Figure 11. Sketch of scour mechanism around USAF under random waves.

Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves

by Ruigeng Hu¹,Hongjun Liu²,Hao Leng¹,Peng Yu³ andXiuhai Wang^1,2,*

¹College of Environmental Science and Engineering, Ocean University of China, Qingdao 266000, China

²Key Lab of Marine Environment and Ecology (Ocean University of China), Ministry of Education, Qingdao 266000, China

³Qingdao Geo-Engineering Survering Institute, Qingdao 266100, China

^*Author to whom correspondence should be addressed.

J. Mar. Sci. Eng. 2021, 9(8), 886; https://doi.org/10.3390/jmse9080886

Received: 6 July 2021 / Revised: 8 August 2021 / Accepted: 13 August 2021 / Published: 17 August 2021

(This article belongs to the Section Ocean Engineering)

Download

Abstract

A series of numerical simulation were conducted to study the local scour around umbrella suction anchor foundation (USAF) under random waves. In this study, the validation was carried out firstly to verify the accuracy of the present model. Furthermore, the scour evolution and scour mechanism were analyzed respectively. In addition, two revised models were proposed to predict the equilibrium scour depth S_eq around USAF. At last, a parametric study was carried out to study the effects of the Froude number F_r and Euler number E_u for the S_eq. The results indicate that the present numerical model is accurate and reasonable for depicting the scour morphology under random waves. The revised Raaijmakers’s model shows good agreement with the simulating results of the present study when KC_s,p < 8. The predicting results of the revised stochastic model are the most favorable for n = 10 when KC_rms,a < 4. The higher F_r and E_u both lead to the more intensive horseshoe vortex and larger S_eq.

Keywords:

scour; numerical investigation; random waves; equilibrium scour depth; KC number

1. Introduction

The rapid expansion of cities tends to cause social and economic problems, such as environmental pollution and traffic jam. As a kind of clean energy, offshore wind power has developed rapidly in recent years. The foundation of offshore wind turbine (OWT) supports the upper tower, and suffers the cyclic loading induced by waves, tides and winds, which exerts a vital influence on the OWT system. The types of OWT foundation include the fixed and floating foundation, and the fixed foundation was used usually for nearshore wind turbine. After the construction of fixed foundation, the hydrodynamic field changes in the vicinity of the foundation, leading to the horseshoe vortex formation and streamline compression at the upside and sides of foundation respectively [1,2,3,4]. As a result, the neighboring soil would be carried away by the shear stress induced by vortex, and the scour hole would emerge in the vicinity of foundation. The scour holes increase the cantilever length, and weaken the lateral bearing capacity of foundation [5,6,7,8,9]. Moreover, the natural frequency of OWT system increases with the increase of cantilever length, causing the resonance occurs when the system natural frequency equals the wave or wind frequency [10,11,12]. Given that, an innovative foundation called umbrella suction anchor foundation (USAF) has been designed for nearshore wind power. The previous studies indicated the USAF was characterized by the favorable lateral bearing capacity with the low cost [6,13,14]. The close-up of USAF is shown in Figure 1, and it includes six parts: 1-interal buckets, 2-external skirt, 3-anchor ring, 4-anchor branch, 5-supporting rod, 6-telescopic hook. The detailed description and application method of USAF can be found in reference [13].

Figure 1. The close-up of umbrella suction anchor foundation (USAF).

Numerical and experimental investigations of scour around OWT foundation under steady currents and waves have been extensively studied by many researchers [1,2,15,16,17,18,19,20,21,22,23,24]. The seabed scour can be classified as two types according to Shields parameter θ, i.e., clear bed scour (θ < θ_cr) or live bed scour (θ > θ_cr). Due to the set of foundation, the adverse hydraulic pressure gradient exists at upstream foundation edges, resulting in the streamline separation between boundary layer flow and seabed. The separating boundary layer ascended at upstream anchor edges and developed into the horseshoe vortex. Then, the horseshoe vortex moved downstream gradually along the periphery of the anchor, and the vortex shed off continually at the lee-side of the anchor, i.e., wake vortex. The core of wake vortex is a negative pressure center, liking a vacuum cleaner. Hence, the soil particles were swirled into the negative pressure core and carried away by wake vortexes. At the same time, the onset of scour at rear side occurred. Finally, the wake vortex became downflow when the turbulence energy could not support the survival of wake vortex. According to Tavouktsoglou et al. [25], the scale of pile wall boundary layer is proportional to 1/ln(R_d) (R_d is pile Reynolds), which means the turbulence intensity induced by the flow-structure interaction would decrease with R_d increases, but the effects of R_d can be neglected only if the flow around the foundation is fully turbulent [26]. According to previous studies [1,15,27,28,29,30,31,32], the scour development around pile foundation under waves was significantly influenced by Shields parameter θ and KC number simultaneously (calculated by Equation (1)). Sand ripples widely existed around pile under waves in the case of live bed scour, and the scour morphology is related with θ and KC. Compared with θ, KC has a greater influence on the scour morphology [21,27,28]. The influence mechanism of KC on the scour around the pile is reflected in two aspects: the horseshoe vortex at upstream and wake vortex shedding at downstream.

KC=UwmTD��=�wm��(1)

where, U_wm is the maximum velocity of the undisturbed wave-induced oscillatory flow at the sea bottom above the wave boundary layer, T is wave period, and D is pile diameter.

There are two prerequisites to satisfy the formation of horseshoe vortex at upstream pile edges: (1) the incoming flow boundary layer with sufficient thickness and (2) the magnitude of upstream adverse pressure gradient making the boundary layer separating [1,15,16,18,20]. The smaller KC results the lower adverse pressure gradient, and the boundary layer cannot separate, herein, there is almost no horseshoe vortex emerging at upside of pile. Sumer et al. [1,15] carried out several sets of wave flume experiments under regular and irregular waves respectively, and the experiment results show that there is no horseshoe vortex when KC is less than 6. While the scale and lifespan of horseshoe vortex increase evidently with the increase of KC when KC is larger than 6. Moreover, the wake vortex contributes to the scour at lee-side of pile. Similar with the case of horseshoe vortex, there is no wake vortex when KC is less than 6. The wake vortex is mainly responsible for scour around pile when KC is greater than 6 and less than O(100), while horseshoe vortex controls scour nearly when KC is greater than O(100).

Sumer et al. [1] found that the equilibrium scour depth was nil around pile when KC was less than 6 under regular waves for live bed scour, while the equilibrium scour depth increased with the increase of KC. Based on that, Sumer proposed an equilibrium scour depth predicting equation (Equation (2)). Carreiras et al. [33] revised Sumer’s equation with m = 0.06 for nonlinear waves. Different with the findings of Sumer et al. [1] and Carreiras et al. [33], Corvaro et al. [21] found the scour still occurred for KC ≈ 4, and proposed the revised equilibrium scour depth predicting equation (Equation (3)) for KC > 4.

Rudolph and Bos [2] conducted a series of wave flume experiments to investigate the scour depth around monopile under waves only, waves and currents combined respectively, indicting KC was one of key parameters in influencing equilibrium scour depth, and proposed the equilibrium scour depth predicting equation (Equation (4)) for low KC (1 < KC < 10). Through analyzing the extensive data from published literatures, Raaijmakers and Rudolph [34] developed the equilibrium scour depth predicting equation (Equation (5)) for low KC, which was suitable for waves only, waves and currents combined. Khalfin [35] carried out several sets of wave flume experiments to study scour development around monopile, and proposed the equilibrium scour depth predicting equation (Equation (6)) for low KC (0.1 < KC < 3.5). Different with above equations, the Khalfin’s equation considers the Shields parameter θ and KC number simultaneously in predicting equilibrium scour depth. The flow reversal occurred under through in one wave period, so sand particles would be carried away from lee-side of pile to upside, resulting in sand particles backfilled into the upstream scour hole [20,29]. Considering the backfilling effects, Zanke et al. [36] proposed the equilibrium scour depth predicting equation (Equation (7)) around pile by theoretical analysis, and the equation is suitable for the whole range of KC number under regular waves and currents combined.

S/D=1.3(1−exp([−m(KC−6)])�/�=1.3(1−exp(−�(��−6))(2)

where, m = 0.03 for linear waves.

S/D=1.3(1−exp([−0.02(KC−4)])�/�=1.3(1−exp(−0.02(��−4))(3)

S/D=1.3γKwaveKhw�/�=1.3��wave�ℎw(4)

where, γ is safety factor, depending on design process, typically γ = 1.5, K_wave is correction factor considering wave action, K_hw is correction factor considering water depth.

S/D=1.5[tanh(hwD)]KwaveKhw�/�=1.5tanh(ℎw�)�wave�ℎw(5)

where, h_w is water depth.

S/D=0.0753(θθcr−−−√−0.5)0.69KC0.68�/�=0.0753(��cr−0.5)0.69��0.68(6)

where, θ is shields parameter, θ_cr is critical shields parameter.

S/D=2.5(1−0.5u/uc)xrelxrel=xeff/(1+xeff)xeff=0.03(1−0.35ucr/u)(KC−6)⎫⎭⎬⎪⎪�/�=2.5(1−0.5�/��)��=��/(1+��)��=0.03(1−0.35�cr/�)(��−6)(7)

where, u is near-bed orbital velocity amplitude, u_c is critical velocity corresponding the onset of sediment motion.

S/D=1.3{1−exp[−0.03(KC2lnn+36)1/2−6]}�/�=1.31−exp−0.03(��2ln�+36)1/2−6(8)

where, n is the 1/n’th highest wave for random waves

For predicting equilibrium scour depth under irregular waves, i.e., random waves, Sumer and Fredsøe [16] found it’s suitable to take Equation (2) to predict equilibrium scour depth around pile under random waves with the root-mean-square (RMS) value of near-bed orbital velocity amplitude U_m and peak wave period T_P to calculate KC. Khalfin [35] recommended the RMS wave height H_rms and peak wave period T_P were used to calculate KC for Equation (6). References [37,38,39,40] developed a series of stochastic theoretical models to predict equilibrium scour depth around pile under random waves, nonlinear random waves plus currents respectively. The stochastic approach thought the 1/n’th highest wave were responsible for scour in vicinity of pile under random waves, and the KC was calculated in Equation (8) with U_m and mean zero-crossing wave period T_z. The results calculated by Equation (8) agree well with experimental values of Sumer and Fredsøe [16] if the 1/10′th highest wave was used. To author’s knowledge, the stochastic approach proposed by Myrhaug and Rue [37] is the only theoretical model to predict equilibrium scour depth around pile under random waves for the whole range of KC number in published documents. Other methods of predicting scour depth under random waves are mainly originated from the equation for regular waves-only, waves and currents combined, which are limited to the large KC number, such as KC > 6 for Equation (2) and KC > 4 for Equation (3) respectively. However, situations with relatively low KC number (KC < 4) often occur in reality, for example, monopile or suction anchor for OWT foundations in ocean environment. Moreover, local scour around OWT foundations under random waves has not yet been investigated fully. Therefore, further study are still needed in the aspect of scour around OWT foundations with low KC number under random waves. Given that, this study presents the scour sediment model around umbrella suction anchor foundation (USAF) under random waves. In this study, a comparison of equilibrium scour depth around USAF between this present numerical models and the previous theoretical models and experimental results was presented firstly. Then, this study gave a comprehensive analysis for the scour mechanisms around USAF. After that, two revised models were proposed according to the model of Raaijmakers and Rudolph [34] and the stochastic model developed by Myrhaug and Rue [37] respectively to predict the equilibrium scour depth. Finally, a parametric study was conducted to study the effects of the Froude number (F_r) and Euler number (E_u) to equilibrium scour depth respectively.

2. Numerical Method

2.1. Governing Equations of Flow

The following equations adopted in present model are already available in Flow 3D software. The authors used these theoretical equations to simulate scour in random waves without modification. The incompressible viscous fluid motion satisfies the Reynolds-averaged Navier-Stokes (RANS) equation, so the present numerical model solves RANS equations:

∂u∂t+1VF(uAx∂u∂x+vAy∂u∂y+wAz∂u∂z)=−1ρf∂p∂x+Gx+fx∂�∂�+1��(��∂�∂�+��∂�∂�+��∂�∂�)=−1�f∂�∂�+��+��(9)

∂v∂t+1VF(uAx∂v∂x+vAy∂v∂y+wAz∂v∂z)=−1ρf∂p∂y+Gy+fy∂�∂�+1��(��∂�∂�+��∂�∂�+��∂�∂�)=−1�f∂�∂�+��+��(10)

∂w∂t+1VF(uAx∂w∂x+vAy∂w∂y+wAz∂w∂z)=−1ρf∂p∂z+Gz+fz∂�∂�+1��(��∂�∂�+��∂�∂�+��∂�∂�)=−1�f∂�∂�+��+��(11)

where, V_F is the volume fraction; u, v, and w are the velocity components in x, y, z direction respectively with Cartesian coordinates; A_i is the area fraction; ρ_f is the fluid density, f_i is the viscous fluid acceleration, G_i is the fluid body acceleration (i = x, y, z).

2.2. Turbulent Model

The turbulence closure is available by the turbulent model, such as one-equation, the one-equation k-ε model, the standard k-ε model, RNG k-ε turbulent model and large eddy simulation (LES) model. The LES model requires very fine mesh grid, so the computational time is large, which hinders the LES model application in engineering. The RNG k-ε model can reduce computational time greatly with high accuracy in the near-wall region. Furthermore, the RNG k-ε model computes the maximum turbulent mixing length dynamically in simulating sediment scour model. Therefore, the RNG k-ε model was adopted to study the scour around anchor under random waves [41,42].

