By Evert-Jan Foeth (MARIN)
Within SATURN the automatic propeller design and optimization suite used at MARIN for day-to-day engineering was updated to estimate the RNL of the cavitation propeller. This suite uses bio-inspired multi-objective optimization algorithms able to balance competing goals while satisfying a range of stringent constraints; successive generations of designs are generated, improving with each iteration. The first goal was maximization of the propeller efficiency, an indicator of the expected power use and, indirectly, green-house gas emissions.
Figure 1: Computed radiated noise level spectrum of a cavitating propeller of a ferry at 20 knots; some of calculated contributions are too low within the scale of this plot.
The results from each propeller analysis were coupled to a series of empirical relations estimating the sound of the sheet and tip vortex cavitation, plus, the tonal contribution at the blade rate frequencies. The total RNL was then corrected by the weighting functions for various mammalian hearing groups from (Southall et al., 2019). The second goal was set to the minimization of the highest weighted RNL levels.
A number of constraint were set to the design process following best-practice engineering guidelines, primarily aimed at obtaining low cavitation erosion risk, satisfying class regulations of blade strength and avoiding flow separation. The method was applied to a twin screw ferry with controllable pitch propellers. To ensure a robust design, each propeller was analyzed at the nominal and maximum rating of both engines per shaft, as well as the nominal rating using only one engine per shaft. These conditions force the propeller to operate at both over and underloading conditions where cavitation may appear on either side of the blade, limiting the viable design space. With a constant rate of revolutions, the blades of the controllable pitch propellers were automatically rotated such that all these conditions were met for any geometry.
Figure 2: Pareto fronts of URN versus efficiency for 2 scenarios of 4-bladed CPPs.
Several design scenarios were run. Figure 2 shows the results of optimization procedure whereby each point represents a fully viable propeller design using stock NACA hydrofoils and using a custom hydrofoil section. For all scenarios a trade-off between URN and efficiency was found, for the NACA sections of about 3dB per percent of efficiency. It was also observed that the propellers could not be mode more quiet without violating cavitation erosion constraints. These custom sections considerably improve the tailor-made propeller design although at an increase of the overall complexity of the design. The exercise showed that at modern design method are well able to handle complicated design scenarios with many variables balancing multiple goals and constraints at once.
.Figure 3 Computed radiated noise level spectra of three cavitating propeller of a ferry at 18 knots
Two propellers from each scenario that had no efficiency loss compared the reference propeller were selected and analyzed at the cruising speed of 18 knots. Figure 2 shows the RNL of the contribution of cavitation of the (two) reference propellers—showing reasonable agreement with the RNL measured for the vessel, as well as the RNL of the two optimized designs. The propeller with stock sections showed a decrease of at least 6dB in RNL over the entire frequency range, while the propeller with custom sections had no appreciable contribution of the cavitation to the RNL at all, leaving only the contribution of our proxy for the non-cavitation and machinery sound. Admittedly, the optimizer was not restricted by any weight constraints and returned a blade design which was twice has heavy as the reference propeller far exceeding design limits of the hub.
Do silent ships perhaps require composite propellers?