Effect of bed roughness on turbulent boundary layer and net sediment transport under asymmetric waves

This paper presents an investigation of the roughness effects in the turbulent boundary layer for asymmetric waves by using the baseline (BSL) k–ω model. This model is validated by a set of the experimental data with different wave non-linearity index, N_i (namely, N_i = 0.67, N_i = 0.60 and N_i = 0.58). It is further used to simulate asymmetric wave velocity flows with several values of the roughness parameter (a_m/k_s) which increase gradually, namely from a_m/k_s = 35 to a_m/k_s = 963. The effect of the roughness tends to increase the turbulent kinetic energy and to decrease the mean velocity distribution in the inner boundary layer, whereas in the outer boundary layer, the roughness alters the turbulent kinetic energy and the mean velocity distribution is relatively unaffected. A new simple calculation method of bottom shear stress based on incorporating velocity and acceleration terms is proposed and applied into the calculation of the rate of bed-load transport induced by asymmetric waves. And further, the effect of bed roughness on the bottom shear stress and bed-load sediment transport under asymmetric waves is examined with the turbulent model, the newly proposed method, and the existing calculation method. It is found that the higher roughness elements increase the magnitude of bottom shear stress along a wave cycle and consequently, the potential net sediment transport rate. Moreover, the wave non-linearity also shows a big impact on the bottom shear stress and the net sediment transport.     

Numerical modeling of turbulent flows over an urban-type surface: Computations for neutral stratification

Computations of turbulent flows over surfaces with explicitly specified roughness elements that imitate an urban built-up area have been performed using a large-eddy simulation (LES) model. Results are presented for neutral stratification. Some statistics of the flow over an inhomogeneous surface are compared with those over a flat surface. Results of spectral analysis performed to identify characteristic length scales are discussed. A relation is established between the Prandtl mixing length and the turbulence scale defined through the cospectrum-weighted-mean wave number. Values of the roughness parameter and displacement height are determined for three different configurations of objects on the surface.     

Numerical study of turbulence and wave damping induced by vegetation canopies

Vegetation canopies control mean and turbulent flow structure as well as surface wave processes in coastal regions. A non-hydrostatic RANS model based on NHWAVE (Ma et al., 2012) is developed to study turbulent mixing, surface wave attenuation and nearshore circulation induced by vegetation. A nonlinear k − ϵ model accounting for vegetation-induced turbulence production is implemented to study turbulent flow within the vegetation field. The model is calibrated and validated using experimental data from vegetated open channel flow, as well as nonbreaking and breaking random wave propagation in vegetation fields. It is found that the drag-related coefficients in the k − ϵ model C_fk and C_fϵ can greatly affect turbulent flow structure, but seldom change the wave attenuation rate. The bulk drag coefficient C_D is the major parameter controlling surface wave damping by vegetation canopies. Using the empirical formula of Mendez and Losada (2004), the present model provides accurate predictions of vegetation-induced wave energy dissipation. Wave propagation through a finite patch of vegetation in the surf zone is investigated as well. It is found that the presence of a finite patch of vegetation may generate strong pressure-driven nearshore currents, with an onshore mean flow in the unvegetated zone and an offshore return flow in the vegetated zone.     

Calculation of wave-induced turbulent flows in estuaries

The k-ε turbulence model which relates the eddy viscosity to turbulent kinetic energy, k, and to the rate of its dissipation, ε, and determines the distribution of these two quantities from modeled transport equations, is employed in calculating the vertical structure of wave-induced turbulent flows in two-dimensional estuaries. The empirical constants in this model are given the standard values cited in the literature and used successfully for calculating a large variety of steady flows; they are not tuned to the oscillating flows. The free surface elevation and the eddy-viscosity distribution are also calculated as a part of the solution. First, calculations performed for oscillatory laminar flows are compared with analytic solutions to ensure the proper performance of the numerical scheme. The turbulent flow in a laboratory-estuary model with one end closed, and in the Humber Estuary is simulated with the numerical model, and the results are compared with the corresponding flume and field measurements. The influence of frequency and roughness is thereby investigated. The results indicate that the time and space variation of the velocity field and the phase lag between the surface slope and the horizontal velocity can be predicted satisfactorily in wave-induced turbulent flows.     