∂kT∂T+1VF(uAx∂kT∂x+vAy∂kT∂y+wAz∂kT∂z)=PT+GT+DiffkT−εkT∂��∂�+1��(��∂��∂�+��∂��∂�+��∂��∂�)=��+��+��−��(12)

∂εT∂T+1VF(uAx∂εT∂x+vAy∂εT∂y+wAz∂εT∂z)=CDIS1εTkT(PT+CDIS3GT)+Diffε−CDIS2ε2TkT∂��∂�+1��(��∂��∂�+��∂��∂�+��∂��∂�)=��1��(��+��3��)+��−��2��2��(13)

where, k_T is specific kinetic energy involved with turbulent velocity, G_T is the turbulent energy generated by buoyancy; ε_T is the turbulent energy dissipating rate, P_T is the turbulent energy, Diff_ε and Diff_kT are diffusion terms associated with V_F, A_i; CDIS1, CDIS2 and CDIS3 are dimensionless parameters, and CDIS1, CDIS3 have default values of 1.42, 0.2 respectively. CDIS2 can be obtained from P_T and k_T.

2.3. Sediment Scour Model

The sand particles may suffer four processes under waves, i.e., entrainment, bed load transport, suspended load transport, and deposition, so the sediment scour model should depict the above processes efficiently. In present numerical simulation, the sediment scour model includes the following aspects:

2.3.1. Entrainment and Deposition

The combination of entrainment and deposition determines the net scour rate of seabed in present sediment scour model. The entrainment lift velocity of sand particles was calculated as [43]:

ulift,i=αinsd0.3∗(θ−θcr)1.5∥g∥di(ρi−ρf)ρf−−−−−−−−−−−−√�lift,i=��*0.3(�−�cr)1.5��(��−�f)�f(14)

where, α_i is the entrainment parameter, n_s is the outward point perpendicular to the seabed, d_* is the dimensionless diameter of sand particles, which was calculated by Equation (15), θ_cr is the critical Shields parameter, g is the gravity acceleration, d_i is the diameter of sand particles, ρ_i is the density of seabed species.

d∗=di(∥g∥ρf(ρi−ρf)μ2f)1/3�*=��(��f(��−�f)�f2)1/3(15)

where μ_f is the fluid dynamic viscosity.

In Equation (14), the entrainment parameter α_i confirms the rate at which sediment erodes when the given shear stress is larger than the critical shear stress, and the recommended value 0.018 was adopted according to the experimental data of Mastbergen and Von den Berg [43]. n_s is the outward pointing normal to the seabed interface, and n_s = (0,0,1) according to the Cartesian coordinates used in present numerical model.

The shields parameter was obtained from the following equation:

θ=U2f,m(ρi/ρf−1)gd50�=�f,m2(��/�f−1)��50(16)

where, U_f,m is the maximum value of the near-bed friction velocity; d₅₀ is the median diameter of sand particles. The detailed calculation procedure of θ was available in Soulsby [44].

The critical shields parameter θ_cr was obtained from the Equation (17) [44]

θcr=0.31+1.2d∗+0.055[1−exp(−0.02d∗)]�cr=0.31+1.2�*+0.0551−exp(−0.02�*)(17)

The sand particles begin to deposit on seabed when the turbulence energy weaken and cann’t support the particles suspending. The setting velocity of the particles was calculated from the following equation [44]:

usettling,i=νfdi[(10.362+1.049d3∗)0.5−10.36]�settling,�=�f��(10.362+1.049�*3)0.5−10.36(18)

where ν_f is the fluid kinematic viscosity.

2.3.2. Bed Load Transport

This is called bed load transport when the sand particles roll or bounce over the seabed and always have contact with seabed. The bed load transport velocity was computed by [45]:

ubedload,i=qb,iδicb,ifb�bedload,�=�b,��b,��b(19)

where, q_b,i is the bed load transport rate, which was obtained from Equation (20), δ_i is the bed load thickness, which was calculated by Equation (21), c_b,i is the volume fraction of sand i in the multiple species, f_b is the critical packing fraction of the seabed.

qb,i=8[∥g∥(ρi−ρfρf)d3i]1/2�b,�=8�(��−�f�f)��31/2(20)

δi=0.3d0.7∗(θθcr−1)0.5di��=0.3�*0.7(��cr−1)0.5��(21)

2.3.3. Suspended Load Transport

Through the following transport equation, the suspended sediment concentration could be acquired.

∂Cs,i∂t+∇(us,iCs,i)=∇∇(DfCs,i)∂�s,�∂�+∇(�s,��s,�)=∇∇(�f�s,�)(22)

where, C_s,i is the suspended sand particles mass concentration of sand i in the multiple species, u_s,i is the sand particles velocity of sand i, D_f is the diffusivity.

The velocity of sand i in the multiple species could be obtained from the following equation:

us,i=u¯¯+usettling,ics,i�s,�=�¯+�settling,��s,�(23)

where, u¯�¯ is the velocity of mixed fluid-particles, which can be calculated by the RANS equation with turbulence model, c_s,i is the suspended sand particles volume concentration, which was computed from Equation (24).

cs,i=Cs,iρi�s,�=�s,��(24)

3. Model Setup

The seabed-USAF-wave three-dimensional scour numerical model was built using Flow-3D software. As shown in Figure 2, the model includes sandy seabed, USAF model, sea water, two baffles and porous media. The dimensions of USAF are shown in Table 1. The sandy bed (210 m in length, 30 m in width and 11 m in height) is made up of uniform fine sand with median diameter d₅₀ = 0.041 cm. The USAF model includes upper steel tube with the length of 20 m, which was installed in the middle of seabed. The location of USAF is positioned at 140 m from the upstream inflow boundary and 70 m from the downstream outflow boundary. Two baffles were installed at two ends of seabed. In order to eliminate the wave reflection basically, the porous media was set at the outflow side on the seabed.

Figure 2. (a) The sketch of seabed-USAF-wave three-dimensional model; (b) boundary condation:Wv-wave boundary, S-symmetric boundary, O-outflow boundary; (c) USAF model.

Table 1. Numerical simulating cases.

3.1. Mesh Geometric Dimensions

In the simulation of the scour under the random waves, the model includes the umbrella suction anchor foundation, seabed and fluid. As shown in Figure 3, the model mesh includes global mesh grid and nested mesh grid, and the total number of grids is 1,812,000. The basic procedure for building mesh grid consists of two steps. Step 1: Divide the global mesh using regular hexahedron with size of 0.6 × 0.6. The global mesh area is cubic box, embracing the seabed and whole fluid volume, and the dimensions are 210 m in length, 30 m in width and 32 m in height. The details of determining the grid size can see the following mesh sensitivity section. Step 2: Set nested fine mesh grid in vicinity of the USAF with size of 0.3 × 0.3 so as to shorten the computation cost and improve the calculation accuracy. The encryption range is −15 m to 15 m in x direction, −15 m to 15 m in y direction and 0 m to 32 m in z direction, respectively. In order to accurately capture the free-surface dynamics, such as the fluid-air interface, the volume of fluid (VOF) method was adopted for tracking the free water surface. One specific algorithm called FAVORTM (Fractional Area/Volume Obstacle Representation) was used to define the fractional face areas and fractional volumes of the cells which are open to fluid flow.

Figure 3. The sketch of mesh grid.

3.2. Boundary Conditions

As shown in Figure 2, the initial fluid length is 210 m as long as seabed. A wave boundary was specified at the upstream offshore end. The details of determining the random wave spectrum can see the following wave parameters section. The outflow boundary was set at the downstream onshore end. The symmetry boundary was used at the top and two sides of the model. The symmetric boundaries were the better strategy to improve the computation efficiency and save the calculation cost [46]. At the seabed bottom, the wall boundary was adopted, which means the u = v = w= 0. Besides, the upper steel tube of USAF was set as no-slip condition.

3.3. Wave Parameters

The random waves with JONSWAP wave spectrum were used for all simulations as realistic representation of offshore conditions. The unidirectional JONSWAP frequency spectrum was described as [47]:

S(ω)=αg2ω5exp[−54(ωpω)4]γexp[−(ω−ωp)22σ2ω2p]�(�)=��2�5exp−54(�p�)4�exp−(�−�p)22�2�p2(25)

where, α is wave energy scale parameter, which is calculated by Equation (26), ω is frequency, ω_p is wave spectrum peak frequency, which can be obtained from Equation (27). γ is wave spectrum peak enhancement factor, in this study γ = 3.3. σ is spectral width factor, σ equals 0.07 for ω ≤ ω_p and 0.09 for ω > ω_p respectively.

α=0.0076(gXU2)−0.22�=0.0076(��2)−0.22(26)

ωp=22(gU)(gXU2)−0.33�p=22(��)(��2)−0.33(27)

where, X is fetch length, U is average wind velocity at 10 m height from mean sea level.

In present numerical model, the input key parameters include X and U for wave boundary with JONSWAP wave spectrum. The objective wave height and period are available by different combinations of X and U. In this study, we designed 9 cases with different wave heights, periods and water depths for simulating scour around USAF under random waves (see Table 2). For random waves, the wave steepness ε and Ursell number U_r were acquired form Equations (28) and (29) respectively

ε=2πgHsT2a�=2��s�a2(28)

Ur=Hsk2h3w�r=�s�2ℎw3(29)

where, H_s is significant wave height, T_a is average wave period, k is wave number, h_w is water depth. The Shield parameter θ satisfies θ _> θ_cr for all simulations in current study, indicating the live bed scour prevails.

Table 2. Numerical simulating cases.

3.4. Mesh Sensitivity

In this section, a mesh sensitivity analysis was conducted to investigate the influence of mesh grid size to results and make sure the calculation is mesh size independent and converged. Three mesh grid size were chosen: Mesh 1—global mesh grid size of 0.75 × 0.75, nested fine mesh grid size of 0.4 × 0.4, and total number of grids 1,724,000, Mesh 2—global mesh grid size of 0.6 × 0.6, nested fine mesh grid size of 0.3 × 0.3, and total number of grids 1,812,000, Mesh 3—global mesh grid size of 0.4 × 0.4, nested fine mesh grid size of 0.2 × 0.2, and total number of grids 1,932,000. The near-bed shear velocity U_* is an important factor for influencing scour process [1,15], so U_* at the position of (4,0,11.12) was evaluated under three mesh sizes. As the Figure 4 shown, the maximum error of shear velocity ∆U_*1,2 is about 39.8% between the mesh 1 and mesh 2, and 4.8% between the mesh 2 and mesh 3. According to the mesh sensitivity criterion adopted by Pang et al. [48], it’s reasonable to think the results are mesh size independent and converged with mesh 2. Additionally, the present model was built according to prototype size, and the mesh size used in present model is larger than the mesh size adopted by Higueira et al. [49] and Corvaro et al. [50]. If we choose the smallest cell size, it will take too much time. For example, the simulation with Mesh3 required about 260 h by using a computer with Intel Xeon Scalable Gold 4214 CPU @24 Cores, 2.2 GHz and 64.00 GB RAM. Therefore, in this case, considering calculation accuracy and computation efficiency, the mesh 2 was chosen for all the simulation in this study.

Figure 4. Comparison of near-bed shear velocity U_* with different mesh grid size.

The nested mesh block was adopted for seabed in vicinity of the USAF, which was overlapped with the global mesh block. When two mesh blocks overlap each other, the governing equations are by default solved on the mesh block with smaller average cell size (i.e., higher grid resolution). It is should be noted that the Flow 3D software used the moving mesh captures the scour evolution and automatically adjusts the time step size to be as large as possible without exceeding any of the stability limits, affecting accuracy, or unduly increasing the effort required to enforce the continuity condition [51].

3.5. Model Validation

In order to verify the reliability of the present model, the results of present study were compared with the experimental data of Khosronejad et al. [52]. The experiment was conducted in an open channel with a slender vertical pile under unidirectional currents. The comparison of scour development between the present results and the experimental results is shown in Figure 5. The Figure 5 reveals that the present results agree well with the experimental data of Khosronejad et al. [52]. In the first stage, the scour depth increases rapidly. After that, the scour depth achieves a maximum value gradually. The equilibrium scour depth calculated by the present model is basically corresponding with the experimental results of Khosronejad et al. [52], although scour depth in the present model is slightly larger than the experimental results at initial stage.

Figure 5. Comparison of time evolution of scour between the present study and Khosronejad et al. [52], Petersen et al. [17].

Secondly, another comparison was further conducted between the results of present study and the experimental data of Petersen et al. [17]. The experiment was carried out in a flume with a circular vertical pile in combined waves and current. Figure 4 shows a comparison of time evolution of scour depth between the simulating and the experimental results. As Figure 5 indicates, the scour depth in this study has good overall agreement with the experimental results proposed in Petersen et al. [17]. The equilibrium scour depth calculated by the present model is 0.399 m, which equals to the experimental value basically. Overall, the above verifications prove the present model is accurate and capable in dealing with sediment scour under waves.

In addition, in order to calibrate and validate the present model for hydrodynamic parameters, the comparison of water surface elevation was carried out with laboratory experiments conducted by Stahlmann [53] for wave gauge No. 3. The Figure 6 depicts the surface wave profiles between experiments and numerical model results. The comparison indicates that there is a good agreement between the model results and experimental values, especially the locations of wave crest and trough. Comparison of the surface elevation instructs the present model has an acceptable relative error, and the model is a calibrated in terms of the hydrodynamic parameters.

Figure 6. Comparison of surface elevation between the present study and Stahlmann [53].

Finally, another comparison was conducted for equilibrium scour depth or maximum scour depth under random waves with the experimental data of Sumer and Fredsøe [16] and Schendel et al. [22]. The Figure 7 shows the comparison between the numerical results and experimental data of Run01, Run05, Run21 and Run22 in Sumer and Fredsøe [16] and test A05 and A09 in Schendel et al. [22]. As shown in Figure 7, the equilibrium scour depth or maximum scour depth distributed within the ±30 error lines basically, meaning the reliability and accuracy of present model for predicting equilibrium scour depth around foundation in random waves. However, compared with the experimental values, the present model overestimated the equilibrium scour depth generally. Given that, a calibration for scour depth was carried out by multiplying the mean reduced coefficient 0.85 in following section.

