THE IMPORTANCE OF A NON-DETERMINISTIC DESIGN OPTIMIZATION FOR PREDICTING REAL-LIFE PROPELLER PERFORMANCES

摘要

The manufacturing of propellers comes with geometrical variability that lies within predefined tolerances decreed by standardized accuracy classes. Although controlled by these tolerances classes, the variation in the design might result in a degradation of the expected propeller behaviour. In a classical deterministic design optimization, engineers improve a digital model, without taking into account these variations. Hence a new type of optimization method is recommended whereby the impact of the manufacturing variability, represented by statistical moments, is minimized. This paper presents a robust non-deterministic optimization of a ducted marine propeller mounted on an inland vessel. In this test case, the statistical moments of the propeller efficiency are optimized while axial thrust is constrained and cavitation occurrence is considered. The manufacturing uncertainties are derived from the ISO geometrical tolerances S-class. Eventually, a robust optimum is compared with a deterministic optimum in order to underline the benefits of the non-deterministic design methodology.