The demonstration of safety and the extension of the lifetime of complex industrial devices (nuclear...) are based on the periodic non-destructive testing (NDT) of welded parts. When there are thick welds (30 to 70 mm), in austenitic stainless steel, the ultrasonic method for defect detection is the only one possible. It is however complex because the heterogeneous and anisotropic nature of these thick multi-pass welds induces strong perturbations in the propagation of the acoustic beam which distort the diagnostic. The best (non-destructive) solution to overcome this difficulty is obtained by modelling the ultrasonic propagation, but this requires the detailed description of the real crystalline structure of the weld. The current 2D weld models, except LMA’s work, provide either a simplified description of the crystallographic growth, based on a symmetry assumption, or a more realistic description, but at the cost of high instrumentation and computation time. Moreover, no model exists for a weld made in position, when the solidification is also governed by gravity. The objective of the project is to produce a realistic 3D model for welds made with GTAW process in all positions, from minimalist input data (those given by the DMOS) and with a calculation speed compatible with industrial needs. Gravity induces inclinations of the texture not only in the direction perpendicular to the weld, but also in the welding direction. The transition from 2D to 3D is therefore not a simple evolution or adaptation of MINA 2D, because the gap is very important. The study will be progressive: some mock-ups will be manufactured for a narrow chamfer (U-type) which allow a stacking of a single pass per layer, and open chamfer (V-type) geometry, in vertical-up and horizontal groove welding position. Specific instrumentations (embedded camera, optical microscopy, EBSD) will help us to understand the solidification kinetics and the grain growth, and then to create the model, the challenge being linked to the various length scales present (weld, grain, dendrite). The objective is to determine a link between the pool shape (gravity, welding energy, ...), the thermal gradient (part temperature, chamfer, preheating, ...), and the crystal growth (crystal competition, ...). The orientation of the grains will be ultimately calculated from information voluntarily restricted to the welding notebook which describes the welding procedure (geometry of the chamfer, sequence of passes, etc.), to be in adequacy with the industrial practice, which cannot afford to instrument each welding carried out in a complex way. The micrographs simulated by the model will be compared to the real micrographs and will thus allow to validate it. A second validation will also be sought by comparing the ultrasonic propagation predictions obtained by associating the MINA 3D model with a 3D ultrasound propagation model, with experimental data. The prediction of the deviations and divisions of the ultrasonic beam will then be mastered, bringing a significant improvement of the ultrasonic testing. The MINA 3D project perfectly fits with the research axis B.4. One innovation concerns the increase in knowledge of the material, but the main innovation is in the application, and therefore in the consequent improvement of the potential of NDT by ultrasound. The 6 partners of the project are the best French specialists in the field and used to work together.

Understanding sediment transport in rivers, lakes and along the ocean floor is key to sustainable management of open water bodies and aquatic ecosystems. Prominent processes are river morphodynamics, turbidity currents, and tsunamis running up a beach. Predicting and managing these processes requires in-depth knowledge of the rheology to describe macroscopic properties of the fluid-sediment mixture. However, the constitutive laws to describe these processes have so far mostly been based on studies of dense suspensions of neutrally buoyant particles in either highly viscous shearing flows or at much larger flow rates where inertial effects play the dominant role. The transition between the two regimes, however, has not been investigated in a systematic manner yet, and, hence, remains only poorly understood. This may be problematic for the predictive modeling of situations that are more relevant for engineering practices and natural flows involving sediment transport. This transitional regime will be the focus of the present study and our objectives are twofold: First, the French and German partners aim to conduct a joint complementary campaign of state-of-the art sediment transport experiments and numerical simulations, respectively. The campaign will yield highly-resolved data of laminar pressure-driven shearing flows across an idealized sediment bed for a wide range of Stokes numbers as the ratio of competing inertial and viscous effects. In a second step, these data will be used to improve existing two-phase modeling approaches that have become popular for macroscopic sediment transport models.