Figure 7. Comparison of equilibrium (or maximum) scour depth between the present study and Sumer and Fredsøe [16], Schendel et al. [22].

Through the various examination for hydrodynamic and morphology parameters, it can be concluded that the present model is a validated and calibrated model for scour under random waves. Thus, the present numerical model would be utilized for scour simulation around foundation under random waves.

4. Numerical Results and Discussions

4.1. Scour Evolution

Figure 8 displays the scour evolution for case 1–9. As shown in Figure 8a, the scour depth increased rapidly at the initial stage, and then slowed down at the transition stage, which attributes to the backfilling occurred in scour holes under live bed scour condition, resulting in the net scour decreasing. Finally, the scour reached the equilibrium state when the amount of sediment backfilling equaled to that of scouring in the scour holes, i.e., the net scour transport rate was nil. Sumer and Fredsøe [16] proposed the following formula for the scour development under waves

St=Seq(1−exp(−t/Tc))�t=�eq(1−exp(−�/�c))(30)

where T_c is time scale of scour process.

Figure 8. Time evolution of scour for case 1–9: (a) Case 1–5; (b) Case 6–9.

The computing time is 3600 s and the scour development curves in Figure 8 kept fluctuating, meaning it’s still not in equilibrium scour stage in these cases. According to Sumer and Fredsøe [16], the equilibrium scour depth can be acquired by fitting the data with Equation (30). From Figure 8, it can be seen that the scour evolution obtained from Equation (30) is consistent with the present study basically at initial stage, but the scour depth predicted by Equation (30) developed slightly faster than the simulating results and the Equation (30) overestimated the scour depth to some extent. Overall, the whole tendency of the results calculated by Equation (30) agrees well with the simulating results of the present study, which means the Equation (30) is applicable to depict the scour evolution around USAF under random waves.

4.2. Scour Mechanism under Random Waves

The scour morphology and scour evolution around USAF are similar under random waves in case 1~9. Taking case 7 as an example, the scour morphology is shown in Figure 9.

Figure 9. Scour morphology under different times for case 7.

From Figure 9, at the initial stage (t < 1200 s), the scour occurred at upstream foundation edges between neighboring anchor branches. The maximum scour depth appeared at the lee-side of the USAF. Correspondingly, the sediments deposited at the periphery of the USAF, and the location of the maximum accretion depth was positioned at an angle of about 45° symmetrically with respect to the wave propagating direction in the lee-side of the USAF. After that, when t > 2400 s, the location of the maximum scour depth shifted to the upside of the USAF at an angle of about 45° with respect to the wave propagating direction.

According to previous studies [1,15,16,19,30,31], the horseshoe vortex, streamline compression and wake vortex shedding were responsible for scour around foundation. The Figure 10 displays the distribution of flow velocity in vicinity of foundation, which reflects the evolving processes of horseshoe vertex.

Figure 10. Velocity profile around USAF: (a) Flow runup and down stream at upstream anchor edges; (b) Horseshoe vortex at upstream anchor edges; (c) Flow reversal during wave through stage at lee side.

As shown in Figure 10, the inflow tripped to the upstream edges of the USAF and it was blocked by the upper tube of USAF. Then, the downflow formed the horizontal axis clockwise vortex and rolled on the seabed bypassing the tube, that is, the horseshoe vortex (Figure 11). The Figure 12 displays the turbulence intensity around the tube on the seabed. From Figure 12, it can be seen that the turbulence intensity was high-intensity with respect to the region of horseshoe vortex. This phenomenon occurred because of drastic water flow momentum exchanging in the horseshoe vortex. As a result, it created the prominent shear stress on the seabed, causing the local scour at the upstream edges of USAF. Besides, the horseshoe vortex moved downstream gradually along the periphery of the tube and the wake vortex shed off continually at the lee-side of the USAF, i.e., wake vortex.

Figure 11. Sketch of scour mechanism around USAF under random waves.

Figure 12. Turbulence intensity: (a) Turbulence intensity of horseshoe vortex; (b) Turbulence intensity of wake vortex; (c) Turbulence intensity of accretion area.

The core of wake vortex is a negative pressure center, liking a vacuum cleaner [11,42]. Hence, the soil particles were swirled into the negative pressure core and carried away by wake vortex. At the same time, the onset of scour at rear side occurred. Finally, the wake vortex became downflow at the downside of USAF. As is shown in Figure 12, the turbulence intensity was low where the downflow occurred at lee-side, which means the turbulence energy may not be able to support the survival of wake vortex, leading to accretion happening. As mentioned in previous section, the formation of horseshoe vortex was dependent with adverse pressure gradient at upside of foundation. As shown in Figure 13, the evaluated range of pressure distribution is −15 m to 15 m in x direction. The t = 450 s and t = 1800 s indicate that the wave crest and trough arrived at the upside and lee-side of the foundation respectively, and the t = 350 s was neither the wave crest nor trough. The adverse gradient pressure reached the maximum value at t = 450 s corresponding to the wave crest phase. In this case, it’s helpful for the wave boundary separating fully from seabed, which leads to the formation of horseshoe vortex with high turbulence intensity. Therefore, the horseshoe vortex is responsible for the local scour between neighboring anchor branches at upside of USAF. What’s more, due to the combination of the horseshoe vortex and streamline compression, the maximum scour depth occurred at the upside of the USAF with an angle of about 45° corresponding to the wave propagating direction. This is consistent with the findings of Pang et al. [48] and Sumer et al. [1,15] in case of regular waves. At the wave trough phase (t = 1800 s), the pressure gradient became positive at upstream USAF edges, which hindered the separating of wave boundary from seabed. In the meantime, the flow reversal occurred (Figure 10) and the adverse gradient pressure appeared at downstream USAF edges, but the magnitude of adverse gradient pressure at lee-side was lower than the upstream gradient pressure under wave crest. In this way, the intensity of horseshoe vortex behind the USAF under wave trough was low, which explains the difference of scour depth at upstream and downstream, i.e., the scour asymmetry. In other words, the scour asymmetry at upside and downside of USAF was attributed to wave asymmetry for random waves, and the phenomenon became more evident for nonlinear waves [21]. Briefly speaking, the vortex system at wave crest phase was mainly related to the scour process around USAF under random waves.

Figure 13. Pressure distribution around USAF.

4.3. Equilibrium Scour Depth

The KC number is a key parameter for horseshoe vortex emerging and evolving under waves. According to Equation (1), when pile diameter D is fixed, the KC depends on the maximum near-bed velocity U_wm and wave period T. For random waves, the U_wm can be denoted by the root-mean-square (RMS) value of near-bed velocity amplitude U_wm,rms or the significant value of near-bed velocity amplitude U_wm,s. The U_wm,rms and U_wm,s for all simulating cases of the present study are listed in Table 3 and Table 4. The T can be denoted by the mean up zero-crossing wave period T_a, peak wave period T_p, significant wave period T_s, the maximum wave period T_m, 1/10′th highest wave period T_{n = 1/10} and 1/5′th highest wave period T_{n = 1/5} for random waves, so the different combinations of U_wm and T will acquire different KC. The Table 3 and Table 4 list 12 types of KC, for example, the KC_rms,s was calculated by U_wm,rms and T_s. Sumer and Fredsøe [16] conducted a series of wave flume experiments to investigate the scour depth around monopile under random waves, and found the equilibrium scour depth predicting equation (Equation (2)) for regular waves was applicable for random waves with KC_rms,p. It should be noted that the Equation (2) is only suitable for KC > 6 under regular waves or KC_rms,p > 6 under random waves.

Table 3. U_wm,rms and KC for case 1~9.

Table 4. U_wm,s and KC for case 1~9.

Raaijmakers and Rudolph [34] proposed the equilibrium scour depth predicting model (Equation (5)) around pile under waves, which is suitable for low KC. The format of Equation (5) is similar with the formula proposed by Breusers [54], which can predict the equilibrium scour depth around pile at different scour stages. In order to verify the applicability of Raaijmakers’s model for predicting the equilibrium scour depth around USAF under random waves, a validation of the equilibrium scour depth S_eq between the present study and Raaijmakers’s equation was conducted. The position where the scour depth S_eq was evaluated is the location of the maximum scour depth, and it was depicted in Figure 14. The Figure 15 displays the comparison of S_eq with different KC between the present study and Raaijmakers’s model.

Figure 14. Sketch of the position where the S_eq was evaluated.

Figure 15. Comparison of the equilibrium scour depth between the present model and the model of Raaijmakers and Rudolph [34]: (a) KC_rms,s, KC_rms,a; (b) KC_rms,p, KC_rms,m; (c) KC_{rms,n = 1/10}, KC_{rms,n = 1/5}; (d) KC_s,s, KC_s,a; (e) KC_s,p, KC_s,m; (f) KC_{s,n = 1/10}, KC_{s,n = 1/5}.

As shown in Figure 15, there is an error in predicting S_eq between the present study and Raaijmakers’s model, and Raaijmakers’s model underestimates the results generally. Although the error exists, the varying trend of S_eq with KC obtained from Raaijmakers’s model is consistent with the present study basically. What’s more, the error is minimum and the Raaijmakers’s model is of relatively high accuracy for predicting scour around USAF under random waves by using KC_s,p. Based on this, a further revision was made to eliminate the error as much as possible, i.e., add the deviation value ∆S/D in the Raaijmakers’s model. The revised equilibrium scour depth predicting equation based on Raaijmakers’s model can be written as

S′eq/D=1.95[tanh(hD)](1−exp(−0.012KCs,p))+ΔS/D�eq′/�=1.95tanh(ℎ�)(1−exp(−0.012��s,p))+∆�/�(31)

As the Figure 16 shown, through trial-calculation, when ∆S/D = 0.05, the results calculated by Equation (31) show good agreement with the simulating results of the present study. The maximum error is about 18.2% and the engineering requirements have been met basically. In order to further verify the accuracy of the revised model for large KC (KC_s,p > 4) under random waves, a validation between the revised model and the previous experimental results [21]. The experiment was conducted in a flume (50 m in length, 1.0 m in width and 1.3 m in height) with a slender vertical pile (D = 0.1 m) under random waves. The seabed is composed of 0.13 m deep layer of sand with d₅₀ = 0.6 mm and the water depth is 0.5 m for all tests. The significant wave height is 0.12~0.21 m and the KC_s,p is 5.52~11.38. The comparison between the predicting results by Equation (31) and the experimental results of Corvaro et al. [21] is shown in Figure 17. From Figure 17, the experimental data evenly distributes around the predicted results and the prediction accuracy is favorable when KC_s,p < 8. However, the gap between the predicting results and experimental data becomes large and the Equation (31) overestimates the equilibrium scour depth to some extent when KC_s,p > 8.

Figure 16. Comparison of S_eq between the simulating results and the predicting values by Equation (31).

Figure 17. Comparison of S_eq/D between the Experimental results of Corvaro et al. [21] and the predicting values by Equation (31).

In ocean environment, the waves are composed of a train of sinusoidal waves with different frequencies and amplitudes. The energy of constituent waves with very large and very small frequencies is relatively low, and the energy of waves is mainly concentrated in a certain range of moderate frequencies. Myrhaug and Rue [37] thought the 1/n’th highest wave was responsible for scour and proposed the stochastic model to predict the equilibrium scour depth around pile under random waves for full range of KC. Noteworthy is that the KC was denoted by KC_rms,a in the stochastic model. To verify the application of the stochastic model for predicting scour depth around USAF, a validation between the simulating results of present study and predicting results by the stochastic model with n = 2,3,5,10,20,500 was carried out respectively.

As shown in Figure 18, compared with the simulating results, the stochastic model underestimates the equilibrium scour depth around USAF generally. Although the error exists, the varying trend of S_eq with KC_rms,a obtained from the stochastic model is consistent with the present study basically. What’s more, the gap between the predicting values by stochastic model and the simulating results decreases with the increase of n, but for large n, for example n = 500, the varying trend diverges between the predicting values and simulating results, meaning it’s not feasible only by increasing n in stochastic model to predict the equilibrium scour depth around USAF.

Figure 18. Comparison of S_eq between the simulating results and the predicting values by Equation (8).

The Figure 19 lists the deviation value ∆S_eq/D′ between the predicting values and simulating results with different KC_rms,a and n. Then, fitted the relationship between the ∆S′and n under different KC_rms,a, and the fitting curve can be written by Equation (32). The revised stochastic model (Equation (33)) can be acquired by adding ∆S_eq/D′ to Equation (8).

ΔSeq/D=0.052*exp(−n/6.566)+0.068∆�eq/�=0.052*exp(−�/6.566)+0.068(32)

S′eq¯/D=S′eq/D+0.052*exp(−n/6.566)+0.068�eq′¯/�=�eq′/�+0.052*exp(−�/6.566)+0.068(33)

Figure 19. The fitting line between ∆S′and n.

The comparison between the predicting results by Equation (33) and the simulating results of present study is shown in Figure 20. According to the Figure 20, the varying trend of S_eq with KC_rms,a obtained from the stochastic model is consistent with the present study basically. Compared with predicting results by the stochastic model, the results calculated by Equation (33) is favorable. Moreover, comparison with simulating results indicates that the predicting results are the most favorable for n = 10, which is consistent with the findings of Myrhaug and Rue [37] for equilibrium scour depth predicting around slender pile in case of random waves.

Figure 20. Comparison of S_eq between the simulating results and the predicting values by Equation (33).