Wave boundary layer over a stone-covered bed

This paper summarizes the results of an experimental investigation on wave boundary layers over a bed with large roughness, simulating stone/rock/armour block cover on the sea bottom. The roughness elements used in the experiments were stones the size of 1.4cm and 3.85cm in one group of experiments and regular ping-pong balls the size 3.6cm in the other. The orbital-motion-amplitude-to-roughness ratio at the bed was rather small, in the range a/k_s = 0.6–3. The mean and turbulence properties of the boundary-layer flow were measured. Various configurations of the roughness elements were used in the ping-pong ball experiments to study the influence of packing pattern, packing density, number of layers and surface roughness of the roughness elements. The results show that the friction factor seems to be not extremely sensitive to these factors. The results also show that the friction factor for small values of the parameter a/k_s does not seem to tend to a constant value as a/k_s → 0 (contrary to the suggestion made by some previous investigators). The present friction-factor data indicates that the friction factor constantly increases with decreasing a/k_s. An empirical expression is given for the friction factor for small values of a/k_s. The results further show that the phase lead of the bed friction velocity over the surface elevation does not seem to change radically with a/k_s, and found to be in the range 12°–23°. Furthermore the results show that the boundary-layer turbulence also is not extremely sensitive to the packing pattern, the packing density, the number of layers and the surface roughness of the roughness elements. There exists a steady streaming near the bed in the direction of wave propagation, in agreement with the existing work. The present data indicate that the steady streaming is markedly smaller in the case where the ping-pong balls are aligned at 45° to the wave direction than in the case with 90° alignment. Likewise, it is found that the steady streaming is relatively smaller in the case of the one-layer ping-pong-ball roughness than in the case of the two-layer situation.     

Numerical simulation of stably stratified turbulent flows over an urban surface: Spectra and scales and parameterization of temperature and wind-velocity profiles

Stably stratified turbulent flows over surfaces with explicit roughness elements have been calculated using an LES model. The results of calculations for different height distributions of external dynamic forcing are compared and discussed. The spatial spectra and cospectra of turbulent wind-velocity fluctuations have been calculated and different methods of normalizing them have been studied. A parameterization allowing mean wind-velocity and temperature profiles to be approximated has been proposed for turbulence scale.     

Extending the k–ω turbulence model towards oceanic applications

Four different two-equation turbulence models for geophysical flows are compared: The k–ϵ model, two new versions of the k–ω model, and the Mellor–Yamada model. An extension of the k–ω model for buoyancy affected and rotating flows is suggested. Model performance is evaluated for a few typical oceanic flows. First, new analytical solutions of the models for the surface layer affected by breaking surface waves are discussed. The deficiencies of earlier attempts are high-lighted, and it is demonstrated why the Mellor–Yamada model and the k–ϵ model fail. It is illustrated that only one version of the k–ω model computes correct decay rates for turbulent quantities under breaking waves. Second, it is demonstrated that all models predict almost identical mixed layer depths and profiles for the turbulent kinetic energy in a classical stratified shear-entrainment experiment if the buoyancy term in the second equation is appropriately weighted. Third, the accuracy and numerical robustness of the new k–ω model in realistic oceanic situations is confirmed by comparison with the data-set of the Ocean Weather Ship ‘Papa’.     

Movable bed roughness and current profiles in the presence of irregular waves with an arbitrary angle to currents

The bed roughness k_s and current velocity profiles in the presence of waves with an arbitrary angle θ to currents are studied. It is found that the movable bed roughness is affected by both the wave and the current and only slightly by the angle θ between the wave propagation and the current, and that existing formulae derived in purely oscillatory flows generally fail to predict k_s. In the present study, a new formula which takes account the effect of the wave and the current on the bed roughness is suggested to calculate k_s in combined wave-current flows. With the present formula, the current profiles calculated by the model of You agree satisfactorily with the laboratory data of van Kampen and Nap and Havinga, and the field measurements of Grant and Williams and Drake et al.     

