18 months post-doctoral position
Numerical methods for massively parallel simulations applied to nanoparticle processing
Institut de Mécanique et d’Ingénierie de Bordeaux (I2M), team MFN
High performance computing, fluid mechanics, numerical methods
Stéphane Glockner, Antoine Lemoine
This post-doctoral position is part of the ANR project SUPERFON which aims to improve
the processing of FON (Fluorescent Organic Nanoparticules) in supercritical flows. These
nanoparticles are produced by the SAS (Supercritical AntiSolvent) process  in a reactor with
complex geometry. To understand the physics of the reactor, especially the inhomogeneities
of the species concentrations and the saturation of the reactor leading to the formation of
crystals, the process must be simulated over a long physical time (few minutes).
We study this problem with a numerical approach with our open-source computational code Notus
which is a multiphysic code, massively parallel and oriented to the
simulation of fluid mechanics.
With the numerical methods implemented in Notus, we can reach a simulated physical time of
about few seconds in a reasonable amount of CPU time . Furthermore, the flow is confined
in a complex geometry that must be reproduced numerically. To take the geometry into
account, the Immersed Boundary method have been improved and implemented into Notus .
With these improvements, the boundary conditions can be represented with a second order of
convergence while the discretization stencil remains small.
This post-doctoral position aims to revisit the physical models and the numerical methods to
drastically shorten the computation time. Many leads have been identified: incompressible or
weakly compressible formulation, partial or total explicitation of the equations, choice of the
numerical schemes for transport, choice of the linear solvers, etc. The performance of these
leads will be evaluated on a base of selected problems.
Since the performances of the numerical methods are also related to their implementation,
a study must be performed on the hybrid parallelization of these methods (with MPI and
OpenMPI) and on some optimizations leading to the vectorization and a better management of
the processor cache.
The candidate must have a solid formation in applied mathematics and good knowledge in
fluid mechanics. In particular, the candidate must know some numerical methods to solve the
Navier-Stokes equations. He/she must master a modern version of the Fortran programming
language and have a very good experience in parallel computing (OpenMPI and OpenMP).
This post-doctoral position will start in January 2018 for 18 months. It will take place in the
MFN (Numerical Fluid Mechanics) team that belongs to the TREFLE department of the I2M
laboratory. The net salary will be 2087 € per month.
Applicants should send their CV, a recommendation letter and a copy of their PhD reviewer
report by e-mail to the contacts in the header of this document.
 C. Neurohr, A. Erriguible, S. Laugier, and P. Subra-Paternault. Challenge of the supercritical
antisolvent technique SAS to prepare cocrystal-pure powders of naproxen-nicotinamide.
Chemical Engineering Journal, 303:238–251, 2016.
 A. Erriguible, T. Fadli, and P. Subra-Paternault. A complete 3D simulation of a crystalliza-
tion process induced by supercritical CO2 to predict particle size. Computers & Chemical
Engineering, 52:1–9, 2013.
 J. Picot, S. Glockner. Discretization stencil reduction of direct forcing immersed boundary
methods on rectangular cells: the ghost node shifting method. Journal of Computational
Physics (en relecture), soumis en septembre 2017.