"Fluid Dynamics: A Powerful Marine Design Tool," ME Online Web Exclusive, Sept. 14, 2005

for 9/14/05

To read
our previous
ME Online Exclusives,
click here.

Computational Fluid Dynamics:
A Powerful Marine Design Tool

by Shin Hyung Rhee
and Greg Stuckert,
Fluent Inc.,
Lebanon, N.H.

Computational fluid dynamics (CFD) is a powerful marine design tool capable of accurately predicting complex flow phenomena. A key advantage is its ability to model the complex and arbitrary geometries that are typical of real-world equipment.

In recent years, CFD capabilities have been expanded through the development of new physical models. Turbulence models have been developed that are well-suited to the complex flow around ship hulls. Free surface models have been improved to the point where they can simulate breaking ocean waves as well as liquid sloshing in a tank, which is common on liquid natural gas (LNG) carriers. Other multiphase models have been developed to simulate cavitation around hull forms and propellers.

Simulating turbulent flow around ship hulls

One of the most critical requirements for any marine design tool is the ability to simulate turbulent flow around ship hulls. Such challenging features as thick boundary layers under the influence of strong cross-flow, pressure gradients, streamline curvatures, and stream-wise vortices pose a challenge for turbulence modeling. The small margins of improvement that are usually targeted in hull form design today require precision tools capable of discerning small differences between alternative designs.

Engineers at Fluent conducted a series of computations with eight turbulence models in popular use on a typical full hull form that is widely used for modern tankers and bulk carriers. The CFD simulations were compared to tests performed on a scale model of a very large crude-oil carrier (VLCC) at the experimental facilities of the Korean Institute of Ship and Ocean Engineering. Measurements of the mean velocity, turbulence, and resistance (or drag) were made in the towing tank and the wind tunnel facilities.

Fully non-linear free-surface wave pattern around DTMB 5415, a naval surface combatant hull form.

Using Fluent, the mean velocity and turbulence fields in the stern and near wake regions of the hull were computed. The three-dimensional turbulent boundary layer flow was computed using Fluent's Reynolds-averaged Navier-Stokes (RANS) solver. The eight turbulence models studied included six eddy-viscosity models and two second-moment closure models. For the near wall treatment, a wall function approach was adopted so that the viscosity-dominated near wall flow did not have to be computed directly. The study examined several major features of the flow, such as the stern boundary layer, cross-flow separation, stream-wise vortices, and some integral quantities such as nominal wake fraction and resistance.

The performance of the isotropic eddy-viscosity based turbulence models varied over a wide range. The Spalart-Allmaras (SA) and standard k-e (SKE) models fell short in predicting the major features of the flow. While other two-equation eddy-viscosity models improved the predictions, their results were not entirely satisfactory.

The two Reynolds stress transport models (RSTMs), on the other hand, were able to reproduce, with remarkable accuracy, all of the salient features of the flow. For example, they captured the cross-flow and the stream-wise vortices emanating from the hull surface very closely. They also captured the cross-flow that causes the limiting wall streamlines in the stern to converge onto a line that runs roughly along the bilge. They reproduced the unique shape of the contours of the mean velocity field in the propeller disc, which is important because it dictates the inflow to the propeller.

The predictions by the two RSTMs were within 2.5 percent of the total drag. The wake fraction predicted was also in excellent agreement with the experimental data.

Simulating hull wave patterns

Another critical factor in modeling the performance of alternative hull designs is the ability to accurately simulate wave patterns around the hull form. The waves have a big impact on the forces acting on the hull. As the shipbuilding industry moves to design hull forms that run faster and generate less noise, computer simulation is playing an increasingly important role. Custom modeling tools that are based on simplified numerical methods and assumptions cannot provide the accuracy that can be obtained with CFD, which offers many inherent advantages. For example, it can handle arbitrary geometries, viscous flow, and compressible flow. Using Fluent, hull wave patterns for the David Taylor Model Basin Form 5415 have been modeled. This hull form is the shape of a surface combatant, and was chosen because it represents the state of the art in naval design and because extensive physical testing data is available.