Numerical modeling investigation on turbulent oscillatory flow over a plane rough bed composed by randomly arrayed particles

A three-dimensional numerical model is established to simulate the turbulent oscillatory boundary layer over a fixed and rough bed composed by randomly arrayed solid spheres based on the lattice Boltzmann method and the large eddy simulation model.The equivalent roughness height,the location of the theoretical bed and the time variation of the friction velocity are investigated using the log-fit method.The time series of turbulent intensity and Reynolds stress are also investigated.The equivalent roughness height of cases with Reynolds numbers of 1×10~4–6×10~4 is approximately 2.81 d(grain size).The time variation of the friction velocity in an oscillatory cycle exhibits sinusoidal-like behavior.The friction factor depends on the relative roughness in the rough turbulent regime,and the pattern of solid particles arrayed as the rough bed in the numerical simulations has no obvious effect on the friction factor.     

Observation of wave-enhanced turbulence in the near-surface layer of the ocean during TOGA COARE

Dissipation rate statistics in the near-surface layer of the ocean were obtained during the month-long COARE Enhanced Monitoring cruise with a microstructure sensor system mounted on the bow of the research vessel. The vibration contamination was cancelled with the Wiener filter. The experimental technique provides an effective separation between surface waves and turbulence, using the difference in spatial scales of the energy-containing surface waves and small-scale turbulence. The data are interpreted in the coordinate system fixed to the ocean surface. Under moderate and high wind-speed conditions, we observed the average dissipation rate of the turbulent kinetic energy in the upper few meters of the ocean to be 3–20 times larger than the logarithmic layer prediction. The Craig and Banner (J. Phys. Oceanogr. 24 (1994) 2546) model of wave-enhanced turbulence with the surface roughness length from the water side z₀ parameterized according to the Terray et al. (J. Phys. Oceanogr. 26 (1996) 792) formula z₀=cH_s provides a reasonable fit to the experimental dissipation profile, where z is the depth (defined here as the distance to the ocean surface), c≈0.6, and H_s is the significant wave height. In the wave-stirred layer, however, the average dissipation profile deviates from the model (supposedly because of extensive removing of the bubble-disturbed areas close to the ocean surface). Though the scatter of individual experimental dissipation rates (10-min averages) is significant, their statistics are consistent with the Kolmogorov's concept of intermittent turbulence and with previous studies of turbulence in the upper ocean mixed layer.     

Combined tidal and wind driven flows and bedload transport over a flat bottom

The combined tidal and wind driven flow and resulting sediment transport in the ocean over a flat bottom at intermediate water depth has been investigated, using a simple one dimensional two-equation turbulence closure model. This model has been verified against field measurements of a tidal flow in the Celtic Sea. The tidal velocity ellipses and the time series of the horizontal velocity components at given elevations above the bottom are well predicted through the water column although there are some deviations between the predicted and measured velocities near the bottom due to the uncertainty of the bottom roughness. For the combined tidal and wind driven flows the velocity profiles, turbulent kinetic energy profiles and surface particle trajectories are predicted for weak and strong winds. Furthermore, the bottom shear stress and the resulting bedload transport have been predicted; the parts of the particle trajectories in the close vicinity of the bottom where the bedload transport exists are displayed. Finally, the direction and magnitude of the surface drift, the depth-averaged mean velocity and the mean bedload transport are given, and the effect of the bottom roughness on the sea surface drift is investigated.     

Evaluation of spatial distribution of turbulent mixing in the central Pacific

A long-term mean turbulent mixing in the depth range of 200–1000 m produced by breaking of internal waves across the middle and low latitudes (40°S–40°N) of the Pacific between 160°W and 140°W is examined by applying fine-scale parameterization depending on strain variance to 8-year (2005–2012) Argo float data. Results show that elevated turbulent dissipation rate (ε) is related to significant topographic regions, along the equator, and on the northern side of 20°N spanning to 24°N throughout the depth range. Two patterns of latitudinal variations of ε and the corresponding diffusivity (K_ρ) for different depth ranges are confirmed: One is for 200–450 m with significant larger ε and K_ρ, and the maximum values are obtained between 4°N and 6°N, where eddy kinetic energy also reaches its maximum; The other is for 350–1000 m with smaller ε and K_ρ, and the maximum values are obtained near the equator, and between 18°S and 12°S in the southern hemisphere, 20°N and 22°N in the northern hemisphere. Most elevated turbulent dissipation in the depth range of 350–1000 m relates to rough bottom roughness (correlation coefficient?=?0.63), excluding the equatorial area. In the temporal mean field, energy flux from surface wind stress to inertial motions is not significant enough to account for the relatively intensified turbulent mixing in the upper layer.     

