2 Applications

4﻿​﻿﻿ Application domains

4.1 Energy​‌﻿﻿ conversion

4.2​​​‌ Schemes for turbulent flow﻿​﻿﻿ simulations using Octrees

4.3 Vascular﻿﻿﻿‌ flows

4.4 Fluid-structure interactions﻿​﻿﻿ using Eulerian non-linear elasticity​‌﻿﻿ models

5 Social and﻿​﻿﻿ environmental responsibility

5.1 Impact​​﻿﻿ of research results

5.1.1 Critical Infrastructure﻿﻿﻿‌ & Defense

6 Highlights﻿‌​‌ of the year

7﻿‌​‌ Latest software developments, platforms,﻿​​﻿ open data

7.1 Latest​​​‌ software developments

7.1.1 COCOFLOW﻿﻿﻿‌

7.1.2 KOPPA

7.1.3﻿‌​‌ NaSCar3D

7.1.4 NS-penal

7.1.5 HiWind

7.1.6 NEOS

8 New results

8.1​​﻿﻿ Nonlinear approximation methods based​​​‌ on coordinate transformations

8.2 Collocation-based​‌﻿﻿ reduced order models: analysis​​﻿﻿ and applications

Structure-preserving﻿​﻿﻿ cROM.

Implicit-explicit​​﻿﻿ (IMEX) schemes for cROM.​​​‌

Neural Network Correction﻿​​﻿ for Explicit Schemes.

CFD-based﻿​​﻿ reduced-order modeling of a​​​‌ turbine stage for parametric﻿﻿﻿‌ optimization.

Optimal Domain﻿​​﻿ Decomposition in Multifidelity Modeling​​​‌ for Uncertain Geometries in﻿﻿﻿‌ Fluid-Structure Interaction.

8.2.1 Component-based model reduction​​​‌ of complex systems

8.3 Reduced-Order​​​‌ Modeling with Active Subspace﻿﻿﻿‌ Closure

8.4 Output reduction:﻿‌​‌ Clustered active subspaces

8.5 Local reduction: Data-Driven​‌﻿﻿ Wall Modeling for Industrial​​﻿﻿ RANS Simulations

8.6 Numerical﻿﻿﻿‌ Schemes for Multiphysics

Non-conservative​​​‌ and Relaxation Schemes for﻿﻿﻿‌ Monolithic Elastic Modeling.

Implicit-explicit strategies﻿﻿﻿‌ for multi-scale flows.

8.6.1 High-order ADER-DG for﻿​﻿﻿ Advection-Diffusion and Navier-Stokes

8.7 Biological​​​‌ Fluid Dynamics and Locomotion﻿​﻿﻿

Data-Driven Simulation of Zebrafish​‌﻿﻿ Locomotion.

Multiphysics​​﻿﻿ Modeling of Aquatic Maneuvers.﻿​​﻿

Optimization and Control﻿﻿﻿‌ of Undulatory Swimming.

Collective​​​‌ Dynamics of Microswimmers.

8.8 Multi-fidelity multi-scale numerical﻿﻿﻿‌ modeling of wave energy﻿‌​‌ converter farms

9 Bilateral​​﻿﻿ contracts and grants with​​​‌ industry

9.1 Bilateral contracts﻿​﻿﻿ with industry

10 Partnerships and​‌﻿﻿ cooperations

10.1 International initiatives​​﻿﻿

10.1.1 Associate Teams in​​​‌ the framework of an﻿​﻿﻿ Inria International Lab or​‌﻿﻿ in the framework of​​﻿﻿ an Inria International Program​​​‌

ROHM

SPADES

10.2 National initiatives

11 Dissemination

11.1​‌﻿﻿ Promoting scientific activities

11.1.1​​﻿﻿ Scientific events: organisation

Reviewer​‌﻿﻿ - reviewing activities

11.1.2 Invited talks​​​‌

11.1.3 Leadership within the﻿​​﻿ scientific community

11.1.4 Scientific​​​‌ expertise

11.2 Teaching -﻿​​﻿ Supervision - Juries -​​​‌ Educational and pedagogical outreach﻿﻿﻿‌

11.2.1 Supervision﻿​​﻿

11.2.2​​​‌ Juries

11.2.3 Specific official responsibilities﻿​﻿﻿ in science outreach structures​‌﻿﻿

11.2.4 Productions (articles,​​﻿﻿ videos, podcasts, serious games,​​​‌ ...)

2025‌Activity reportProject-TeamMEMPHIS‌‌

Computer Science and Digital Science‌

Other Research‌ Topics and Application Domains

1 Team members, visitors, external collaborators

Research Scientists‌

Faculty‌ Members

Post-Doctoral‌ Fellows

Interns and Apprentices‌

Administrative Assistant

2 Overall objectives‌‌

2.1 Multi-physics numerical modeling

2.1.1 Reduced-order models: convergence‌ between PDE models and‌ data

2.1.2 Schemes for‌ Hierarchical and Chimera meshes,‌‌ multi-physics and asymptotic limits

3 Research program‌

4 Application domains

4.1 Energy‌ conversion

4.2‌ Schemes for turbulent flow simulations using Octrees

4.3 Vascular‌ flows

4.4 Fluid-structure interactions using Eulerian non-linear elasticity‌ models

5 Social and environmental responsibility

5.1 Impact of research results

5.1.1 Critical Infrastructure‌ & Defense

6 Highlights‌‌ of the year

7‌‌ Latest software developments, platforms, open data

7.1 Latest‌ software developments

7.1.1 COCOFLOW‌

7.1.3‌‌ NaSCar3D

8.1 Nonlinear approximation methods based‌ on coordinate transformations

8.2 Collocation-based‌ reduced order models: analysis and applications

Structure-preserving cROM.

Implicit-explicit (IMEX) schemes for cROM.‌

Neural Network Correction for Explicit Schemes.

CFD-based reduced-order modeling of a‌ turbine stage for parametric‌ optimization.

Optimal Domain Decomposition in Multifidelity Modeling‌ for Uncertain Geometries in‌ Fluid-Structure Interaction.

8.2.1 Component-based model reduction‌ of complex systems

8.3 Reduced-Order‌ Modeling with Active Subspace‌ Closure

8.4 Output reduction:‌‌ Clustered active subspaces

8.5 Local reduction: Data-Driven‌ Wall Modeling for Industrial RANS Simulations

8.6 Numerical‌ Schemes for Multiphysics

Non-conservative‌ and Relaxation Schemes for‌ Monolithic Elastic Modeling.

Implicit-explicit strategies‌ for multi-scale flows.

8.6.1 High-order ADER-DG for Advection-Diffusion and Navier-Stokes

8.7 Biological‌ Fluid Dynamics and Locomotion

Data-Driven Simulation of Zebrafish‌ Locomotion.

Multiphysics Modeling of Aquatic Maneuvers.

Optimization and Control‌ of Undulatory Swimming.

Collective‌ Dynamics of Microswimmers.

8.8 Multi-fidelity multi-scale numerical‌ modeling of wave energy‌‌ converter farms

9 Bilateral contracts and grants with‌ industry

9.1 Bilateral contracts with industry

10 Partnerships and‌ cooperations

10.1 International initiatives

10.1.1 Associate Teams in‌ the framework of an Inria International Lab or‌ in the framework of an Inria International Program‌

11.1‌ Promoting scientific activities

11.1.1 Scientific events: organisation

Reviewer‌ - reviewing activities

11.1.2 Invited talks‌

11.1.3 Leadership within the scientific community

11.1.4 Scientific‌ expertise

11.2 Teaching - Supervision - Juries -‌ Educational and pedagogical outreach‌

11.2.1 Supervision

11.2.2‌ Juries

11.2.3 Specific official responsibilities in science outreach structures‌

11.2.4 Productions (articles, videos, podcasts, serious games,‌ ...)

12 Scientific‌ production

12.1 Major publications

12.2 Publications‌ of the year

International journals

Invited conferences‌

International peer-reviewed conferences‌‌

Conferences‌‌ without proceedings

Doctoral dissertations and habilitation theses

Reports & preprints

12.3 Cited‌ publications