The free surface modeling capabilities of Fluent were used to model the hull wave profile. For problems of this type, the greatest challenge is provided by waves that break against the ship. The challenge of modeling breaking waves is increased by the fact that they typically entrain air bubbles. Air entrainment increases the chance that the ship will be detected, and can degrade propeller performance. The results were able to accurately predict the shape of the breaking wave in the immediate vicinity of the bow and stern, representing a major improvement over the Navy's custom tools. CFD tools are not yet capable of modeling the impact of the bubble formation, but substantial improvements have been made in bubble formation models in recent years and these advances can be expected to continue.

Modeling a cavitating propeller

In another project, Fluent software was used to simulate a marine propeller in cavitating operating conditions. Two benchmark propeller models were studied, for which extensive experimental data is available. The propellers were meshed using tetrahedral cells except for the region near the surface where prismatic cells were used to resolve the boundary layer. Both uniform and nonuniform inflow conditions were simulated, and the results were compared to physical measurements. A cavitation model [Singhal et al., ASME J. Fluids Engineering, 124(3), 2002] was used that tracks the interpenetrating (liquid and vapor) fluids using a single momentum equation, a slip velocity equation, and a volume fraction equation. Bubbles form when the local pressure becomes less than the vaporization pressure, and these bubbles may grow and form cavities.

Thrust breakdown is one of the major issues in cavitating propellers. At a large angle of attack, both thrust and torque start decreasing at the onset of cavitation. The simulation accurately reproduced this behavior. The blade backside pressure contours show clearly that cavitation occurs in the tip area. The pressure coefficient in the cavitating area is maintained constant at the opposite of the cavitation number, as expected from general cavitation theory. This prediction of cavitation inception can be confirmed by the contours of vapor volume fraction on the backside of the blade, in which the high vapor volume fraction area closely matches the low-pressure area. The computed iso-surface of vapor volume fraction of 0.1 and the observed cavity shapes were compared. Although the tip vortex cavity is missing in the simulation, primarily due to the mesh resolution in the region, the simulation and experimental measurements are in close agreement on the cavity shape on the blade.

Addressing tank sloshing problems

Tanking sloshing problems in ships or offshore structures are increasingly of concern to naval architects and ocean engineers because of their impact on the safety of the marine structure and cargo. Tanks on LNG carriers are growing larger, and this increases the natural period of sloshing and increases the impact on the ship's motion. In addition, demand is increasing for floating production storage and offloading (FPSO) units, which need to be designed to withstand severe sea states with a sloshing load inside. Until recently, experimental measurements were used to design vessels subjected to sloshing. However, it takes an inordinate amount of time and money to carry out tests, even for simpler geometries and motions. Scaling down for the physical test also can cause accuracy problems.

The volume of fluid (VOF) method in Fluent has been used for many years to model free surface flows, including marine tank sloshing applications. The VOF formulation is designed for cases where there are two or more immiscible fluids. The method tracks the volume fraction of each phase in each computational cell. The fields for all variables and properties are shared by the fluids present in each cell, and represent volume-averaged values. The tracking of the interfaces between the phases is also computed when the VOF model is used.

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

For validation purposes, the experiments carried out at the National Maritime Research Institute of Japan were selected. The shape of the free surface as computed by CFD was found to be in very close agreement with the experimental results. The static pressure histories at three pressure taps, located in a region where the effects of sloshing liquid on the wall are large, were favorably predicted by the simulation. The agreement between measured and computed average impact pressures and force histories on the tank walls was also excellent. These are important for the interaction of the sloshing liquid with other objects, and from a structural analysis standpoint.

As these examples show, CFD has advanced to the point where it can now accurately predict some of the most difficult marine engineering problems. Simulation provides engineers with the ability to accurately determine the performance of design concepts, reducing the need for physical testing and the building of prototypes. This makes it possible to evaluate many more designs, resulting in a substantial improvement in performance. At the same time, the lower cost and shorter lead-times of simulation provide faster time to market and reduced development costs. When you add the fact that CFD simulation provides more design data than physical testing, you end up with an unbeatable proposition for its use in marine research and development.

Reference:

1. Singhal et al., ASME J. Fluids Engineering, 124(3), 2002.