Comparison of laminar model,RANS, LES and VLES for simulation of liquid sloshing

Liquid sloshing in storage tank is a fundamental problem of great engineering importance. Sloshing motion can be laminar or turbulent. However, the necessity for inclusion of turbulence in CFD simulation of sloshing flows has not yet been established. In this paper, three roll–induced sloshing cases are studied to assess the merits and shortcomings of the laminar model and three most–commonly used turbulence models (RANS k–ε, LES and Very LES). To overcome the deficiencies in the RANS and LES, the new Very LES (VLES) model, which combines the RANS k–ε and LES, is developed in this paper. The free surface profiles are reconstructed by a coupled Level–Set and Volume–of–Fluid (CLSVOF) method. To the authors’ knowledge, the comprehensive and systematical assessment of the effect of turbulence on sloshing simulation has not been reported in the literature. The numerical results are evaluated using experimental measurements from Delorme and Souto−Iglesias. The present study indicates that the inclusion of an appropriate turbulence model has a profound influence on the simulations of violent and non–violent sloshing flows. The VLES and LES models can provide accurate predictions of free surface profiles and impact pressures, whereas the laminar flow assumption and the RANS model cannot adequately capture the energy dissipation in the sloshing simulation and lead to the inaccurate flow predictions.     

Force and flow characteristics of a circular cylinder with uniform surface roughness at subcritical Reynolds numbers

The main purpose of this study is to establish a better understanding of the relationship between drag reduction and surface roughness. Experiments were conducted to measure the force and flow characteristics of a circular cylinder with different types of artificial surface roughness over the range 6 × 10³ < Re < 8 × 10⁴ (Re is based on the cylinder diameter D). The roughness cylinder was formed by covering the exterior surface of the cylinder with uniformly distributed (1) sandpaper, (2) netting, and (3) dimples. The roughness coefficient ranged from k/D = 0.0028 to 0.025 (k is the roughness height). A detailed quantitative measurement of the flow field around the cylinder using Particle Imaging Velocimetry (PIV) was carried out. The hydrodynamic force coefficients (drag and lift) of the rough cylinders are compared against those of a smooth cylinder measured under the same flow conditions. It is found that certain configuration of surface roughness significantly reduces the mean drag coefficient of the cylinder, particularly at large Reynolds numbers. In addition, the root-mean-square (r.m.s.) lift coefficient of the rough cylinders is considerably lower than that of a smooth cylinder.     

On the turbulent prandtl number in a stably stratified atmospheric boundary layer

The results obtained from both atmospheric and laboratory measurements and from LES data show that, in the stably stratified flows of the atmospheric boundary layer, turbulent mixing occurs at gradient Richardson numbers Ri _g that significantly exceed one: the inverse turbulent Prandtl number Pr _t ⁻¹ decreases with an increase in the thermal flow stability. The decreasing trend of the inverse turbulent Ptandtl number is reproduced in a stably stratified atmospheric boundary layer in agreement with measurement data with the aid of an improved three-parameter turbulence model. In this model, a modified model that takes into account the effect of stratification in the expression for the time scale of the scalar field is used for the pressure-scalar correlation.     

Modeling the underwater light field fluctuations in coastal oceanic waters: Validation with experimental data

