Computational design and optimization of marine systems
Development and application of computational methodologies for the design and optimization of marine shapes and systems. Research activities include geometric parametrization, integration of numerical models with optimization algorithms, and automated exploration of design configurations, with applications to hull forms, appendages and marine vehicles.
Design-space learning and dimensionality reduction
Development of methods to identify compact and informative representations of high-dimensional design spaces. Particular attention is devoted to latent representations associated with geometric and physical variability and to their use for improving the efficiency of shape exploration and optimization.
Surrogate and multi-fidelity modelling
Development of efficient models for combining information from simulations with different computational costs and levels of accuracy. Surrogate and multi-fidelity methods are used to reduce the computational cost of analysis, design-space exploration, uncertainty quantification and optimization.
Data-driven and machine-learning methods for engineering design
Development and application of machine-learning methods to support the analysis and design of complex engineering systems, with particular emphasis on physics-aware representations, dimensionality reduction, dynamical-system modelling and the integration of data-driven and physics-based models.
Robust design and optimization under uncertainty
Development of design methodologies accounting for variability in operating conditions, environmental conditions and numerical models, with the aim of identifying efficient solutions that remain robust under realistic engineering operating scenarios.