In order to further verify the accuracy of the Equation (33) for large KC (KC_rms,a > 4) under random waves, a validation was conducted between the Equation (33) and the previous experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21]. The details of experiments conducted by Corvaro et al. [21] were described in above section. Sumer and Fredsøe [16] investigated the local scour around pile under random waves. The experiments were conducted in a wave basin with a slender vertical pile (D = 0.032, 0.055 m). The seabed is composed of 0.14 m deep layer of sand with d₅₀ = 0.2 mm and the water depth was maintained at 0.5 m. The JONSWAP wave spectrum was used and the KC_rms,a was 5.29~16.95. The comparison between the predicting results by Equation (33) and the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21] are shown in Figure 21. From Figure 21, contrary to the case of low KC_rms,a (KC_rms,a < 4), the error between the predicting values and experimental results increases with decreasing of n for KC_rms,a > 4. Therefore, the predicting results are the most favorable for n = 2 when KC_rms,a > 4.

Figure 21. Comparison of S_eq between the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21] and the predicting values by Equation (33).

Noteworthy is that the present model was built according to prototype size, so the errors between the numerical results and experimental data of References [16,21] may be attribute to the scale effects. In laboratory experiments on scouring process, it is typically impossible to ensure a rigorous similarity of all physical parameters between the model and prototype structure, leading to the scale effects in the laboratory experiments. To avoid a cohesive behaviour, the bed material was not scaled geometrically according to model scale. As a consequence, the relatively large-scaled sediments sizes may result in the overestimation of bed load transport and underestimation of suspended load transport compared with field conditions. What’s more, the disproportional scaled sediment presumably lead to the difference of bed roughness between the model and prototype, and thus large influences for wave boundary layer on the seabed and scour process. Besides, according to Corvaro et al. [21] and Schendel et al. [55], the pile Reynolds numbers and Froude numbers both affect the scour depth for the condition of non fully developed turbulent flow in laboratory experiments.

4.4. Parametric Study

4.4.1. Influence of Froude Number

As described above, the set of foundation leads to the adverse pressure gradient appearing at upstream, leading to the wave boundary layer separating from seabed, then horseshoe vortex formatting and the horseshoe vortex are mainly responsible for scour around foundation (see Figure 22). The Froude number F_r is the key parameter to influence the scale and intensity of horseshoe vortex. The F_r under waves can be calculated by the following formula [42]

Fr=UwgD−−−√�r=�w��(34)

where U_w is the mean water particle velocity during 1/4 cycle of wave oscillation, obtained from the following formula. Noteworthy is that the root-mean-square (RMS) value of near-bed velocity amplitude U_wm,rms is used for calculating U_wm.

Uw=1T/4∫0T/4Uwmsin(t/T)dt=2πUwm�w=1�/4∫0�/4�wmsin(�/�)��=2��wm(35)

Figure 22. Sketch of flow field at upstream USAF edges.

Tavouktsoglou et al. [25] proposed the following formula between F_r and the vertical location of the stagnation y

yh∝Fer�ℎ∝�r�(36)

where e is constant.

The Figure 23 displays the relationship between S_eq/D and F_r of the present study. In order to compare with the simulating results, the experimental data of Corvaro et al. [21] was also depicted in Figure 23. As shown in Figure 23, the equilibrium scour depth appears a logarithmic increase as F_r increases and approaches the mathematical asymptotic value, which is also consistent with the experimental results of Corvaro et al. [21]. According to Figure 24, the adverse pressure gradient pressure at upstream USAF edges increases with the increase of F_r, which is benefit for the wave boundary layer separating from seabed, resulting in the high-intensity horseshoe vortex, hence, causing intensive scour around USAF. Based on the previous study of Tavouktsoglou et al. [25] for scour around pile under currents, the high F_r leads to the stagnation point is closer to the mean sea level for shallow water, causing the stronger downflow kinetic energy. As mentioned in previous section, the energy of downflow at upstream makes up the energy of the subsequent horseshoe vortex, so the stronger downflow kinetic energy results in the more intensive horseshoe vortex. Therefore, the higher F_r leads to the more intensive horseshoe vortex by influencing the position of stagnation point y presumably. Qi and Gao [19] carried out a series of flume tests to investigate the scour around pile under regular waves, and proposed the fitting formula between S_eq/D and F_r as following

lg(Seq/D)=Aexp(B/Fr)+Clg(�eq/�)=�exp(�/�r)+�(37)

where A, B and C are constant.

Figure 23. The fitting curve between S_eq/D and F_r.

Figure 24. Sketch of adverse pressure gradient at upstream USAF edges.

Took the Equation (37) to fit the simulating results with A = −0.002, B = 0.686 and C = −0.808, and the results are shown in Figure 23. From Figure 23, the simulating results evenly distribute around the Equation (37) and the varying trend of S_eq/D and F_r in present study is consistent with Equation (37) basically, meaning the Equation (37) is applicable to express the relationship of S_eq/D with F_r around USAF under random waves.

4.4.2. Influence of Euler Number

The Euler number E_u is the influencing factor for the hydrodynamic field around foundation. The E_u under waves can be calculated by the following formula. The E_u can be represented by the Equation (38) for uniform cylinders [25]. The root-mean-square (RMS) value of near-bed velocity amplitude U_m,rms is used for calculating U_m.

Eu=U2mgD�u=�m2��(38)

where U_m is depth-averaged flow velocity.

The Figure 25 displays the relationship between S_eq/D and E_u of the present study. In order to compare with the simulating results, the experimental data of Sumer and Fredsøe [16] and Corvaro et al. [21] were also plotted in Figure 25. As shown in Figure 25, similar with the varying trend of S_eq/D and F_r, the equilibrium scour depth appears a logarithmic increase as E_u increases and approaches the mathematical asymptotic value, which is also consistent with the experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21]. According to Figure 24, the adverse pressure gradient pressure at upstream USAF edges increases with the increasing of E_u, which is benefit for the wave boundary layer separating from seabed, inducing the high-intensity horseshoe vortex, hence, causing intensive scour around USAF.

Figure 25. The fitting curve between S_eq/D and E_u.

Therefore, the variation of F_r and E_u reflect the magnitude of adverse pressure gradient pressure at upstream. Given that, the Equation (37) also was used to fit the simulating results with A = 8.875, B = 0.078 and C = −9.601, and the results are shown in Figure 25. From Figure 25, the simulating results evenly distribute around the Equation (37) and the varying trend of S_eq/D and E_u in present study is consistent with Equation (37) basically, meaning the Equation (37) is also applicable to express the relationship of S_eq/D with E_u around USAF under random waves. Additionally, according to the above description of F_r, it can be inferred that the higher F_r and E_u both lead to the more intensive horseshoe vortex by influencing the position of stagnation point y presumably.

5. Conclusions

A series of numerical models were established to investigate the local scour around umbrella suction anchor foundation (USAF) under random waves. The numerical model was validated for hydrodynamic and morphology parameters by comparing with the experimental data of Khosronejad et al. [52], Petersen et al. [17], Sumer and Fredsøe [16] and Schendel et al. [22]. Based on the simulating results, the scour evolution and scour mechanisms around USAF under random waves were analyzed respectively. Two revised models were proposed according to the model of Raaijmakers and Rudolph [34] and the stochastic model developed by Myrhaug and Rue [37] to predict the equilibrium scour depth around USAF under random waves. Finally, a parametric study was carried out with the present model to study the effects of the Froude number F_r and Euler number E_u to the equilibrium scour depth around USAF under random waves. The main conclusions can be described as follows.(1)

The packed sediment scour model and the RNG k−ε turbulence model were used to simulate the sand particles transport processes and the flow field around UASF respectively. The scour evolution obtained by the present model agrees well with the experimental results of Khosronejad et al. [52], Petersen et al. [17], Sumer and Fredsøe [16] and Schendel et al. [22], which indicates that the present model is accurate and reasonable for depicting the scour morphology around UASF under random waves.(2)

The vortex system at wave crest phase is mainly related to the scour process around USAF under random waves. The maximum scour depth appeared at the lee-side of the USAF at the initial stage (t < 1200 s). Subsequently, when t > 2400 s, the location of the maximum scour depth shifted to the upside of the USAF at an angle of about 45° with respect to the wave propagating direction.(3)

The error is negligible and the Raaijmakers’s model is of relatively high accuracy for predicting scour around USAF under random waves when KC is calculated by KC_s,p. Given that, a further revision model (Equation (31)) was proposed according to Raaijmakers’s model to predict the equilibrium scour depth around USAF under random waves and it shows good agreement with the simulating results of the present study when KC_s,p < 8.(4)

Another further revision model (Equation (33)) was proposed according to the stochastic model established by Myrhaug and Rue [37] to predict the equilibrium scour depth around USAF under random waves, and the predicting results are the most favorable for n = 10 when KC_rms,a < 4. However, contrary to the case of low KC_rms,a, the predicting results are the most favorable for n = 2 when KC_rms,a > 4 by the comparison with experimental results of Sumer and Fredsøe [16] and Corvaro et al. [21].(5)

The same formula (Equation (37)) is applicable to express the relationship of S_eq/D with E_u or F_r, and it can be inferred that the higher F_r and E_u both lead to the more intensive horseshoe vortex and larger S_eq.

Author Contributions

Conceptualization, H.L. (Hongjun Liu); Data curation, R.H. and P.Y.; Formal analysis, X.W. and H.L. (Hao Leng); Funding acquisition, X.W.; Writing—original draft, R.H. and P.Y.; Writing—review & editing, X.W. and H.L. (Hao Leng); The final manuscript has been approved by all the authors. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Fundamental Research Funds for the Central Universities (grant number 202061027) and the National Natural Science Foundation of China (grant number 41572247).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

