The report outlines the procedure and results of response surface-based propeller optimisation. This work continues the propeller design studies presented in VIRTUE D4.5.1b. Specifically, the performed extensive exploration of the design space in VIRTUE D4.5.1b allowed to derive response surfaces for each design objective/constraint. The response surface models have been composed of first order, twoway interactions and pure quadratic terms. The response surface-based optimisation has turned out to be a promising approach for a propeller optimisation problem. The ability to instantly evaluate design alternatives during the optimisation allowed to use optimisation algorithms that are less time efficient but effective at thorough exploration of the solution space. Specifically, a simulated annealing algorithm has been successfully used as a global optimiser. A substantial improvement of the initial design from the efficiency, cavitation performance and structural strength points of view has been achieved. That is, the efficiency has been risen by 1.26%, the likelihood of the erosive sheet cavitation shedding (expressed in terms of the cavity length growth) has reduced by 49%, the likelihood of mid chord cavitation on the suction side has been decreased by ca. 30%, the cavity volume has dropped ca. two orders of magnitude in amount and twice in period, and in contrast to the initial design, the constraint on the minimum thickness has been satisfied and the new design exhibits more even load distribution over all four blades. Hence, a response surface-based propeller optimisation represents a strong alternative to fist-principles optimisation, shortening the delivery of results from weeks to days.
Response surface-based propeller optimisation - EU-FP6 VIRTUE, Deliverable D4.5.2
Francesco Salvatore
2009
Abstract
The report outlines the procedure and results of response surface-based propeller optimisation. This work continues the propeller design studies presented in VIRTUE D4.5.1b. Specifically, the performed extensive exploration of the design space in VIRTUE D4.5.1b allowed to derive response surfaces for each design objective/constraint. The response surface models have been composed of first order, twoway interactions and pure quadratic terms. The response surface-based optimisation has turned out to be a promising approach for a propeller optimisation problem. The ability to instantly evaluate design alternatives during the optimisation allowed to use optimisation algorithms that are less time efficient but effective at thorough exploration of the solution space. Specifically, a simulated annealing algorithm has been successfully used as a global optimiser. A substantial improvement of the initial design from the efficiency, cavitation performance and structural strength points of view has been achieved. That is, the efficiency has been risen by 1.26%, the likelihood of the erosive sheet cavitation shedding (expressed in terms of the cavity length growth) has reduced by 49%, the likelihood of mid chord cavitation on the suction side has been decreased by ca. 30%, the cavity volume has dropped ca. two orders of magnitude in amount and twice in period, and in contrast to the initial design, the constraint on the minimum thickness has been satisfied and the new design exhibits more even load distribution over all four blades. Hence, a response surface-based propeller optimisation represents a strong alternative to fist-principles optimisation, shortening the delivery of results from weeks to days.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


