Bonaventura

Personal website

Research

I work at the intersection of high-fidelity numerical simulation and machine learning, applied to offshore wind and wave energy systems. The central challenge I address reconciles physics-based simulation of floating offshore energy systems and physics based models. I am developing surrogate frameworks that can close this gap for fast, GPU-native, and physically consistent simulations.


Flagship project

IM-POWER (2024–2026)

Marie Skłodowska-Curie Postdoctoral Fellowship · Uppsala University · PI In cooperation with Hexicon AB, Stockholm

IM-POWER develops an integrated numerical model for the power output of floating offshore wind farms under real sea conditions, including storm events, with full grid connection.

The core technical contribution is a physics-aware neural surrogate for hydrodynamic and aerodynamic load prediction on floating platforms. A feedforward neural network learns the nonlinear hydrodynamic forces from high-fidelity SPH simulation data, and is embedded within a time-domain integration loop. At farm scale, loads across all turbines are predicted in a single parallel GPU call, which will enable simulation sub-linear scaling up until 100 turbines.

Key results (in preparation):

HPC resources: so far 51,400 node-hours on MareNostrum5 (EuroHPC); 200,000 GPU-hours on Pelle UPPMAX


Offshore wind & wave energy

High-fidelity numerical modeling of floating offshore wind turbines (FOWT) and wave energy converters (WEC), using SPH methods (DualSPHysics) coupled with multi-body solvers (Project Chrono).


Open-Source Contributions


Home · Profile · Research · Publications · Experience · Contact