Sumer, B.M.; Fredsøe, J.; Christiansen, N. Scour Around Vertical Pile in Waves. J. Waterw. Port. Coast. Ocean Eng. 1992, 118, 15–31. [Google Scholar] [CrossRef]
Rudolph, D.; Bos, K. Scour around a monopile under combined wave-current conditions and low KC-numbers. In Proceedings of the 6th International Conference on Scour and Erosion, Amsterdam, The Netherlands, 1–3 November 2006; pp. 582–588. [Google Scholar]
Nielsen, A.W.; Liu, X.; Sumer, B.M.; Fredsøe, J. Flow and bed shear stresses in scour protections around a pile in a current. Coast. Eng. 2013, 72, 20–38. [Google Scholar] [CrossRef]
Ahmad, N.; Bihs, H.; Myrhaug, D.; Kamath, A.; Arntsen, Ø.A. Three-dimensional numerical modelling of wave-induced scour around piles in a side-by-side arrangement. Coast. Eng. 2018, 138, 132–151. [Google Scholar] [CrossRef]
Li, H.; Ong, M.C.; Leira, B.J.; Myrhaug, D. Effects of Soil Profile Variation and Scour on Structural Response of an Offshore Monopile Wind Turbine. J. Offshore Mech. Arct. Eng. 2018, 140, 042001. [Google Scholar] [CrossRef]
Li, H.; Liu, H.; Liu, S. Dynamic analysis of umbrella suction anchor foundation embedded in seabed for offshore wind turbines. Géoméch. Energy Environ. 2017, 10, 12–20. [Google Scholar] [CrossRef]
Fazeres-Ferradosa, T.; Rosa-Santos, P.; Taveira-Pinto, F.; Vanem, E.; Carvalho, H.; Correia, J.A.F.D.O. Editorial: Advanced research on offshore structures and foundation design: Part 1. Proc. Inst. Civ. Eng. Marit. Eng. 2019, 172, 118–123. [Google Scholar] [CrossRef]
Chavez, C.E.A.; Stratigaki, V.; Wu, M.; Troch, P.; Schendel, A.; Welzel, M.; Villanueva, R.; Schlurmann, T.; De Vos, L.; Kisacik, D.; et al. Large-Scale Experiments to Improve Monopile Scour Protection Design Adapted to Climate Change—The PROTEUS Project. Energies 2019, 12, 1709. [Google Scholar] [CrossRef][Green Version]
Wu, M.; De Vos, L.; Chavez, C.E.A.; Stratigaki, V.; Fazeres-Ferradosa, T.; Rosa-Santos, P.; Taveira-Pinto, F.; Troch, P. Large Scale Experimental Study of the Scour Protection Damage Around a Monopile Foundation Under Combined Wave and Current Conditions. J. Mar. Sci. Eng. 2020, 8, 417. [Google Scholar] [CrossRef]
Sørensen, S.P.H.; Ibsen, L.B. Assessment of foundation design for offshore monopiles unprotected against scour. Ocean Eng. 2013, 63, 17–25. [Google Scholar] [CrossRef]
Prendergast, L.; Gavin, K.; Doherty, P. An investigation into the effect of scour on the natural frequency of an offshore wind turbine. Ocean Eng. 2015, 101, 1–11. [Google Scholar] [CrossRef][Green Version]
Fazeres-Ferradosa, T.; Chambel, J.; Taveira-Pinto, F.; Rosa-Santos, P.; Taveira-Pinto, F.; Giannini, G.; Haerens, P. Scour Protections for Offshore Foundations of Marine Energy Harvesting Technologies: A Review. J. Mar. Sci. Eng. 2021, 9, 297. [Google Scholar] [CrossRef]
Yang, Q.; Yu, P.; Liu, Y.; Liu, H.; Zhang, P.; Wang, Q. Scour characteristics of an offshore umbrella suction anchor foundation under the combined actions of waves and currents. Ocean Eng. 2020, 202, 106701. [Google Scholar] [CrossRef]
Yu, P.; Hu, R.; Yang, J.; Liu, H. Numerical investigation of local scour around USAF with different hydraulic conditions under currents and waves. Ocean Eng. 2020, 213, 107696. [Google Scholar] [CrossRef]
Sumer, B.M.; Christiansen, N.; Fredsøe, J. The horseshoe vortex and vortex shedding around a vertical wall-mounted cylinder exposed to waves. J. Fluid Mech. 1997, 332, 41–70. [Google Scholar] [CrossRef]
Sumer, B.M.; Fredsøe, J. Scour around Pile in Combined Waves and Current. J. Hydraul. Eng. 2001, 127, 403–411. [Google Scholar] [CrossRef]
Petersen, T.U.; Sumer, B.M.; Fredsøe, J. Time scale of scour around a pile in combined waves and current. In Proceedings of the 6th International Conference on Scour and Erosion, Paris, France, 27–31 August 2012. [Google Scholar]
Petersen, T.U.; Sumer, B.M.; Fredsøe, J.; Raaijmakers, T.C.; Schouten, J.-J. Edge scour at scour protections around piles in the marine environment—Laboratory and field investigation. Coast. Eng. 2015, 106, 42–72. [Google Scholar] [CrossRef]
Qi, W.; Gao, F. Equilibrium scour depth at offshore monopile foundation in combined waves and current. Sci. China Ser. E Technol. Sci. 2014, 57, 1030–1039. [Google Scholar] [CrossRef][Green Version]
Larsen, B.E.; Fuhrman, D.R.; Baykal, C.; Sumer, B.M. Tsunami-induced scour around monopile foundations. Coast. Eng. 2017, 129, 36–49. [Google Scholar] [CrossRef][Green Version]
Corvaro, S.; Marini, F.; Mancinelli, A.; Lorenzoni, C.; Brocchini, M. Hydro- and Morpho-dynamics Induced by a Vertical Slender Pile under Regular and Random Waves. J. Waterw. Port. Coast. Ocean Eng. 2018, 144, 04018018. [Google Scholar] [CrossRef]
Schendel, A.; Welzel, M.; Schlurmann, T.; Hsu, T.-W. Scour around a monopile induced by directionally spread irregular waves in combination with oblique currents. Coast. Eng. 2020, 161, 103751. [Google Scholar] [CrossRef]
Fazeres-Ferradosa, T.; Taveira-Pinto, F.; Romão, X.; Reis, M.; das Neves, L. Reliability assessment of offshore dynamic scour protections using copulas. Wind. Eng. 2018, 43, 506–538. [Google Scholar] [CrossRef]
Fazeres-Ferradosa, T.; Welzel, M.; Schendel, A.; Baelus, L.; Santos, P.R.; Pinto, F.T. Extended characterization of damage in rubble mound scour protections. Coast. Eng. 2020, 158, 103671. [Google Scholar] [CrossRef]
Tavouktsoglou, N.S.; Harris, J.M.; Simons, R.R.; Whitehouse, R.J.S. Equilibrium Scour-Depth Prediction around Cylindrical Structures. J. Waterw. Port. Coast. Ocean Eng. 2017, 143, 04017017. [Google Scholar] [CrossRef][Green Version]
Ettema, R.; Melville, B.; Barkdoll, B. Scale Effect in Pier-Scour Experiments. J. Hydraul. Eng. 1998, 124, 639–642. [Google Scholar] [CrossRef]
Umeda, S. Scour Regime and Scour Depth around a Pile in Waves. J. Coast. Res. Spec. Issue 2011, 64, 845–849. [Google Scholar]
Umeda, S. Scour process around monopiles during various phases of sea storms. J. Coast. Res. 2013, 165, 1599–1604. [Google Scholar] [CrossRef]
Baykal, C.; Sumer, B.; Fuhrman, D.R.; Jacobsen, N.; Fredsøe, J. Numerical simulation of scour and backfilling processes around a circular pile in waves. Coast. Eng. 2017, 122, 87–107. [Google Scholar] [CrossRef][Green Version]
Miles, J.; Martin, T.; Goddard, L. Current and wave effects around windfarm monopile foundations. Coast. Eng. 2017, 121, 167–178. [Google Scholar] [CrossRef][Green Version]
Miozzi, M.; Corvaro, S.; Pereira, F.A.; Brocchini, M. Wave-induced morphodynamics and sediment transport around a slender vertical cylinder. Adv. Water Resour. 2019, 129, 263–280. [Google Scholar] [CrossRef]
Yu, T.; Zhang, Y.; Zhang, S.; Shi, Z.; Chen, X.; Xu, Y.; Tang, Y. Experimental study on scour around a composite bucket foundation due to waves and current. Ocean Eng. 2019, 189, 106302. [Google Scholar] [CrossRef]
Carreiras, J.; Larroudé, P.; Seabra-Santos, F.; Mory, M. Wave Scour Around Piles. In Proceedings of the Coastal Engineering 2000, American Society of Civil Engineers (ASCE), Sydney, Australia, 16–21 July 2000; pp. 1860–1870. [Google Scholar]
Raaijmakers, T.; Rudolph, D. Time-dependent scour development under combined current and waves conditions—Laboratory experiments with online monitoring technique. In Proceedings of the 4th International Conference on Scour and Erosion, Tokyo, Japan, 5–7 November 2008; pp. 152–161. [Google Scholar]
Khalfin, I.S. Modeling and calculation of bed score around large-diameter vertical cylinder under wave action. Water Resour. 2007, 34, 357. [Google Scholar] [CrossRef][Green Version]
Zanke, U.C.; Hsu, T.-W.; Roland, A.; Link, O.; Diab, R. Equilibrium scour depths around piles in noncohesive sediments under currents and waves. Coast. Eng. 2011, 58, 986–991. [Google Scholar] [CrossRef]
Myrhaug, D.; Rue, H. Scour below pipelines and around vertical piles in random waves. Coast. Eng. 2003, 48, 227–242. [Google Scholar] [CrossRef]
Myrhaug, D.; Ong, M.C.; Føien, H.; Gjengedal, C.; Leira, B.J. Scour below pipelines and around vertical piles due to second-order random waves plus a current. Ocean Eng. 2009, 36, 605–616. [Google Scholar] [CrossRef]
Myrhaug, D.; Ong, M.C. Random wave-induced onshore scour characteristics around submerged breakwaters using a stochastic method. Ocean Eng. 2010, 37, 1233–1238. [Google Scholar] [CrossRef]
Ong, M.C.; Myrhaug, D.; Hesten, P. Scour around vertical piles due to long-crested and short-crested nonlinear random waves plus a current. Coast. Eng. 2013, 73, 106–114. [Google Scholar] [CrossRef]
Yakhot, V.; Orszag, S.A. Renormalization group analysis of turbulence. I. Basic theory. J. Sci. Comput. 1986, 1, 3–51. [Google Scholar] [CrossRef]
Yakhot, V.; Smith, L.M. The renormalization group, the e-expansion and derivation of turbulence models. J. Sci. Comput. 1992, 7, 35–61. [Google Scholar] [CrossRef]
Mastbergen, D.R.; Berg, J.V.D. Breaching in fine sands and the generation of sustained turbidity currents in submarine canyons. Sedimentology 2003, 50, 625–637. [Google Scholar] [CrossRef]
Soulsby, R. Dynamics of Marine Sands; Thomas Telford Ltd.: London, UK, 1998. [Google Scholar] [CrossRef]
Van Rijn, L.C. Sediment Transport, Part I: Bed Load Transport. J. Hydraul. Eng. 1984, 110, 1431–1456. [Google Scholar] [CrossRef][Green Version]
Zhang, Q.; Zhou, X.-L.; Wang, J.-H. Numerical investigation of local scour around three adjacent piles with different arrangements under current. Ocean Eng. 2017, 142, 625–638. [Google Scholar] [CrossRef]
Yu, Y.X.; Liu, S.X. Random Wave and Its Applications to Engineering, 4th ed.; Dalian University of Technology Press: Dalian, China, 2011. [Google Scholar]
Pang, A.; Skote, M.; Lim, S.; Gullman-Strand, J.; Morgan, N. A numerical approach for determining equilibrium scour depth around a mono-pile due to steady currents. Appl. Ocean Res. 2016, 57, 114–124. [Google Scholar] [CrossRef]
Higuera, P.; Lara, J.L.; Losada, I.J. Three-dimensional interaction of waves and porous coastal structures using Open-FOAM®. Part I: Formulation and validation. Coast. Eng. 2014, 83, 243–258. [Google Scholar] [CrossRef]
Corvaro, S.; Crivellini, A.; Marini, F.; Cimarelli, A.; Capitanelli, L.; Mancinelli, A. Experimental and Numerical Analysis of the Hydrodynamics around a Vertical Cylinder in Waves. J. Mar. Sci. Eng. 2019, 7, 453. [Google Scholar] [CrossRef][Green Version]
Flow3D User Manual, version 11.0.3; Flow Science, Inc.: Santa Fe, NM, USA, 2013.
Khosronejad, A.; Kang, S.; Sotiropoulos, F. Experimental and computational investigation of local scour around bridge piers. Adv. Water Resour. 2012, 37, 73–85. [Google Scholar] [CrossRef]
Stahlmann, A. Experimental and Numerical Modeling of Scour at Foundation Structures for Offshore Wind Turbines. Ph.D. Thesis, Franzius-Institute for Hydraulic, Estuarine and Coastal Engineering, Leibniz Universität Hannover, Hannover, Germany, 2013. [Google Scholar]
Breusers, H.N.C.; Nicollet, G.; Shen, H. Local Scour Around Cylindrical Piers. J. Hydraul. Res. 1977, 15, 211–252. [Google Scholar] [CrossRef]
Schendel, A.; Hildebrandt, A.; Goseberg, N.; Schlurmann, T. Processes and evolution of scour around a monopile induced by tidal currents. Coast. Eng. 2018, 139, 65–84. [Google Scholar] [CrossRef]

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Share and Cite

MDPI and ACS Style

Hu, R.; Liu, H.; Leng, H.; Yu, P.; Wang, X. Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves. J. Mar. Sci. Eng. 2021, 9, 886. https://doi.org/10.3390/jmse9080886

AMA Style

Hu R, Liu H, Leng H, Yu P, Wang X. Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves. Journal of Marine Science and Engineering. 2021; 9(8):886. https://doi.org/10.3390/jmse9080886Chicago/Turabian Style

Hu, Ruigeng, Hongjun Liu, Hao Leng, Peng Yu, and Xiuhai Wang. 2021. “Scour Characteristics and Equilibrium Scour Depth Prediction around Umbrella Suction Anchor Foundation under Random Waves” Journal of Marine Science and Engineering 9, no. 8: 886. https://doi.org/10.3390/jmse9080886

Find Other Styles

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

For more information on the journal statistics, click here.

Multiple requests from the same IP address are counted as one view.

원본 다운로드

Effect of tailwater depth on non-cohesive earth dam failure due to overtopping

범람으로 인한 비점착성 흙댐 붕괴에 대한 테일워터 깊이의 영향

ShaimaaAman^aMohamedAbdelrazek Rezk^bRabieaNasr^c

Abstract

본 연구에서는 범람으로 인한 토사댐 붕괴에 대한 테일워터 깊이의 영향을 실험적으로 조사하였다. 테일워터 깊이의 네 가지 다른 값을 검사합니다. 각 실험에 대해 댐 수심 측량 프로파일의 진화, 고장 기간, 침식 체적 및 유출 수위곡선을 관찰하고 기록합니다.

결과는 tailwater 깊이를 늘리면 고장 시간이 최대 57% 감소하고 상대적으로 침식된 마루 높이가 최대 77.6% 감소한다는 것을 보여줍니다. 또한 상대 배수 깊이가 3, 4, 5인 경우 누적 침식 체적의 감소는 각각 23, 36.5 및 75%인 반면 최대 유출량의 감소는 각각 7, 14 및 17.35%입니다.

실험 결과는 침식 과정을 복제할 때 Flow 3D 소프트웨어의 성능을 평가하는 데 활용됩니다. 수치 모델은 비응집성 흙댐의 침식 과정을 성공적으로 시뮬레이션합니다.

The influence of tailwater depth on earth dam failure due to overtopping is investigated experimentally in this work. Four different values of tailwater depths are examined. For each experiment, the evolution of the dam bathymetry profile, the duration of failure, the eroded volume, and the outflow hydrograph are observed and recorded. The results reveal that increasing the tailwater depth reduces the time of failure by up to 57% and decreases the relative eroded crest height by up to 77.6%. In addition, for relative tailwater depths equal to 3, 4, and 5, the reduction in the cumulative eroded volume is 23, 36.5, and 75%, while the reduction in peak discharge is 7, 14, and 17.35%, respectively. The experimental results are utilized to evaluate the performance of the Flow 3D software in replicating the erosion process. The numerical model successfully simulates the erosion process of non-cohesive earth dams.

Keywords

Earth dam, Eroded volume, Flow 3D model, Non-cohesive soil, Overtopping failure, Tailwater depth

Notation

d₅₀

Mean partical diameterW_c

Optimum water contentZ_o

Dam height (cm)d_o

Tailwater depth (cm)Z_eroded

Eroded height of the dam measured at distance of 0.7 m from the dam heel (cm)t

Total time of failure (sec)t₁

Time of crest width erosion (sec)Z_crest

The crest height (cm)V_total

Total volume of the dam (m³)V_eroded

Cumulative eroded volume (m³)RMSE

The statistical variable root- mean- square errord

Degree of agreement indexy_u.s.

The upstream water depth (cm)y_d.s

The downstream water depth (cm)H

Water surface elevation over sharp crested weir (cm)Q

Outflow discharge (liter/sec)Q_peak

Peak discharge (liter/sec)

1. Introduction

Earth dams are compacted structures composed of natural materials that are usually mined or quarried from local locations. The failures of the earth dams have proven to be deadly, destructive, and costly. According to People’s Daily, two earthen dams, Yong’an Dam and Xinfa Dam located in Hulun Buir City in North China’s Inner Mongolia failed on 2021, due to a surge in the water level of the Nuomin River caused by heavy rain. The dam breach affected 16,660 people, flooded 325,622 mu of farmland (21708.1 ha), and destroyed 22 bridges, 124 culverts, and 15.6 km of roadways. Also, the failure of south fork dam (earth and rock fill dam) near Johnstown on 1889 is considered the worst U.S dam disaster in terms of loss of life. The dam was overtopped and washed away due to unexpected heavy rains, releasing 20 million tons of water which destroyed Johnstown and resulted in 2209 deaths, [1], [2]. Piping or shear sliding, failure due to natural factors, and failure due to overtopping are all possible causes of earth dam failure. However, overtopping failure is the most frequent cause of dam failure. According to The International Committee on Large Dams (ICOLD, 1995), and [3], more than one-third of the total known dam failures were caused by dam overtopping.