BlooDrop aims to develop a biomedical test for auto-diagnosis and follow up of dyslipidaemia and quantitative and qualitative haemoglobin abnormalities, by using single human blood drops and a smartphone application able to decipher an image and calculate blood biological parameters. This test will be cheap, easy and accessible in any context and location. BlooDrop aims: 1) to understand processes that govern the drying of human blood drops on glass microscope plates and to correlate them to classical biochemical analysis; 2) to predict blood components segregation in a dried drop; 3) to develop a complete automated processing of drop patterns images via a smartphone application. BlooDrop is built around one administrative and two scientific work-packages (WP1 & WP2). WP1 aims to perform relevant experiments of drying human blood drops of known donors with known biological parameters (blood lipids and haemoglobin values). Blood from healthy subjects and patients (affected with dyslipidaemia, paludism or anaemia) will be used to create a collection of blood dried drop images. In parallel, ex vivo experiments of in vitro lipids or haemoglobin content modulation will be also performed with healthy donors’ blood. In WP2, using this database and our image processing skills, we will establish a correlation between morphological and biological parameters. Then, such correlation will be applied on anonymous images to find and calculate biological parameters and to check the reverse engineering approach. The development of a smartphone application is also included in WP2. Our approach is successful because: 1) it relies only on data-driven models main parameters extraction; 2) to obtain blood drop patterns, we use a scientific engineering driven approach; 3) the skills of clinicians and medical biologists is used to perform the experimental campaign under biomedical strict protocols. Preliminary results from the ex vivo approach have already shown the specificity and the feasibility of our study.

Macroscale diffusion effects are well described by Navier-Stokes equations. In meso- and micropores, however, wall effects control gas flows and thus often catalytic reactions. Under dilute conditions, additional mass transport effects occur (thermal creep, Soret and Dufour effects) which are summarized as temperature-driven mass flow (TDF). TDF, which increases with temperature gradient may have a significant impact on catalytic reactions. These differences are determined by several parameters, (heat of reaction, thermal conductivity and dimensions). However, in reactor models, TDF was ignored or, in rare cases neglected. There is no empirical study showing quantification of TDF for catalytic reactions. We aim to fill this gap and evaluate conditions requiring TDF model use for exothermic reactions such as hydrogenations. It will be shown whether TDF can improve the model predictions, and whether mass transfer limitations in catalytic reactors can be reduced by superimposed TDF. We aim to establish a model building block for the assignment of thermal transport effects based on a scale bridging of TDF effects from the mesopore level macropore and from pore level to bulk flow. To relate thermal effects to structural and physical properties, experimental studies start with single channels, progress to multiple parallel ones (geometrically well-defined membrane), and to mesoporous membranes for catalytic reactions. As complexity increases, the individual contributions of thermal effects can be gradually revealed. TDF effects will be studied in individual channels and transferred to parallelized channel networks and mesoporous membranes impregnated with a Pt catalyst and used for hydrogenation reactions, accompanied by operando NMR measurements of concentration and temperature profiles. These non-invasive reactor- scale measurements will be used to cross-validate simulations of the effects of thermally induced mass transport in porous materials on catalytic reactions.

The SNIP Project (Numerical Simulation of Impact in Porous Materials) is proposed by researchers from AMU (Aix-Marseille University) and from the Direction des Applications Militaires of the Centre à l’énergie Atomique et aux énergies alternatives (CEA-DAM) CESTA (Centre d’Etude Scientifiques et Techniques d’Aquitaine). These researchers are specialists in mathematical and numerical modeling in fluid and solid mechanics. The aim of this project is to understand the behaviour of porous materials under impacts. During the impacts, shock waves are present accompanied with high pressures and temperatures. Many phenoma have to be taken into account such as compaction of the porous materials, plasticity, damag. Interaction with a surrounding fluid or a fluid in the pores have also to be taken into account. These problems are crucial in astrophysics (asteroid impacts), for military applications (bunker buster simulation, shock detonation transition in solid explosives), in petrol industry (fracturation of rocks ), and in civil industry (security of chemical facilities). For the modeling of porous materials, two approches can be found in the litterature. From one side, there are models obtained by the homogenization technique. In this approach the elasto-plastic matrix is usually considered as incompressible, and the dynamic behaviour is rarely taken into account. From the other side, there are models developped for multi-component solid explosives. These models take into account the phase compressibility but are, most of the time, unable to take into account the shear effects in solids. The models are thus purely hydrodynamical. Our aim is the developement of a mathematical model for porous materials with a multiphase flow approach. The multiphase flow models have shown their ability for the simulation of fluid mixtures, fluid - solid interfaces and the treatment of various physical effects (spallation, phase transition, detonation waves, capillarity effects). The new model must be able to describe the compaction process in the porous solids, take into account the compressibility of all the phases with both closed and open porosity. The model and the corresponding numerical methods will be able to recover classical homogenization results, mixture Hugoniot curves and quasi-static compaction experiments. The model will also be able to take into account micro-inertia effects. Damage effects will be then studied. The numerical results will be compared with impact experiments from the litterature and those performed in the CEA DAM. The comparison with experiments (plate impact, ballistic impact) will define the validity domain of the model and will help to determine the other physical effects to deal with (hardening, phase transition). The influence of micro-inertia effects on the shock wave structure will also be studied.