A Generalized Potential/RANS Interactive Method for the Prediction of Propulsor Performance

Abstract

This article introduces a generalized potential/Reynolds-averaged Navier-Stokes (RANS) interactive method for the prediction of propulsor performance with an emphasis on the calculation of the steady (axisymmetric) or the time-averaged (non-axisymmetric) effective wake fields. In the RANS solver, the propeller boundary is represented by local source terms. Therefore, this method can be used to efficiently handle different types of propulsor-related multicomponent interaction problems. The scheme is first validated by an open propeller application with a uniform inflow and then applied to a vortical-inflow-propeller interaction case, a contra-rotating propeller case, as well as a hull-rudder-propeller interaction case. Results are compared with RANS (with the propeller modeled as wall boundaries) and experiments. 1. Introduction Predictions of propeller efficiency, propeller forces, and propeller cavitation patterns are of great importance in maritime and offshore industry. These predictions are directly related to fuel efficiency, structural vibration analysis, and noise control. To accurately evaluate the propeller performance, it is essential to consider the mutual influence between the ship hull and the propeller, between different propellers if multiple propellers are used, and between different rows of a propeller if multirow propulsion is employed. To predict the flow field around multiple bodies, finite volume methods are commonly used. In these methods, the fluid domain is usually split into one static domain and one (or several) revolving domain(s) and a mixing plane condition is applied to the sliding interfaces between different domains (Zhuang et al. 2004; Sanchez-Caja et al. 2009). The mixing plane condition circumferentially averages certain fluid quantities on each side of the interface and uses these averaged quantities as the boundary condition to the other side. This treatment neglects the unsteady component of the interaction effect between different bodies (between ship hull/propeller or between different rows of the propeller) so that the problem can be solved as steady. The computational cost of this method can range from several hours to several days depending on the size of the mesh, the type of the propeller, and the computing power (Kinnas et al. 2015). In fact, when the mixing plane condition is applied to the sliding interfaces, the Reynolds-averaged Navier-Stokes method (RANS) actually solves for the mean performance of the propeller. On the other hand, if the mixing plane condition is not applied, RANS can be solved as unsteady, but the computational cost will be much higher.