Overtopping occurs as the result of insufficient flood design or freeboard in some cases. Extreme rainstorms can cause floods which can overtop the dam and cause it to fail. The size and geometry of the reservoir or the dam (side slopes, top width, height, etc.), the homogeneity of the material used in the construction of the dam, overtopping depth, and the presence or absence of tailwater are all elements that influence this type of failure which will be illustrated in the following literature. Overtopping failures of earth dams may be divided into several failure mechanisms based on the material composition and the inner structure of the dam. For cohesive earth dams because of low permeability, no seepage exists on the slopes. Erosion often begins at the earth dam toe during turbulent erosion and moves upstream, undercutting the slope, causing the removal of large chunks of materials. While for non-cohesive earth dams the downstream face of the dam flattens progressively and is often said to rotate around a point near the downstream toe [4], [5], [6] In the last few decades, the study of failures due to overtopping has gained popularity among researchers. The overtopping failure, in fact, has been widely investigated in coastal and river hydraulics and morpho dynamic. In addition, several laboratory experimental studies have been conducted in this field in order to better understand different involved factors. Also, many numerical types of research have been conducted to investigate the process of overtopping failure as well as the elements that influence this type of failure.

Tabrizi et al. [5] conducted a series of embankment overtopping tests to find the effect of compaction on the failure of a homogenous sand embankment. A plane breach process occurred across the flume width due to the narrow flume width. They measured the downstream hydrographs and embankment surface profile for every case. They concluded that the peak discharge decreased with a high compaction level, while the time to peak increased. Kansoh et al. [6] studied experimentally the failure of compacted homogeneous non-cohesive earthen embankment due to overtopping. They investigated the influence of different shape parameters including the downstream slope, the crest width, and the height of the embankment on the erosion process. The erosion process was initiated by carving a pilot channel into the embankment crest. They evaluated the time of embankment failure for different shape parameters. They concluded that the failure time increases with increasing the downstream slope and the crest width. Zhu et al. [7] investigated experimentally the breaching of five embankments, one constructed with pure sand, and four with different sand-silt–clay mixtures. The erosion pattern was similar across the flume width. They stated that for cohesive soil mixtures the head cut erosion was the most important factor that affected the breach growth, while for non-cohesive soil the breach erosion was affected by shear erosion.

Amaral et al. [8] studied experimentally the failure by overtopping for two embankments built from silt sand material. They studied the effect of the degree of compaction of the embankment and the geometry of the pilot channel carved at the centre of the dam crest. They studied two shapes of pilot channel a rectangular shape and triangular shape. They stated that the breach development is influenced by a higher degree of compaction, however, the pilot channel geometry did not influence the breach’s final form. Bereta et al. [9] studied experimentally the breach formation of five dam models, three of them were homogenous clay soil while two were sandy-clay mixtures. The erosion process was initiated by cutting a pilot channel at the centre of the dam crest. They observed the initiation of erosion, flow shear erosion, sidewall bottom erosion, and distinguished the soil mechanical slope mass failure from the head cut vertically and laterally during these tests. Verma et al. [10] investigated experimentally a two-dimensional erosion phenomenon due to overtopping by using a wooden fuse plug model and five different soils. They concluded that the erosion process was affected mostly by cohesiveness and degree of compaction. For cohesive soils, a head cut erosion was observed, while for non-cohesive soils surface erosion occurred gradually. Also, the dimensions of fuse plug, type of fill material, reservoir capacity, and inflow were found to affect the behaviour of the overall breaching process.

Wu and Qin [11] studied the effect of adding coarse grains to the downstream face of a non-cohesive dam as a result of tailings deposition. The process of overtopping during tailings dam failures is analyzed and its effect on delaying the dam-break process and disaster mitigation are investigated. They found that the tested protective measures decreased the breach area, the maximum breaching flow discharge and flow velocity, and the downstream inundated area. Khankandi et al. [12] studied experimentally the effect of reservoir geometry on dam break flow in case of dry and wet bed conditions. They considered four different reservoir shapes, a long reservoir, a wide, a trapezoidal shaped and one with a 90◦ bend all with identical water volume and horizontal bed. The dam break is simulated by the sudden gate removal using a pneumatic jack. They measured the variation of water level over time with ultrasonic sensors and flow velocity component with an acoustic Doppler velocimeter. Also, the experimental results of water level variation are compared with Ritters solution (1892) [13]. They stated that for dry bed condition the long and 90 bend reservoirs results are close to the analytical solution by ritter also in these two shapes a 1D flow is noticed. However, for wide and trapezoidal reservoirs a 2D effect is significant due to flow contraction at channel entrance.

Rifai et al. [14] conducted a series of experiments to investigate the effect of tailwater depth on the outflow discharge and breach geometry during non-cohesive homogenous fluvial dikes overtopping failure. They cut an initial notch in the crest at 0.8 m from the upstream end of the dike to initiate overtopping. They compared their results to previous experiments under different main channel inflow discharges combined with a free floodplain. They divided the dike breaching process into three stages: gradual start of overtopping flow resulting in slow initiation of dike erosion, deepening and widening breach due to large flow depth and velocity, finally the flow depth starts stabilizing at its minimal level with or without sustained breach expansion. They stated that breach discharge has lower values than in free floodplain tests. Jiang [15] studied the effect of bed slope on breach parameters and peak discharge in non-cohesive embankment failure. An initial triangular breach with a depth and width of 4 cm was pre-set on one side of the dam. He stated that peak discharge increases with the increase of bed slope and then decreases.

Ozmen-cagatay et al. [16] studied experimentally flood wave propagation resulted from a sudden dam break event. For dam-break modelling, they used a mechanism that permitted the rapid removal of a vertical plate with a thickness of 4 mm and made of rigid plastic. They conducted three tests, one with dry bed condition and two tests with tailwater depths equal 0.025 m and 0.1 m respectively. They recorded the free surface profile during initial stages of dam break by using digital image processing. Finally, they compared the experimental results with the with a commercially available VOF-based CFD program solving the Reynolds-averaged Navier –Stokes equations (RANS) with the k– Ɛ turbulence model and the shallow water equations (SWEs). They concluded that Wave breaking was delayed with increasing the tailwater depth to initial reservoir depth ratio. They also stated that the SWE approach is sufficient more to represent dam break flows for wet bed condition. Evangelista [17] investigated experimentally and numerically using a depth-integrated two-phase model, the erosion of sand dike caused by the impact of a dam break wave. The dam break is simulated by a sudden opening of an upstream reservoir gate resulting in the overtopping of a downstream trapezoidal sand dike. The evolution of the water wave caused from the gate opening and dike erosion process are recorded by using a computer-controlled camera. The experimental results demonstrated that the progression of the wave front and dike erosion have a considerable influence on each other during the process. In addition, the dike constructed from fine sands was more resistant to erosion than the one built with coarse sand. They also stated that the numerical model can is capable of accurately predicting wave front position and dike erosion. Also, Di Cristo et al. [18] studied the effect of dam break wave propagation on a sand embankment both experimentally and numerically using a two-phase shallow-water model. The evolution of free surface and of the embankment bottom are recorded and used in numerical model assessment. They stated that the model allows reasonable simulation of the experimental trends of the free surface elevation regardeless of the geofailure operator.

Lots of numerical models have been developed over the past few years to simulate the dam break flooding problem. A one-dimensional model, such as Hec-Ras, DAMBRK and MIKE 11, ect. A two-dimensional model such as iRIC Nay2DH is used in earth embankment breach simulation. Other researchers studied the failure process numerically using (3D) computational fluid dynamics (CFD) models, such as FLOW-3D, and FLUENT. Goharnejad et al. [19] determined the outflow hydrograph which results from the embankment dam break due to overtopping. Hu et al. [20] performed a comparison between Flow-3D and MIKE3 FM numerical models in simulating a dam break event under dry and wet bed conditions with different tailwater depths. Kaurav et al. [21] simulated a planar dam breach process due to overtopping. They conducted a sensitivity analysis to find the effect of dam material, dam height, downstream slope, crest width, and inlet discharge on the erosion process and peak discharge through breach. They concluded that downstream slope has a significant influence on breaching process. Yusof et al. [22] studied the effect of embankment sediment sizes and inflow rates on breaching geometric and hydrodynamic parameters. They stated that the peak outflow hydrograph increases with increasing sediment size and inflow rates while time of failure decreases.

In the present work, the effect of tailwater depth on earth dam failure during overtopping is studied experimentally. The relation between the eroded volume of the dam and the tailwater depth is presented. Also, the percentage of reduction in peak discharge due to tailwater existence is calculated. An assessment of Flow 3D software performance in simulating the erosion process during earth dam failure is introduced. The statistical variable root- mean- square error, RMSE, and the agreement degree index, d, are used in model assessment.

2. Material and methods

The tests are conducted in a straight rectangular flume in the laboratory of Irrigation Engineering and Hydraulics Department, Faculty of Engineering, Alexandria University, Egypt. The flume dimensions are 10 m long, 0.86 m wide, and 0.5 m deep. The front part of the flume is connected to a storage basin 1 m long by 0.86 m wide. The storage basin is connected to a collecting tank for water recirculation during the experiments as shown in Fig. 1, Fig. 2. A sharp-crested weir is placed at a distance of 4 m downstream the constructed dam to keep a constant tailwater depth in each experiment and to measure the outflow discharge.

To measure the eroded volume with time a rods technique is used. This technique consists of two parallel wooden plates with 10 cm distance in between and five rows of stainless-steel rods passing vertically through the wooden plates at a spacing of 20 cm distributed across flume width. Each row consists of four rods with 15 cm spacing between them. Also, a graph board is provided to measure the drop in each rod with time as shown in Fig. 3, Fig. 4. After dam construction the rods are carefully rested on the dam, with the first line of rods resting in the middle of the dam crest and then a constant distance of 15 cm between rods lines is maintained.

A soil sample is taken and tested in the laboratory of the soil mechanics to find the soil geotechnical parameters. The soil particle size distribution is also determined by sieve analysis as shown in Fig. 5. The soil mean diameter d₅₀,equals 0.38 mm and internal friction angle equals 32.6°.

2.1. Experimental procedures

To investigate the effect of the tailwater depth (d_o), the tailwater depth is changed four times 5, 15, 20, and 25 cm on the sand dam model. The dam profile is 35 cm height, with crest width = 15 cm, the dam base width is 155 cm, and the upstream and downstream slopes are 2:1 as shown in Fig. 6. The dam dimensions are set as the flume permitted to allow observation of the dam erosion process under the available flume dimensions and conditions. All of the conducted experiments have the same dimensions and configurations.

The optimum water content, W_c, from the standard proctor test is found to be 8 % and the maximum dry unit weight is 19.42 kN/m³. The soil and water are mixed thoroughly to ensure consistency and then placed on three horizontal layers. Each layer is compacted according to ASTM standard with 25 blows by using a rammer (27 cm × 20.5 cm) weighing 4 kg. Special attention is paid to the compaction of the soil to guarantee the repeatability of the tests.

After placing and compacting the three layers, the dam slopes are trimmed carefully to form the trapezoidal shape of the dam. A small triangular pilot channel with 1 cm height and 1:1 side slopes is cut into the dam crest to initiate the erosion process. The position of triangular pilot channel is presented in Fig. 1. Three digital video cameras with a resolution of 1920 × 1080 pixels and a frame rate of 60 fps are placed in three different locations. One camera on one side of the flume to record the progress of the dam profile during erosion. Another to track the water level over the sharp-crested rectangular weir placed at the downstream end of the flume. And the third camera is placed above the flume at the downstream side of the dam and in front of the rods to record the drop of the tip of the rods with time as shown previously in Fig. 1.

Before starting the experiment, the water is pumped into the storage basin by using pump with capacity 360 m³/hr, and then into the upstream section of the flume. The upstream boundary is an inflow condition. The flow discharge provided to the storage basin is kept at a constant rate of 6 L/sec for all experiments, while the downstream boundary is an outflow boundary condition.

Also, the required tailwater depth for each experiment is filled to the desired depth. A dye container valve is opened to color the water upstream of the dam to make it easy to distinguish the dam profile from the water profile. A wooden board is placed just upstream of the dam to prevent water from overtopping the dam until the water level rises to a certain level above the dam crest and then the wooden board is removed slowly to start the experiment.

2.2. Repeatability

To verify the accuracy of the results, each experiment is repeated two times under the same conditions. Fig. 7 shows the relative eroded crest height, Z_eroded / Z_o, with time for 5 cm tailwater depth. From the Figure, it can be noticed that results for all runs are consistent, and accuracy is achieved.

3. Numerical model

The commercially available numerical model, Flow 3D is used to simulate the dam failure due to overtopping for the cases of 15 cm, 20 cm and 25 cm tailwater depths. For numerical model calibration, experimental results for dam surface evolution are used. The numerical model is calibrated for selection of the optimal turbulence model (RNG, K-e, and k-w) and sediment scour equations (Van Rin, Meyer- peter and Muller, and Nielsen) that produce the best results. In this, the flow field is solved by the RNG turbulence model, and the van Rijn equation is used for the sediment scour model. A geometry file is imported before applying the mesh.

A Mesh sensitivity is analyzed and checked for various cell sizes, and it is found that decreasing the cell size significantly increases the simulation time with insignificant differences in the result. It is noticed that the most important factor influencing cell size selection is the value of the dam’s upstream and downstream slopes. For example, the slopes in the dam model are 2:1, thus the cell size ratio in X and Z directions should be 2:1 as well. The cell size in a mesh block is set to be 0.02 m, 0.025 m, and 0.01 m in X, Y and Z directions respectively.

