By David Bellèvre, Bassin d'Essai des Carènes (French Ministry of Defense), Val de Reuil, France

View the pdf of this Supplement

The Bassin d'Essais des Carènes in Val de Reuil, France

The "Bassin d'Essais des Carènes" (BEC) is a large facility, originally built in the early 1900s in Paris. It was designed to study the resistance of ship hulls so that the power needed to propel the ships could be estimated. Over the years, improved testing devices and larger tanks were developed at the site. In the early 1990s, the facility, operated by the French Ministry of Defense, moved to Val de Reuil, about 100km from Paris. A 550m long towing tank with two test sections is the centerpiece of the new site, and a water tunnel for open water propeller tests is also available. The BEC serves as the primary propeller designer for the French Navy and for selected civilian clients who are developing research or prototype ships.

In addition to the experimental operation at BEC, numerical studies play an important role, representing about 40% of the engineering activity. In-house CFD codes have been in use since the early 1990s, and FLUENT has been used for about ten years as well. The typical simulation process begins with model building using PALAOS, BEC's in-house preprocessor. The simulation is performed with either an in-house solver or FLUENT, and postprocessing is done by another of BEC's in-house codes. The first of BEC's in-house solvers were potential codes, but they have adopted the Navier-Stokes methodology since then. The steady-state runs of the past have gradually been replaced by unsteady simulations, when needed, and the use of CFD results for stress and displacement calculations has been developed as well.

A major area of focus at BEC is propeller design. The bulk of the calculations performed to date have been in the non-cavitating regime. The quality of the mesh is considered a key factor in the success of any given simulation, and the experience base within the engineering staff has proved to be extremely valuable. Only hexahedral meshes are used, and an attempt is made to maintain a consistent topology from one project to the next. The grids are refined in areas where large gradients are expected, such as the blade root and tip (where vortices form). The near-wall spacing required to resolve the turbulent boundary layer is an important parameter as well. To make the CFD work proceed in an efficient manner, engineers often try to use meshes of the same size when studying a given propeller under a number of operating conditions, or when studying a group of propellers with small geometric differences. In this manner, an earlier solution can be used as a starting guess, reducing the total time needed to reach convergence.

CFD is used for both design and validation work, and the accuracy of the mesh and boundary conditions are chosen accordingly.

DESIGN: During the design process, numerous computations are made, so run times have to be short. Propellers are initially designed in open-water (stationary mode) with a grid size (for a single blade) of about 200,000 cells, which corresponds to approximately 30 cells in the chord-wise direction and 40 cells in the radial direction.

VALIDATION IN OPEN-WATER: To validate the final design and obtain accurate performance characteristics, the mesh is refined and the propeller rotation is taken into account. Single-blade models with large grids of about 900,000 cells are used, with 50 and 60 cells in the chord-wise and radial directions, respectively. The number of cells in the boundary layer is increased, and mesh independent solutions are ensured.

VALIDATION IN INCLINATION (UNSTEADY MODE): In the final set of tests, the entire propeller is modeled, so a coarser mesh (150,000 to 200,000 cells) is used for each blade. The inlet boundary conditions and other geometric details are adjusted to correspond to a more realistic environment for the propeller.

Once the propeller blade has undergone this series of tests, modifications to components like the hub (with cylindrical, conical, spherical, or nose cone shapes) and nozzle can be introduced using PALAOS, and revised solutions of the modified geometries run.

In one recent study, a propeller was designed at BEC that was intended to replace propellers on the P400 patrol boats used by the French Naval Forces. A one-seventh model of the propeller was built, and CFD predictions of thrust (K_T) and torque (K_Q) coefficients, calculated by integration of pressure loading and viscous stresses on the blades, were compared with experimental test results. Very good agreement was obtained, despite the fact that the 900,000 cell CFD model was not a perfectly accurate representation of the test conditions.

Table 1: Thrust and torque coefficients for a replacement propeller for the P400 patrol boat

Propeller mounted on a pod; surface pressure contours and tip vortices demonstrated with a helicity iso-surface; the axial velocity is shown in the background; the tip vortex is reduced for this propeller design because of a more advanced control of the flow

Simple propeller in open water condition; pressure contours on the blade surfaces, and tip vortices illustrated using an iso-surface of helicity

In another example, a propeller with a complex hub shape was mounted on a pod and tested at a frigate ship design point. The 5m diameter propeller consisted of five blades and was operated at 200rpm at an advance ratio, J (free stream speed/tip speed), of 0.926. To assess the role of the pod strut in reducing the thrust, CFD models at 1/20 scale were developed with and without this component, while open water experiments were run for the pod-mounted propeller only. A single blade model of 280,000 cells was created for the unmounted propeller. The CFD results for thrust coefficient for the mounted propeller were found to be in good agreement with experiment. The error is somewhat larger than for the P400 propeller studies. This may have been due to the relative grid densities used for the different cases (280,000 for a single blade for the frigate case, compared to 900,000 for a single blade for the P400 case). When compared to the predicted thrust coefficient without the pod mounting, the CFD results show that the pod drag reduces the thrust by 40%, which brings about a corresponding decrease in efficiency.

Table 2: Thrust coefficient predictions and measurement, showing the effect of pod mounting

The computed thrust coefficient, K_T, as a function of the number of nodes (per blade) in the CFD model

Many other propellers have been studied at BEC over the years. Classic propellers have been developed for French navy frigates and multi-stage propulsors have been developed for submarines. Usually, the difference between numerical and experimental test results is 1 - 2% for classical propellers (the numerical approach often overestimates the drag) and under 5% for complex propellers (which may include a mounting pod or be a pump, consisting of a stator, rotor, and nozzle). Structural and acoustic studies are playing an increasingly important role in the testing, using ABAQUS and SYSNOISE. The cavitation and LES models in FLUENT have seen increased use as well.

FluentNEWS Supplement