This page gives an overview of my research activities.
The goal of my PhD was to model Micro PlastoHydroDynamic lubrication (MPH lubrication). This concept, which was first introduced by Mizuno and Okamoto , consists in lubricant outflow from cavities at the asperity level. Azushima  directly observed this phenomenon in plane strip drawing using a transparent die. Bech  carried out experiments on a similar experimental set-up. He studied the effect of several parameters such as the strip thickness reduction, the drawing speed, the lubricant viscosity, etc.
I used the in-house Finite Element (FE) software Metafor to reproduce Bech's experiments numerically. The model is able to predict lubricant escapes of the pocket backward and forward (see video). As in the experiments, several parametric studies were carried out on the thickness reduction effect, the slope of the lower tool and the drawing velocity, etc. These results show the model is capable of reproducing the trends observed experimentally by Bech.
My goal is now to couple the FE software Metafor to the numerical rolling model MetaLub in order to take MPH lubrication effects in cold-rolling.
The experimental setup used by Bech to highlight MPH lubrication is depicted below. It is made of a transparent upper tool and a steel lower tool having a small angle $\alpha$ respective to the horizontal axis ($\alpha$ being either 2°, 3° or 5° depending on the tests). Using a camera, Bech was able to observe and record lubricant outflow from cavities. In the various conditions he tested, Bech observed backward and/or forward lubricant escapes.
The FE numerical model developped in Metafor is able to predict backward and forward lubricant flow as shown by the video below. Moreover, parametric studies conducted numerically show the trends predicted by the model match experimental observations.
FE simulations results computed with Metafor
MPH lubrication requires to take into account fluid and solid at the same time in the simulation. To assess Metafor's ability to deal with fluid simulation. I modelled fluid flows on test cases of increasing complexity such as the square cavity and Rayleigh's step (see pictures below). To perform these simulations, I implemented a new material law in order to model the behaviour of Newtonian fluids and I took advantage of the Arbitrary Lagrangian Eulerian formalism - allowing to uncouple the motion material's motion from the mesh - which was already implemented in Metafor (see Romain Boman for the current implementation). The comparison between Metafor results and reference solutions show a good agreement between these results.
Since fluid-structure interaction simulations were required, I assessed Metafor's validity by studying a benchmark where a flexible beam bends due to the fluid flow around it. The video below shows large displacement of the beam and its impact on fluid velocity.
MetaLub is a cold-rolling model taking into account mixed lubrication regime. The main objective is to enhance the performances of rolling mills from a lubrication point of view. It means that lubricant rheology but also roll diameters and roughness, etc. can be optimized to improve stability and efficiency of the rolling tool.
The first version of this algorithm was implemented, in Fortran, by Nicolas Marsault at CEMEF. It was then optimized and implemented in C language by Romain Boman during is PhD at the University of Liège. Then, Antoine Stephany - another PhD-student - added new features such as roll-bite starvation (the amount of oil in the roll-bite is specified by the software user), large deformations of the work rolls, new lubricant rheology laws, etc.
I changed the structure of the code by means of C++ language. Thanks to the new object oriented structure, features addition are more straightforward and code modifications are more localised. It allowed me to adapt the set of equations when I implemented the coupling procedure involving MetaLub and Metafor.
This coupling procedure is functional in dry cases: analytical asperity crushing equations used in MetaLub can be replaced by FE simulations of asperity crushing. FE results, such as the relative contact area between the tool and the strip, are then sent back to MetaLub for another computation. This coupling procedure allows us to consider more realistic asperity profiles than in the analytical laws used so far.
I implemented a Graphical User Interface (GUI - see image below) with PyQt module. Numerical parameters of the model can now be defined easily, even by people who do not know programming languages. This tool was set-up in order to spread the use of the software within ArcelorMittal R&D center.
I also designed a plotter which displays intermediate results as the software is running (see figure below). This is helpful to optimise numerical parameters of the model depending on the rolling conditions that are being studied. Moreover, functionalities has been added to ease the post-processing stage. Results curves can be drawn and compare to each other in a few clicks.
 Mizuno, T. and Okamoto, M. Effects of lubricant viscosity at pressure and sliding velocity on lubricating conditions in the compression-friction test on sheet metals. Journal of Lubrication Technology, 104: 53–59, 1982.
 Azushima, A. Experimental confirmation of the micro-plasto-hydrodynamic lubrication mechanism at the interface between work-piece and forming die. Journal of the Japan Society for Technology of Plasticity, 30: 1631–1638, 1989.
 Bech, J., Bay, N., and Eriksen, M. Entrapment and escape of liquid lubricant in metal forming. Wear, 232(2): 134–139, 1999.