Treballem amb simulació avançada per millorar la seguretat nuclear

Descobreix com la Càtedra UNESCO de Mètodes Numèrics lidera la innovació de frontera al Sud Global

Projecte destacat

Enrere

FEMUSS

FEMUSS is an advanced computational platform that uses sub-grid scale stabilized finite element methods to deliver high-fidelity, multiscale simulations for complex engineering problems. It allows for precise analysis and optimization across fluid flow, heat transfer, and structural mechanics.

In action

Imperial Front Wing

This video presents a large-scale fluid dynamics simulation of a McLaren Formula 1 front wing, performed with our in-house code FEMUSS. The simulation employs an implicit Large Eddy Simulation (LES) approach based on the Variational Multiscale (VMS) method to capture the complex, unsteady turbulent structures around the wing. The computation was executed on the LUMI supercomputer using an MPI-based parallel implementation, showcasing FEMUSS’s performance and scalability for high-fidelity simulations in realistic aerodynamic configurations.

Tunnel

This video shows a high-fidelity fluid dynamics simulation of airflow inside a tunnel, performed with our in-house code FEMUSS. The simulation applies an implicit Large Eddy Simulation (LES) approach based on the Variational Multiscale (VMS) method to resolve complex turbulent structures and flow interactions within the tunnel. Using an MPI-based parallel implementation, the simulation enables a detailed comparison of two ventilation system configurations, highlighting differences in flow distribution, recirculation zones, and overall ventilation performance.

Integration with ADD2MAN

ADD2MAN is a high-performance computational platform developed by CIMNE and ArcerlorMittal for large-scale industrial simulations. It features scalable parallel adaptive algorithms, automatic mesh refinement, and G-code-driven analysis. ADD2MAN uses FEMUSS to provide highly-efficient and accurate solid mechanics simulations.

Integration with HP4FSW

HP4FSW is a high-performance simulation tool developed at CIMNE for analysing and optimising Friction Stir Welding (FSW) processes in industrial environments. It enables the accurate and efficient modelling of the thermal and mechanical aspects of the process by leveraging the capabilities of the FEMUSS platform for solid mechanics simulations.

Topology Optimization

FEMUSS supports topology optimization under uncertainty for large-scale continuum structures, combining sparse grid stochastic collocation to compute statistical metrics with parallel adaptive mesh refinement for efficient resolution of each stochastic node. A two-level parallel strategy (TOUU-PS2) distributes stochastic nodes across a distributed memory system and solves each node using domain decomposition with adaptive refinement. Dynamic load balancing improves efficiency by reducing processor idle time. The optimization relies on the topological derivative and level-set method, with demonstrated scalability to thousands of processors.

Features

Scalability up to thousands of cores, with load rebalancing

Stabilized Unifitted Finite Element Methods

Automatic mesh generation from imported CAD files

Both octree-based and anisotropic mesh refinement.

Usable with both standard and HPC architectures

Coupled Multiphysics Simulations.

Applications

Data-Driven and AI-based ROMs

MultiPhase and Phase-Field Models

Implicit LES Turbulence Simulation

Aeroacoustics

Large Thermo-Mechanical Solid Mechanics

Topology Optimization

Fluid Structure Interaction

Industrial Applications

Vehicle and component aerodynamics

Design and validation of friction stir welding (FSW)

Ventilation of urban and railway tunnels

Fluid-structure interaction in civil and naval engineering

Thermal and mechanical processes in materials

Additive Manufacturing

Interested?

If you want to learn more about the the capabilities of FEMUSS or how it can help your organization, send us an email!

 

Discover CIMNE’s research cluster in Solid and Fluid Simulation for Industrial Processes

Compartiu: