Perspectives and Challenges for Shape Optimisation in Marine Engineering
Matteo Diez, CNR-INM
Sala conferenze IMATI-CNR, Pavia - Tuesday, April 9, 2019 h.15:00
Abstract. In the context of marine engineering products whose performance is highly affected by shape
parameters (such as surface/underwater vehicles, sustainable-energy conversion machines such as wind
farms, wave-energy harvesting devices, etc.) computer simulations have played an increasing significant
role in the design process. Addressing real-world complex industrial applications requires high-fidelity
physics-based computational tools especially if innovative products/designs are sought after. In this
context, the simulation-based design optimisation (SBDO) paradigm has demonstrated the capability of
supporting the design decision process, not only providing large sets of design options but also exploring
operational spaces by assessing design performance for large sets of operating/environmental
conditions. In shape design, SBDO consists of three main elements: (i) a deterministic and/or stochastic
simulation tool (integrating multidisciplinary physics-based solvers possibly with UQ), (ii) an optimisation
algorithm, and (iii) a shape modification tool. The computational cost associated to solving a high-fidelity
SBDO problem still remains quite unaffordable for most problems and users, especially for large design
spaces in the quest for the global optimum. To overcome this limitation, recent approaches have been
developed and their potential demonstrated for application of interest. Specifically, design-space
dimensionality reduction methods have gained an increasing interest, along with global metamodels,
and global optimisation methods, providing the foundation for the development of modular tools for
massive computational-cost reduction in high-fidelity SBDO for shape design in marine engineering. An
overview with examples will be discussed during the seminar.
Università degli Studi di Pavia -
Via Ferrata, 5 - 27100 Pavia
Tel +39.0382.985600 - Fax +39.0382.985602