In the numerical computations, the boundary conditions employed are the walls for sidewalls and the channel bottom. The pressure boundary condition is applied at the top, at the air–water interface, to account for atmospheric pressure on the free surface. The upstream boundary is volume flow rate while the downstream boundary is outflow discharge.

The initial condition is a fluid region, which is used to define fluid areas both upstream and downstream of the dam. To assess the model accuracy, the statistical variable root- mean- square error, RMSE, and the agreement degree index, d, are calculated as(1)RMSE=1N∑i=1N(Pi-Mi)2(2)d=1-∑Mi-Pi2∑Mi-M¯+Pi-P¯2

where N is the number of samples, P_i and M_i are the models and experimental values, P and M are the means of the model and experimental values. The best fit between the experimental and model results would have an RMSE = 0 and degree of agreement, d = 1.

4. Results of experimental work

The results of the total time of failure, t (defined as the time from when the water begins to overtop the dam crest until the erosion reaches a steady state, when no erosion occurs), time of crest width erosion t₁, cumulative eroded volume V_eroded, and peak discharge Q_peak for each experiment are listed in Table 1. The case of 5 cm tailwater depth is considered as a reference case in this work.

Table 1. Results of experimental work.

*Tailwater depth, d_o* (cm)**	*Total time of failure, t (sec)*	*Time of crest width erosion, t₁ (sec)*	*cumulative eroded volume, V_eroded* (m³)**	*Peak discharge, Q_peak* (liter/sec)**
5	255	22	0.21	13.12
15	165	30	0.16	12.19
20	140	34	0.13	11.29
25	110	39	0.05	10.84

5. Discussion

5.1. Side erosion

The evolution of the bathymetry of the erosion line recorded by the video camera1. The videos are split into frames (60 frames/sec) by the Free Video to JPG Converter v.5.063 build and then converted into an excel spreadsheet using MATLAB code as shown in Fig. 8.

Fig. 9 shows a sample of numerical model output. Fig. 10, Fig. 11, Fig. 12 show a dam profile development for different time steps from both experimental and numerical model, for tailwater depths equal 15 cm, 20 cm and 25 cm. Also, the values of RMSE and d for each figure are presented. The comparison shows that the Flow 3D software can simulate the erosion process of non-cohesive earth dam during overtopping with an RMSE value equals 0.023, 0.0218, and 0.0167 and degree of agreement, d, equals 0.95, 0.968, and 0.988 for relative tailwater depths, d_o/(d_o)_ref, = 3, 4 and 5, respectively. The low values of RMSE and high values of d show that the Flow 3D can effectively simulate the erosion process. From Fig. 10, Fig. 11, Fig. 12, it can be noticed that the model is not capable of reproducing the head cut, while it can simulate well the degradation of the crest height with a minor difference from experimental work. The reason of this could be due to inability of simulation of all physical conditions which exists in the experimental work, such as channel friction and the grain size distribution of the dam soil which is surely has a great effect on the erosion process and breach development. In the experimental work the grain size distribution is shown in Fig. 5, while the numerical model considers that the soil is uniform and exactly 50 % of the dam particles diameter are equal to the d₅₀ value. Another reason is that the model is not considering the increased resistance of the dam due to the apparent cohesion which happens due to dam saturation [23].

It is clear from both the experimental and numerical results that for a 5 cm tailwater depth, d_o/(d_o)_ref = 1.0, erosion begins near the dam toe and continues upward on the downstream slope until it reaches the crest. After eroding the crest width, the crest is lowered, resulting in increased flow rates and the speeding up of the erosion process. While for relative tailwater depths, d_o/(d_o)_ref = 3, 4, and 5 erosion starts at the point of intersection between the downstream slope and tailwater. The existence of tailwater works as an energy dissipater for the falling water which reduces the erosion process and prevents the dam from failure as shown in Fig. 13. It is found that the time of the failure decreases with increasing the tailwater depth because most of the dam height is being submerged with water which decreases the erosion process. The reduction in time of failure from the referenced case is found to be 35.3, 45, and 57 % for relative tailwater depth, d_o /(d_o)_ref equals 3, 4, and 5, respectively.

The relation between the relative eroded crest height, Z_eroded /Z_o, with time is drawn as shown in Fig. 14. It is found that the relative eroded crest height decreases with increasing tailwater depth by 10, 41, and 77.6 % for relative tailwater depth, d_o /(d_o)_ref equals 3, 4, and 5, respectively. The time required for the erosion of the crest width, t₁, is calculated for each experiment. The relation between relative tailwater depth and relative time of crest width erosion is shown in Fig. 15. It is found that the time of crest width erosion increases linearly with increasing, d_o /Z_o. The percent of increase is 36.4, 54.5 and 77.3 % for relative tailwater depth, d_o /(d_o)_ref = 3, 4 and 5, respectively.

Crest height, Z_crest is calculated from the experimental results and the Flow 3D results for relative tailwater depths, d_o/(d_o)_ref, = 3, 4, and 5. A relation between relative crest height, Z_crest/Z_o with time from experimental and numerical results is presented in Fig. 16. From Fig. 16, it is seen that there is a good consistency between the results of numerical model and the experimental results in the case of tracking the erosion of the crest height with time.

5.2. Upstream and downstream water depths

It is noticed that at the beginning of the erosion process, both upstream and downstream water depths increase linearly with time as long as erosion of the crest height did not take place. However, when the crest height starts to lower the upstream water depth decreases with time while the downstream water depth increases. At the end of the experiment, the two depths are nearly equal. A relation between relative downstream and upstream water depths with time is drawn for each experiment as shown in Fig. 17.

5.3. Eroded volume

A MATLAB code is used to calculate the cumulative eroded volume every time interval for each experiment. The total volume of the dam, V_total is 0.256 m³. The cumulative eroded volume, V_eroded is 0.21, 0.16, 0.13, and 0.05 m³ for tailwater depths, d_o = 5, 15, 20, and 25 cm, respectively. Fig. 18 presents the relation between cumulative eroded volume, V_eroded and time. From Fig. 18, it is observed that the cumulative eroded volume decreases with increasing the tailwater depth. The reduction in cumulative eroded volume is 23, 36.5, and 75 % for relative tailwater depth, d_o /(d_o)_ref = 3, 4, and 5, respectively. The relative remained volume of the dam equals 0.18, 0.375, 0.492, and 0.8 for tailwater depths = 5, 15, 20, and 25 cm, respectively. Fig. 19 shows a relation between relative tailwater depth and relative cumulative eroded volume from experimental results. From that figure, it is noticed that the eroded volume decreases exponentially with increasing relative tailwater depth.

5.4. The outflow discharge

The inflow discharge provided to the storage tank is maintained constant for all experiments. The water surface elevation, H, over the sharp-crested weir placed at the downstream side is recorded by the video camera 2. For each experiment, the outflow discharge is then calculated by using the sharp-crested rectangular weir equation every 10 sec.

The outflow discharge is found to increase rapidly until it reaches its peak then it decreases until it is constant. For high values of tailwater depths, the peak discharge becomes less than that in the case of small tailwater depth as shown in Fig. 20 which agrees well with the results of Rifai et al. [14] The reduction in peak discharge is 7, 14, and 17.35 % for relative tailwater depth, d_o /(d_o)_ref = 3, 4, and 5, respectively.

The scenario presented in this article in which the tailwater depth rises due to unexpected heavy rainfall, is investigated to find the effect of rising tailwater depth on earth dam failure. The results revealed that rising tailwater depth positively affects the process of dam failure in terms of preventing the dam from complete failure and reducing the outflow discharge.

6. Conclusions

The effect of tailwater depth on earth dam failure due to overtopping is investigated experimentally in this work. The study focuses on the effect of tailwater depth on side erosion, upstream and downstream water depths, eroded volume, outflow hydrograph, and duration of the failure process. The Flow 3D numerical software is used to simulate the dam failure, and a comparison is made between the experimental and numerical results to find the ability of this software to simulate the erosion process. The following are the results of the investigation:

The existence of tailwater with high depths prevents the dam from completely collapsing thereby turning it into a broad crested weir. The failure time decreases with increasing the tailwater depth and the reduction from the reference case is found to be 35.3, 45, and 57 % for relative tailwater depth, d_o /(d_o)_ref = 3, 4, and 5, respectively. The difference between the upstream and downstream water depths decreases with time till it became almost negligible at the end of the experiment. The reduction in cumulative eroded volume is 23, 36.5, and 75 % for relative tailwater depth, d_o /(d_o)_ref = 3, 4, and 5, respectively. The peak discharge decreases by 7, 14, and 17.35 % for relative tailwater depth, d_o /(d_o)_ref = 3, 4, and 5, respectively. The relative eroded crest height decreases linearly with increasing the tailwater depth by 10, 41, and 77.6 % for relative tailwater depth, d_o /(d_o)_ref = 3, 4, and 5, respectively. The numerical model can reproduce the erosion process with a minor deviation from the experimental results, particularly in terms of tracking the degradation of the crest height with time.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Reference

[1]

D. McCullough

The Johnstown Flood

Simon and Schuster, NY (1968)

Google Scholar [2]Rose AT. The influence of dam failures on dam safety laws in Pennsylvania. Association of State Dam Safety Officials Annual Conference 2013, Dam Safety 2013. 2013;1:738–56.

Google Scholar [3]

M. Foster, R. Fell, M. Spannagle

The statistics of embankment dam failures and accidents

Can Geotech J, 37 (5) (2000), pp. 1000-1024, 10.1139/t00-030 View PDF

View Record in Scopus Google Scholar [4]Pickert, G., Jirka, G., Bieberstein, A., Brauns, J. Soil/water interaction during the breaching process of overtopped embankments. In: Greco, M., Carravetta, A., Morte, R.D. (Eds.), Proceedings of the Conference River-Flow 2004, Balkema.

Google Scholar [5]

A. Asghari Tabrizi, E. Elalfy, M. Elkholy, M.H. Chaudhry, J. Imran

Effects of compaction on embankment breach due to overtopping

J Hydraul Res, 55 (2) (2017), pp. 236-247, 10.1080/00221686.2016.1238014 View PDF

View Record in Scopus Google Scholar [6]

R.M. Kansoh, M. Elkholy, G. Abo-Zaid

Effect of Shape Parameters on Failure of Earthen Embankment due to Overtopping

KSCE J Civ Eng, 24 (5) (2020), pp. 1476-1485, 10.1007/s12205-020-1107-x View PDF

View Record in Scopus Google Scholar [7]

YongHui Zhu, P.J. Visser, J.K. Vrijling, GuangQian Wang

Experimental investigation on breaching of embankments

Experimental investigation on breaching of embankments, 54 (1) (2011), pp. 148-155 View PDF

CrossRef View Record in Scopus Google Scholar [8]Amaral S, Jónatas R, Bento AM, Palma J, Viseu T, Cardoso R, et al. Failure by overtopping of earth dams. Quantification of the discharge hydrograph. Proceedings of the 3rd IAHR Europe Congress: 14-15 April 2014, Portugal. 2014;(1):182–93.

Google Scholar [9]

G. Bereta, P. Hui, H. Kai, L. Guang, P. Kefan, Y.Z. Zhao

Experimental study of cohesive embankment dam breach formation due to overtopping

Periodica Polytechnica Civil Engineering, 64 (1) (2020), pp. 198-211, 10.3311/PPci.14565 View PDF

View Record in Scopus Google Scholar [10]

D.K. Verma, B. Setia, V.K. Arora

Experimental study of breaching of an earthen dam using a fuse plug model

Int J Eng Trans A, 30 (4) (2017), pp. 479-485, 10.5829/idosi.ije.2017.30.04a.04 View PDF

View Record in Scopus Google Scholar [11]Wu T, Qin J. Experimental Study of a Tailings Impoundment Dam Failure Due to Overtopping. Mine Water and the Environment [Internet]. 2018;37(2):272–80. Available from: doi: 10.1007/s10230-018-0529-x.

Google Scholar [12]

A. Feizi Khankandi, A. Tahershamsi, S. Soares-Frazo

Experimental investigation of reservoir geometry effect on dam-break flow

J Hydraul Res, 50 (4) (2012), pp. 376-387 View PDF

CrossRef View Record in Scopus Google Scholar [13]

A. Ritter

Die Fortpflanzung der Wasserwellen (The propagation of water waves)

Zeitschrift Verein Deutscher Ingenieure, 36 (33) (1892), pp. 947-954

[in German]

View Record in Scopus Google Scholar [14]

I. Rifai, K. El Kadi Abderrezzak, S. Erpicum, P. Archambeau, D. Violeau, M. Pirotton, et al.

Floodplain Backwater Effect on Overtopping Induced Fluvial Dike Failure

Water Resour Res, 54 (11) (2018), pp. 9060-9073 View PDF

This article is free to access.

CrossRef View Record in Scopus Google Scholar [15]

X. Jiang

Laboratory Experiments on Breaching Characteristics of Natural Dams on Sloping Beds

Advances in Civil Engineering, 2019 (2019), pp. 1-14

View Record in Scopus Google Scholar [16]

H. Ozmen-Cagatay, S. Kocaman

Dam-break flows during initial stage using SWE and RANS approaches

J Hydraul Res, 48 (5) (2010), pp. 603-611 View PDF

CrossRef View Record in Scopus Google Scholar [17]

S. Evangelista

Experiments and numerical simulations of dike erosion due to a wave impact

Water (Switzerland), 7 (10) (2015), pp. 5831-5848 View PDF

CrossRef View Record in Scopus Google Scholar [18]

C. Di Cristo, S. Evangelista, M. Greco, M. Iervolino, A. Leopardi, A. Vacca

Dam-break waves over an erodible embankment: experiments and simulations

J Hydraul Res, 56 (2) (2018), pp. 196-210 View PDF

CrossRef View Record in Scopus Google Scholar [19]Goharnejad H, Sm M, Zn M, Sadeghi L, Abadi K. Numerical Modeling and Evaluation of Embankment Dam Break Phenomenon (Case Study : Taleghan Dam) ISSN : 2319-9873. 2016;5(3):104–11.

