Strategies to Minimise Numerical Ventilation in CFD Simulations of High-Speed Planing Hulls

Numerical Ventilation (NV) is a well-known problem that occurs when the Volume of Fluid method is used to model vessels with a bow that creates a small, acute entrance angle with the free surface. These are typical of both planing hulls and yachts. There is a general lack of discussion focusing upon Numerical Ventilation available within the public domain, which is attributable to the fact that it only affects such a niche area of naval architecture. The information available is difficult to find, often fleetingly mentioned in papers with a different focus. Numerical Ventilation may be considered one of the main sources of error in numerical simulations of planing hulls and as such warrants an in-depth analysis. This paper sets out to bring together the available work, as well as performing its own investigation into the problem to develop a better understanding of Numerical Ventilation and present alternate solutions. Additionally, the success and impact of different approaches is presented in an attempt to help other researchers avoid and correct for Numerical Ventilation. Interface smearing caused by the simulations inability to track the free surface is identified as the main source of Numerical Ventilation. This originates from the interface between the volume mesh and the prism layer mesh. This study looks into the interface to identify strategies that minimise Numerical Ventilation, presenting a novel solution to prism layer meshing that was found to have a positive impact. Through the implementation of a modified High Resolution Interface Capture (HRIC) scheme and the correct mesh refinements, it is possible to minimise the impact of Numerical Ventilation to a level that will not affect the results of a simulation and is acceptable for engineering applications.

Three-Dimensional Numerical Analysis of Horizontal and Vertical Coalescence of Bubbles at Two Submerged Horizontal Orifices on the Wall

In the process of bubbling from two submerged adjacent orifices, bubbles coalescence becomes inevitable. But the study of the evolution and interaction of bubbles from submerged orifices is little, especially numerical simulation. In this paper, combined with mesh smoothing technique, mesh subdivision technique and the technique of axisymmetric coalescence and 3D coalescence, a three-dimensional model of bubbles coalescence at two submerged adjacent orifices on the wall is established by the boundary element method. Then, numerical simulations were carried out for horizontal and vertical coalescence before detachment. Finally, by changing the ventilation rate and the Froude number, the effects of different ventilation rates and buoyancy on the process of bubbles coalescence at two adjacent orifices were investigated. The results show that for horizontal coalescence, the effect of ventilation rate is more pronounced than buoyancy. As the ventilation rate increases or the influence of buoyancy is decreased, the amplitude of internal pressure fluctuation of the bubble decreases and the coalescence time decreases. For vertical coalescence, the effect of buoyancy is more pronounced than ventilation rate. With the influence of buoyancy is decreased, the vertical coalescence time is increased, the internal pressure of the bubble is decreased. The influence of ventilation rate is similar to that of horizontal coalescence.

REEF3D::FNPF: A Flexible Fully Nonlinear Potential Flow Solver

In situations where the calculation of ocean wave propagation and impact on offshore structures is required, fast numerical solvers are desired in order to find relevant wave events in a first step. After the identification of the relevant events, Computational Fluid Dynamics (CFD) based Numerical Wave Tanks (NWT) with an interface capturing two-phase flow approach can be used to resolve the complex wave structure interaction, including breaking wave kinematics. CFD models emphasize detail of the hydrodynamic physics, which makes them not the ideal candidate for the event identification due to the large computational resources involved. In the current paper a new numerical wave model is represented that solves the Laplace equation for the flow potential and the nonlinear kinematic and dynamics free surface boundary conditions. This approach requires reduced computational resources compared to CFD based NWTs. In contrast to existing approaches, the resulting fully nonlinear potential flow solver REEF3D::FNPF uses a σ-coordinate grid for the computations. Solid boundaries are incorporated through a ghost cell immersed boundary method. The free surface boundary conditions are discretized using fifth-order WENO finite difference methods and the third-order TVD Runge-Kutta scheme for time stepping. The Laplace equation for the potential is solved with Hypres stabilized bi-conjugated gradient solver preconditioned with geometric multi-grid. REEF3D::FNPF is fully parallelized following the domain decomposition strategy and the MPI communication protocol. The model is successfully tested for wave propagation benchmark cases for shallow water conditions with variable bottom as well as deep water.

