PhD opportunities

Computational Platform for Additive Manufacturing by Multiscale Method and Model Reduction Technique.

Thesis proposal

Area of expertiseComputational mechanics and Materials
Doctoral SchoolSFA - Sciences Fondamentales et Appliques
SupervisorM. Yancheng ZHANG
Co-supervisorM. Michel BELLET
Research unitCentre for material forming
Contact
http://www.cemef.mines-paristech.fr/sections/formations/doctorats/pour-postuler/these-cemef-8926
KeywordsAdditive manufacturing, selective laser melting process, thermo-mechanical evolution, multi-scale coupling, finite-element method, model reduction
AbstractThe CEMEF laboratory has developed two approaches for the simulation of SLM additive manufacturing processes during the past years. A meso- approach dedicated to the simulation of elementary track formation and a macro- approach at the part scale (see Figure). The two methods are part of the C++ library CimLib developed in the laboratory. The numerical developments done in the AM_MULTI project will extend the capacity of this library and specifically its application to AM simulation.
Three main tasks will be included in this project:

Multiscale model coupling meso- and macro- scale models. The useful information extracted from the developed meso-scale model can be considered as an input of the macro-scale model. Conversely, during the construction process, the thermal and mechanical boundary conditions to be applied in the meso-scale model can be extracted from the macro-scale model to get more accurate local simulation results.
Extension of the macro-scale model capacities: inherent strain technique [4]. In the macro-scale model, the deposition of material and laser energy by fractions of layer, layer by layer and by super layer (several layers at a time) is now available. However, the non-linear mechanical resolution is still too costly for parts having very complex shape. The inherent strain method is proposed as a candidate for an efficient macro-scale model, as it could be run through linear resolutions only: calculation time could be reduced by a factor 5 or more.
Full-order/reduced-order model for parametric studies. If the simulated system is huge, the above proposed strategies are still costly for parametric studies to define optimized process windows in industry. Based on the authors expertise [5], the hybrid full-order/reduced-order model [6] is proposed for modelling the selective laser melting process in the macro-scale in the frame of level set and mesh adaptation.

The ambition of the present project is to increase the visibility of CEMEF and MINES ParisTech in the field, and take leadership with respect to potential competitors, by extending our software capacities.

The PhD student will also benefit from advanced teaching and expertise of different advisors in scientific computing, and programming in C++ language. In addition, he will have access to lectures and team experiences in fluid and solid mechanics, material science and digital metallurgy. These capacities will provide opportunities to develop future activities in various R&D sectors in industry. An alternative perspective could be to start an academic scientific career.
ProfileEngineer / Master student in the field of computational mechanics, or applied mathematics. Student attracted by problematic linked with the modelling and simulation of physical phenomena.
FundingFinancement d'un Etablissement d'enseignement suprieur
PartnershipEnviron 1650 net / mois