Gustavo Celis

L'CADAME

The applications of the high performance computing (HPC) and machine learning in the engineering laboratory (L'CADAME) is part of the COPPE, at Federal University of Rio de Janeiro (UFRJ). After many years of experience in industrial research projects, especially for the pertoleum industry in Brazil, the lack of a specilized group working on advanced numerical and computational techniques was evident. The goal of the L'CADAME is to work and grow in this direction. L'CADAME is mainly focused on developing advanced numerical techniques and modeling complicated phenomena (such as inorganic scaling) for flow assurance applications. Our current projects are related to developing a new unconditionally hyperbolic "Slug Capturing" method to solve compositional flows (n-phase). The code developed for this purpose could handle as many phases, mixtures and components which is necessary and the whole process will be accelerated using the GPU platform. To simulate the thermodynamical behaviour of the multi-phase/multi-component flow, a new PVT solver is also under development. This solver can handle 3-phase settings (liquid-CO2 liquid-gas), will be data aware by using machine learning to re-calibrate the coefficients on fly, and is boosted using GPU computation.