Numerical Manoeuvrable Tank on Wave Based Moving Domain

The maneuvering motion simulation of ship in waves needs large space, the simulation based static domain requires a large amount of mesh. On the contrary, the moving domain can reduce the grid amount and improve the simulation speed. This paper studies the establishment of numerical wave tank based moving domain. The wave-generation method of defining inlet boundary conditions is proposed. The level set method is used to solve the free surface. The URANS equation is solved by finite difference method and the projection algorithm. For the wave-generation method of defining inlet boundary conditions, the velocity inlet direction is always fixed because the direction of incident wave is always fixed. Specifically, the yaw should be turned off avoid the disturbance of wave in the y direction. Comparing the simulated wave with the target wave, the numerical results show that the simulated value is coincident with the theoretical value in the moving domain, and there are no obvious signs of dissipation over time. The established moving domain can produce a continuous and qualified wave, which provides a theoretical basis for the study of maneuvering simulation of ship in waves.

Forced Vibration Tests for In-Line VIV to Assess Partially Strake-Covered Pipeline Spans

A series of experiments is performed in which a strake-covered rigid cylinder undergoes harmonic purely in-line motion while subject to a uniform “flow” created by towing the test rig along SINTEF Ocean’s towing tank. These tests are performed for a range of frequencies and amplitudes of the harmonic motion, to generate added-mass and excitation functions are derived from the in-phase and 90° out-of-phase components of the hydrodynamic force on the pipe, respectively. Using these excitation- and added-mass functions in VIVANA together with those from experiments on bare pipe by Aronsen (2007), the in-line VIV response of partially strake-covered pipeline spans is calculated. It is found that as little as 10% strake coverage at the optimal location effectively suppresses pure in-line VIV. Further advantages of strakes rather than intermediate supports to suppress in-line VIV include: strakes are not affected by the scour which can lower an intermediate support (in addition to creating the span in the first place). Further they do not prevent self-lowering of the pipeline or act as a point of concentration of VIV damage as the spans to each side of the intermediate support grow again.

An Investigation Into the Effect of Biofouling on Full-Scale Propeller Performance Using CFD

The negative effect of biofouling on ship resistance has been investigated since the early days of naval architecture. However, for more precise prediction of fuel consumption of ships, understanding the effect of biofouling on ship propulsion performance is also important. In this study, CFD simulations for the full-scale performance of KP505 propeller in open water, including the presence of marine biofouling, were conducted. To predict the effect of barnacle fouling on the propeller performance, experimentally obtained roughness functions of barnacle fouling were employed in the wall-function of the CFD software. The roughness effect of barnacles of varying sizes and coverages on the propeller open water performance was predicted for advance coefficients ranging from 0.2 to 0.8. From the simulations, drastic effects of barnacle fouling on the propeller open water performance were found. The result suggests that the thrust coefficient decreases while the torque coefficient increases with increasing level of surface fouling, which leads to a reduction of the open water efficiency of the propeller. Further investigations into the roughness effect on the pressure and velocity field, surface pressure and wall shear stress, and propeller vortices were examined.

Assessment of LNG Pump Tower Loads

The assessment of design loads acting on Liquefied Natural Gas (LNG) pump tower are widely based on Morison equation. However, the Morison equation lacks consideration of transverse flow, impact loads and the interaction between fluid and structure. Studies dealing with a direct simulation of LNG pump tower loads by means of Computational Fluid Dynamics (CFD), which can cover the aforementioned effects, are currently not available. A comparative numerical study on LNG pump tower loads is presented in this paper focusing on the following two questions: Are impact loads relevant for the structural design of LNG pump towers? In which way does the fluid-structure interaction influence the loads? Numerical simulations of the multiphase problem were conducted using field methods. Firstly, Unsteady Reynolds-Averaged Navier-Stokes (URANS) equations, extended by the Volume of Fluid (VoF) approach were used to simulate the flow inside a three-dimensional LNG tank in model scale without tower structure. The results were used to validate the numerical model against model tests. Motion periods and amplitudes were systematically varied. Velocities and accelerations along the positions of the main structural members of the pump tower were extracted and used as input data for load approximations with the Morison equation. Morison equation, URANS and Delayed Detached Eddy Simulation (DDES) computed tower loads were compared. Time histories as well as statistically processed data were used. Global loads acting on the full (with tower structure) and simplified structure (no tower structure, but using Morison equation) are in the same order of magnitude. However, their time evolution is different, especially at peaks, which is considered significant for the structural design.

