GRANIT – Research Group on Numerical Applications in Engineering and Technology
The mission of the GRANIT research group is to develop and apply high-efficiency computational methods to problems related to fluid and solid dynamics on a massive scale.
Research projects focus on advanced modeling methods based on finite elements-volumes and machine learning, along with stochastic methods for quantifying uncertainties. The methods that are developed are applied to a variety of complex engineering problems, including flood simulation, dam stability, 3D printing, the performance of hydraulic machines and external aerodynamics.
Our research is focused of the following areas:
- High-performance parallel computing using CPU (central processing unit) and GPU (graphic processing unit) clusters;
- Analysis of the propagation of uncertainties using numerical solvers;
- Optimization of additive manufacturing processes;
- Stochastic optimization of the parameters of rockfill dams;
- Stochastic analysis of the outflows of dam failures;
- Establishment of probabilistic flood maps;
- New non-intrusive stochastic model reduction approaches to high-dimension problems (polynomial chaos, B-splines, deep networks);
- Three-dimensional modeling of cracks in solids;
- Deep learning algorithms.