Methods for Constrained Optimisation of Propellers
Global market development, increasing environmental concerns and incessant rising
fuel prices demand for high-grade efficient propulsion systems. This puts high pressure
on the propeller designer to develop well-engineered and customized propeller
design in a short timeframe. The design under development is always unique and
requires individual consideration on several constraints. The objective of this research
project is to improve the academic state-of-the-art in automatic propeller
design optimisation in terms of several design constraints and optimisation procedures.
The first part of the work, presented in this thesis, contains the further development
of cavitation constraints that can be achieved by low-fidelity numerical methods, e.g.
potential methods like vortex lattice method (VLM). Attempts were made to recognize
and constrain certain types of cavities, based on sheet cavity prediction like
primary cavities that tend to evolve to erosive cavitation, rather than the total
amount of cavitation. Automated optimisations of the blade geometry were carried
out, utilizing a combination of a Reynolds-Averaged Navier-Stokes (RANS) solver
and a VLM based propeller analysis code, in a multi-objective setup, incorporating
this concept and allow for harmless cavitation.
A second part addresses the improvements of computational costs in automated
optimisation approaches. Several response surface methodologies (RSM) were investigated
to exclusively determine the propeller performance, including constraints,
in a multi-objective optimisation.
response surface methodology
propeller hull interaction
artificial neural network