Multi-Phase Simulation of Droplet Trajectories of Wave-Impact Sea Spray Over a Vessel

Sea spray, generated by ship-wave collisions, is the main source of marine icing. In certain, but not all, circumstances a cloud of spray forms after a wave impacts a ship. The spray cloud comprises numerous water droplets of various sizes. These droplets are dispersed and transported over the vessel deck by the surrounding wind and fall onto the deck or into the ocean under the effect of gravity. The motion of these droplets is important since they determine the extent of the spray cloud and its duration over the deck, which consequently affects the distribution of icing accumulation on a ship in freezing weather. In this paper, a multi-phase air-water simulation of droplet trajectory is used to predict the cloud motion of various size droplets. A smooth particle hydrodynamics (SPH) computational fluid dynamics (CFD) model is implemented and the simulation is accelerated using GPU computing. The field observation data is used to simulate the trajectory. The results of the simulations are compared with an available theoretical model and reasonable agreement is found. The inverse dependence of size and velocity for droplets after the breakup process is examined. The simulation results are consistent with the theoretical model in that neither the largest nor the smallest droplets reach the maximum height of the spray cloud, but the mid-size droplets do. The spray cloud spreads faster and crosses the front of the vessel quicker than predicted by the theoretical model. It is also found that the trajectory of a single droplet is significantly affected by surrounding droplets in a multi-droplet trajectory model. A mono-droplet theoretical trajectory model, therefore, is not as accurate as the multi-droplet CFD model.

Numerical and Mechanistic Modelling of Two-Phase Liquid-Gas Flow’s Pressure Drop Across Sharp-Edged Orifices

Pressure drop estimation across orifices for two-phase liquid-gas flow is essential to size valves and pipelines and decrease the probability of unsafe consequences or high costs in petroleum, chemical, and nuclear industries. While numerically modeling flow across orifices is a complex task, it can assess the effect of numerous orifice designs and operation parameters. In this paper, two-phase flow across orifices has been numerically modeled to investigate the effect of different fluid combinations and orifice geometries on pressure drop. The orifice is assumed to be located in a pipe with fully-developed upstream and downstream flow. Two liquid-gas fluid combinations, namely water-air, and gasoil liquid-gas mixture were investigated for different orifice to pipe area ratios ranging from 0.01 to 1 for the superficial velocity of 10 m/s. Volume of Fluid multiphase flow model along with k-epsilon turbulence model were used to estimate the pressure distribution of liquid-gas mixture along the pipe. The numerical model was validated for water-air with mean relative error less than 10.5%. As expected, a decrease in orifice to pipe area ratio resulted in larger pressure drops due to an increase in the contraction coefficients of the orifice assembly. It was also found that water-air had larger pressure drops relative to gasoil mixture due to larger vortex formation downstream of orifices. In parallel, a mechanistic model to directly estimate the local two-phase pressure drop across orifices was developed. The gas void fraction was predicted using a correlation by Woldesemayat and Ghajar, and applied to separated two-phase flow undergoing contraction and expansion due to an orifice. The model results were validated for different orifices and velocities, with the overall relative error of less than 40%, which is acceptable due to the uncertainties associated with measuring experimental pressure drop. Comparison of the developed numerical and mechanistic model showed that the numerical model is able to achieve a higher accuracy, while the mechanistic model requires minimal computation.

REEF3D Wave Generation Interface for Commercial CFD Codes

In recent years CFD developments have shown a trend to combine RANS CFD simulation with other methods such as wave theories or velocity potential based numerical wave tanks, in order to reduce to computation costs. This is however not a new approach, and there exists a large amount of literature about domain decomposition techniques describing a two way coupling between the RANS CFD models and other methods. One can also observe an increasing popularity in the use of a less sophisticated technique where different fluid solvers are combined with one-way coupling. In these methods a predefined solution is provided in the far-field (ignoring the structure), while a three-dimensional (3D) CFD simulation is applied in a limited zone near the structure. The predefined solution is used to specify the background far-field solution. The governing equations are extended by the addition of a source term. The published solutions use wave theory or a numerical wave tank where the predefined solution is calculated parallel to the RANS solver. In this way it is possible to reduce the interpolation inaccuracy and the amount of transferred data to the CFD simulation. The disadvantage of this technique is that the far field solver has to be prepared in order to run in parallel with the CFD solver. Due to the one way coupling it is possible to predefine this information in tables before the CFD simulation. This technique makes it possible to define a general interface between difference solvers without modifying existing codes. This paper presents such a technique where the predefined solution is stored into files. An interpolation function delivers all data to the far-field solution for the CFD simulation. The paper analyses the necessary accuracy of the interpolation and the costs of the input/output operation of the CFD simulation through several verification cases.