Development of Monte-Carlo methods for the simulation of radiative transfer: application to severe accidents
This post-doctoral subject concerns the development of Monte-Carlo ray-tracing methods for modeling radiation heat transfer in the context of severe accidents. Starting from a well-developed software framework for Monte Carlo simulation of particle transport in the context of reactor physics and radiation protection, we will seek to adapt existing methods to the problem of radiative heat transfer, in a high-performance computing framework. To do this, we will develop a hierarchy of approximations associated with radiative heat transfer that are intended to allow the validation of simplified models implemented in the context of the numerical simulation of severe accidents in nuclear reactors. Focusing on algorithm and simulation performance, this work is intended to be a "proof of principle" of the possible software mutualization around the Monte-Carlo method for particle transport on the one hand and radiative heat transfer on the other hand.
Modelling of valley winds by statistical downscaling
To model and monitor atmospheric emissions in an area with significant relief, it is essential to represent the winds at the scale of this relief. Cadarache's operational meteorological model only has a horizontal resolution of 1km, which does not allow it to resolve the orographic effects of the valley.
However, obtaining simulation results with a high resolution model requires calculation times that are still incompatible with the constraints of operational weather forecasting (6 hours of calculation on our servers for 1 hour of forecast for Cadarache in 2020). This constrains the horizontal resolution of the calculations and does not make it possible to resolve the orographic valley effects.
The object of the post-doc is therefore to develop a downscaling model applied to a 3D mesh of the valley, with a sufficient resolution to, at the same time, model the aerology of the valley and follow a pollution plume using an atmospheric dispersion model. It will be implemented through the use of an artificial neural network, the learning of which will be based on measurements of local climatology and aerology, supplemented by synthetic data using a local high-resolution model.
The candidate will work within a small, attentive and benevolent CEA team while remaining connected to university research via the Toulouse Aerology Laboratory. He will be able to both become a specialist in applied research in the meteorological field and acquire digital and scientific skills that can be used in business.
Mesh level modeling for disperse phases in diphasic and compressible flows
The frame is the leaks in secondary/tertiary heat exchangers. Nowadays with CANOP code it is possible to know state variables of fluids and the void fraction of their mixing. The interfaces are too numerous to be followed individually. The diffuse interfaces method does not allow to access to the interfaces total area. Its influence on heat and on momentum transfers between the phases added to pressure effects are of the
The tracking of disperse phase estimation will allow to access to :
- an accurate interfacial area by the knowledge of the evolution of the disperse phase population and its localization inside of the flow,
- bubble population generated,
- droplet population generated.
The proposed work consists in the development of a mesh level model for disperse phases tracking and its implementation within CANOP code in order to have a better estimation of the exchanges between the phases.
Parallel optimization of a distributed solver for fast transient fluid-structure dynamics with adaptive mesh refinement
The proposed work deals with the optimization of the distributed solver in EUROPLEXUS software (EPX, http://www-epx.cea.fr) when adaptive mesh refinement is used. It extends some previous recent work, dedicated on the one hand to the implementation of the scalable solver able to handle large topological changes in coupled systems with large displacements through a dynamic domain decomposition (for contact after impact or floating structures inside a fluid domain for instance), and on the other hand to the design of the versatile and efficient adaptive approach compatible with fluid-structure interaction with wave fronts, interfaces and structural failure. The compatibility of the adaptive framework with the distributed solver has been implemented, but it remains a significant step for parallel optimization, requesting a specific domain decomposition update strategy to design and program to retrieve the scalability levels achieved without adaptivity.
Development of processing by Artificial Intelligence of a measuring and forecasting station
This post-doctoral proposal is part of the French atomic commission (CEA) project "MultiMod'Air", which involves developing an « intelligent » prototype of air quality measurement and forecasting station within two years. The work proposal is to develop the bricks of Artificial Intelligence (AI) of the project: correction by ANN (Artificial Neuronal network) of the measurements obtained through low cost sensors, correction ANN of weather forecasts at the station level, which are simple treatments to implement. The actual research work will concern the development of a AI based pollution forecast at the station by learning from past events.