Modeling of the wave-induced underwater light fluctuations at near-surface depths in coastal oceanic waters is challenging because of the surface roughness and strong anisotropic effects of the light field. In the present work, a simple and computationally efficient radiative transfer model is used for the wind-driven sea surface for simulating underwater light fields such as downwelling irradiance (E_d), upwelling irradiance (E_u), and upwelling radiance (L_u) in a spatial domain. It is an extension of our previous work that essentially combines the air–sea interface of the wind-driven sea surface with transmittance and reflectance along with the diffuse and direct components of the homogenous and inhomogeneous water column. The present model simulates underwater light fields for any possible values of absorption and backscattering coefficients. To assess the performance of the model, the E_d, E_u, and L_u profiles predicted by the model are compared with experimental data from relatively clear and turbid coastal waters. Statistical results show significantly low mean relative differences regardless of the wavelength. Comparison of the simulated and in-situ time series data measured over rough sea surfaces demonstrates that model-observation agreement is good for the present model. The Hydrolight model when implemented with the modified bottom reflectance and phase function provides significantly better results than the original Hydrolight model without consideration of the bottom slope and vertically varying phase function. However, these results are non-spatial and have errors fluctuating at different wavelengths. To further demonstrate the efficiency of the present model, spatial distribution patterns of the underwater light fields are simulated based on the measured data from a coastal station for different solar zenith angles (under sunny condition). Simulated wave-induced fluctuations of the underwater lights fields show a good consistency with in-situ data for a few near-surface depths. The present model also provides a reasonable approximation for simulating wave-induced effects on the downward irradiance field and its anisotropic conditions caused by the surface roughness, wavelength and angle of incidence.     

A new efficient finite volume modeling of small amplitude free surface flows with unstructured grid

A staggered finite-volume technique for non-hydrostatic, small amplitude free surface flow governed by the incompressible Navier-Stokes equations is presented there is a proper balance between accuracy and computing time. The advection and horizontal diffusion terms in the momentum equation are discretized by an integral interpolation method on the orthogonal unstructured staggered mesh and, while it has the attractive property of being conservative. The pressure-correction algorithm is employed for the non-hydrostatic pressure in order to achieve second-order temporal accuracy. A conservative scalar transport algorithm is also applied to discretize k-ε equations in this model. The eddy viscosity is calculated from the k-ε turbulent model. The resulting model is mass and momentum conservative. The model is verified by two examples to simulate unsteady small amplitude free surface flows where non-hydrostatic pressures have a considerable effect on the velocity field,and then applied to simulate the tidal flow in the Bohai Sea.     

Relevance of transition turbulent model for hydrodynamic characteristics of low Reynolds number propeller

The propeller of an Autonomous Underwater Vehicle (AUV) operates at low Reynolds number in laminar to turbulent transition region. The performance of these propellers can be calculated accurately using RANSE solver with γ − Re_θ transition model. In this study, the global and local hydrodynamic characteristics of open and ducted propeller are investigated using the γ − Re_θ transition model. The capability of the γ − Re_θ transition model to capture laminar to turbulent transition on the surface of the open propeller is demonstrated by comparison with published experimental results. The application of transition model for the propeller K_a-4-70 inside the duct 19A shows that the centrifugal forces are dominant at low Reynolds number and the flow is mainly directed in the radial direction. The transition model is able to predict complex flow physics such as leading-edge separation, tip leakage vortex, and the separation bubble on outer surface of the duct. The accurate prediction of these flow phenomenon can lead to correct calculation of global hydrodynamic forces and moments acting on the propeller at low Reynolds number.     

The “slip law” of the free surface

A “slip law” connects the excess velocity or “slip” of a wind-blown water surface, relative to the motion in the middle of the mixed layer, to the wind stress, the wind-wave field, and buoyancy flux. An inner layer-outer layer model of the turbulent shear flow in the mixed layer is appropriate, as for a turbulent boundary layer or Ekman layer over a solid surface, allowing, however, for turbulent kinetic energy transfer from the air-side via breaking waves, and for Stokes drift. Asymptotic matching of the velocity distributions in inner and outer portions of the mixed layer yields a slip law of logarithmic form, akin to the drag law of a turbulent boundary layer. The dominant independent variable is the ratio of water-side roughness length to mixed layer depth or turbulent Ekman depth. Convection due to surface cooling is also an important influence, reducing surface slip. Water-side roughness length is a wind-wave property, varying with wind speed similarly to air-side roughness. Slip velocity is typically 20 times water-side friction velocity or 3% of wind speed, varying within a range of about 2 to 4.5%. A linearized model of turbulent kinetic energy distribution shows much higher values near the surface than in a wall layer. Nondimensional dissipation peaks at a value of about eight, a short distance below the surface.