Google Scholar [20]Hu H, Zhang J, Li T. Dam-Break Flows : Comparison between Flow-3D , MIKE 3 FM , and Analytical Solutions with Experimental Data. 2018;1–24. doi: 10.3390/app8122456.

Google Scholar [21]

R. Kaurav, P.K. Mohapatra, D. Ph

Studying the Peak Discharge through a Planar Dam Breach, 145 (6) (2019), pp. 1-8 View PDF

CrossRef [22]

Z.M. Yusof, Z.A.L. Shirling, A.K.A. Wahab, Z. Ismail, S. Amerudin

A hydrodynamic model of an embankment breaching due to overtopping flow using FLOW-3D

IOP Conference Series: Earth and Environmental Science, 920 (1) (2021)

Google Scholar [23]

G. Pickert, V. Weitbrecht, A. Bieberstein

Breaching of overtopped river embankments controlled by apparent cohesion

J Hydraul Res, 49 (2) (Apr. 2011), pp. 143-156, 10.1080/00221686.2011.552468 View PDF

View Record in Scopus Google Scholar

Cited by (0)

My name is Shaimaa Ibrahim Mohamed Aman and I am a teaching assistant in Irrigation and Hydraulics department, Faculty of Engineering, Alexandria University. I graduated from the Faculty of Engineering, Alexandria University in 2013. I had my MSc in Irrigation and Hydraulic Engineering in 2017. My research interests lie in the area of earth dam Failures.

Peer review under responsibility of Ain Shams University.

PDF 원문보기

Fig. 5. The predicted shapes of initial breach (a) Rectangular (b) V-notch. Fig. 6. Dam breaching stages.

Investigating the peak outflow through a spatial embankment dam breach

공간적 제방댐 붕괴를 통한 최대 유출량 조사

Mahmoud T.Ghonim Magdy H.Mowafy Mohamed N.Salem AshrafJatwaryFaculty of Engineering, Zagazig University, Zagazig 44519, Egypt

Abstract

Investigating the breach outflow hydrograph is an essential task to conduct mitigation plans and flood warnings. In the present study, the spatial dam breach is simulated by using a three-dimensional computational fluid dynamics model, FLOW-3D. The model parameters were adjusted by making a comparison with a previous experimental model. The different parameters (initial breach shape, dimensions, location, and dam slopes) are studied to investigate their effects on dam breaching. The results indicate that these parameters have a significant impact. The maximum erosion rate and peak outflow for the rectangular shape are higher than those for the V-notch by 8.85% and 5%, respectively. Increasing breach width or decreasing depth by 5% leads to increasing maximum erosion rate by 11% and 15%, respectively. Increasing the downstream slope angle by 4° leads to an increase in both peak outflow and maximum erosion rate by 2.0% and 6.0%, respectively.

유출 유출 수문곡선을 조사하는 것은 완화 계획 및 홍수 경보를 수행하는 데 필수적인 작업입니다. 본 연구에서는 3차원 전산유체역학 모델인 FLOW-3D를 사용하여 공간 댐 붕괴를 시뮬레이션합니다. 이전 실험 모델과 비교하여 모델 매개변수를 조정했습니다.

다양한 매개변수(초기 붕괴 형태, 치수, 위치 및 댐 경사)가 댐 붕괴에 미치는 영향을 조사하기 위해 연구됩니다. 결과는 이러한 매개변수가 상당한 영향을 미친다는 것을 나타냅니다. 직사각형 형태의 최대 침식율과 최대 유출량은 V-notch보다 각각 8.85%, 5% 높게 나타났습니다.

위반 폭을 늘리거나 깊이를 5% 줄이면 최대 침식률이 각각 11% 및 15% 증가합니다. 하류 경사각을 4° 증가시키면 최대 유출량과 최대 침식률이 각각 2.0% 및 6.0% 증가합니다.

Keywords

Spatial dam breach; FLOW-3D; Overtopping erosion; Computational fluid dynamics (CFD)

1. Introduction

There are many purposes for dam construction, such as protection from flood disasters, water storage, and power generation. Embankment failures may have a catastrophic impact on lives and infrastructure in the downstream regions. One of the most common causes of embankment dam failure is overtopping. Once the overtopping of the dam begins, the breach formation will start in the dam body then end with the dam failure. This failure occurs within a very short time, which threatens to be very dangerous. Therefore, understanding and modeling the embankment breaching processes is essential for conducting mitigation plans, flood warnings, and forecasting flood damage.

The analysis of the dam breaching process is implemented by different techniques: comparative methods, empirical models with dimensional and dimensionless solutions, physical-based models, and parametric models. These models were described in detail [1]. Parametric modeling is commonly used to simulate breach growth as a time-dependent linear process and calculate outflow discharge from the breach using hydraulics principles [2]. Alhasan et al. [3] presented a simple one-dimensional mathematical model and a computer code to simulate the dam breaching process. These models were validated by small dams breaching during the floods in 2002 in the Czech Republic. Fread [4] developed an erosion model (BREACH) based on hydraulics principles, sediment transport, and soil mechanics to estimate breach size, time of formation, and outflow discharge. Říha et al. [5] investigated the dam break process for a cascade of small dams using a simple parametric model for piping and overtopping erosion, as well as a 2D shallow-water flow model for the flood in downstream areas. Goodarzi et al. [6] implemented mathematical and statistical methods to assess the effect of inflows and wind speeds on the dam’s overtopping failure.

Dam breaching studies can be divided into two main modes of erosion. The first mode is called “planar dam breach” where the flow overtops the whole dam width. While the second mode is called “spatial dam breach” where the flow overtops through the initial pilot channel (i.e., a channel created in the dam body). Therefore, the erosion will be in both vertical and horizontal directions [7].

The erosion process through the embankment dams occurs due to the shear stress applied by water flows. The dam breaching evolution can be divided into three stages [8], [9], but Y. Yang et al. [10] divided the breach development into five stages: Stage I, the seepage erosion; Stage II, the initial breach formation; Stage III, the head erosion; Stage IV, the breach expansion; and Stage V, the re-equilibrium of the river channel through the breach. Many experimental tests have been carried out on non-cohesive embankment dams with an initial breach to examine the effect of upstream inflow discharges on the longitudinal profile evolution and the time to inflection point [11].

Zhang et al. [12] studied the effect of changing downstream slope angle, sediment grain size, and dam crest length on erosion rates. They noticed that increasing dam crest length and decreasing downstream slope angle lead to decreasing sediment transport rate. While the increase in sediment grain size leads to an increased sediment transport rate at the initial stages. Höeg et al. [13] presented a series of field tests to investigate the stability of embankment dams made of various materials. Overtopping and piping were among the failure tests carried out for the dams composed of homogeneous rock-fill, clay, or gravel with a height of up to 6.0 m. Hakimzadeh et al. [14] constructed 40 homogeneous cohesive and non-cohesive embankment dams to study the effect of changing sediment diameter and dam height on the breaching process. They also used genetic programming (GP) to estimate the breach outflow. Refaiy et al. [15] studied different scenarios for the downstream drain geometry, such as length, height, and angle, to minimize the effect of piping phenomena and therefore increase dam safety.

Zhu et al. [16] examined the effect of headcut erosion on dam breach growth, especially in the case of cohesive dams. They found that the breach growth in non-cohesive embankments is slower than cohesive embankments due to the little effect of headcut. Schmocker and Hager [7] proposed a relationship for estimating peak outflow from the dam breach process.(1)QpQin-1=1.7exp-20hc23d5013H0

where: Q_p = peak outflow discharge.

Q_in = inflow discharge.

h_c = critical flow depth.

d₅₀ = mean sediment diameter.

H_o = initial dam height.

Yu et al. [17] carried out an experimental study for homogeneous non-cohesive embankment dams in a 180° bending rectangular flume to determine the effect of overtopping flows on breaching formation. They found that the main factors influencing breach formation are water level, river discharge, and embankment material diameter.

Wu et al. [18] carried out a series of experiments to investigate the effect of breaching geometry on both non-cohesive and cohesive embankment dams in a U-bend flume due to overtopping flows. In the case of non-cohesive embankments, the non-symmetrical lateral expansion was noticed during the breach formation. This expansion was described by a coefficient ranging from 2.7 to 3.3.

The numerical models of the dam breach can be categorized according to different parameters, such as flow dimensions (1D, 2D, or 3D), flow governing equations, and solution methods. The 1D models are mainly used to predict the outflow hydrograph from the dam breach. Saberi et al. [19] applied the 1D Saint-Venant equation, which is solved by the finite difference method to investigate the outflow hydrograph during dam overtopping failure. Because of the ability to study dam profile evolution and breach formation, 2D models are more applicable than 1D models. Guan et al. [20] and Wu et al. [21] employed both 2D shallow water equations (SWEs) and sediment erosion equations, which are solved by the finite volume method to study the effect of the dam’s geometry parameters on outflow hydrograph and dam profile evolution. Wang et al. [22] also proposed a second-order hybrid-type of total variation diminishing (TVD) finite-difference to estimate the breach outflow by solving the 2D (SWEs). The accuracy of (SWEs) for both vertical flow contraction and surface roughness has been assessed [23]. They noted that the accuracy of (SWEs) is acceptable for milder slopes, but in the case of steeper slopes, modelers should be more careful. Generally, the accuracy of 2D models is still low, especially with velocity distribution over the flow depth, lateral momentum exchange, density-driven flows, and bottom friction [24]. Therefore, 3D models are preferred. Larocque et al. [25] and Yang et al. [26] started to use three-dimensional (3D) models that depend on the Reynolds-averaged Navier-Stokes (RANS) equations.

Previous experimental studies concluded that there is no clear relationship between the peak outflow from the dam breach and the initial breach characteristics. Some of these studies depend on the sharp-crested weir fixed at the end of the flume to determine the peak outflow from the breach, which leads to a decrease in the accuracy of outflow calculations at the microscale. The main goals of this study are to carry out a numerical simulation for a spatial dam breach due to overtopping flows by using (FLOW-3D) software to find an empirical equation for the peak outflow discharge from the breach and determine the worst-case that leads to accelerating the dam breaching process.

2. Numerical simulation

The current study for spatial dam breach is simulated by using (FLOW-3D) software [27], which is a powerful computational fluid dynamics (CFD) program.

2.1. Geometric presentations

A stereolithographic (STL) file is prepared for each change in the initial breach geometry and dimensions. The CAD program is useful for creating solid objects and converting them to STL format, as shown in Fig. 1.

2.2. Governing equations

The governing equations for water flow are three-dimensional Reynolds Averaged Navier-Stokes equations (RANS).

The continuity equation:(2)∂ui∂xi=0

The momentum equation:(3)∂ui∂t+1VFuj∂ui∂xj=1ρ∂∂xj-pδij+ν∂ui∂xj+∂uj∂xi-ρu`iu`j¯

where u is time-averaged velocity,ν is kinematic viscosity, VF is fractional volume open to flow, p is averaged pressure and -u`iu`j¯ are components of Reynold’s stress. The Volume of Fluid (VOF) technique is used to simulate the free surface profile. Hirt et al. [28] presented the VOF algorithm, which employs the function (F) to express the occupancy of each grid cell with fluid. The value of (F) varies from zero to unity. Zero value refers to no fluid in the grid cell, while the unity value refers to the grid cell being fully occupied with fluid. The free surface is formed in the grid cells having (F) values between zero and unity.(4)∂F∂t+1VF∂∂xFAxu+∂∂yFAyv+∂∂zFAzw=0

where (u, v, w) are the velocity components in (x, y, z) coordinates, respectively, and (A_x, A_y, A_z) are the area fractions.

2.3. Boundary and initial conditions

To improve the accuracy of the results, the boundary conditions should be carefully determined. In this study, two mesh blocks are used to minimize the time consumed in the simulation. The boundary conditions for mesh block 1 are as follows: The inlet and sides boundaries are defined as a wall boundary condition (wall boundary condition is usually used for bound fluid by solid regions. In the case of viscous flows, no-slip means that the tangential velocity is equal to the wall velocity and the normal velocity is zero), the outlet is defined as a symmetry boundary condition (symmetry boundary condition is usually used to reduce computational effort during CFD simulation. This condition allows the flow to be transferred from one mesh block to another. No inputs are required for this boundary condition except that its location should be defined accurately), the bottom boundary is defined as a uniform flow rate boundary condition, and the top boundary is defined as a specific pressure boundary condition with assigned atmospheric pressure. The boundary conditions for mesh block 2 are as follows: The inlet is defined as a symmetry boundary condition, the outlet is defined as a free flow boundary condition, the bottom and sides boundaries are defined as a wall boundary condition, and the top boundary is defined as a specific pressure boundary condition with assigned atmospheric pressure as shown in Fig. 2. The initial conditions required to be set for the fluid (i.e., water) inside of the domain include configuration, temperature, velocities, and pressure distribution. The configuration of water depends on the dimensions and shape of the dam reservoir. While the other conditions have been assigned as follows: temperature is normal water temperature (25 °c) and pressure distribution is hydrostatic with no initial velocity.

2.4. Numerical method

FLOW-3D uses the finite volume method (FVM) to solve the governing equation (Reynolds-averaged Navier-Stokes) over the computational domain. A finite-volume method is an Eulerian approach for representing and evaluating partial differential equations in algebraic equations form [29]. At discrete points on the mesh geometry, values are determined. Finite volume expresses a small volume surrounding each node point on a mesh. In this method, the divergence theorem is used to convert volume integrals with a divergence term to surface integrals. After that, these terms are evaluated as fluxes at each finite volume’s surfaces.

2.5. Turbulent models

Turbulence is the chaotic, unstable motion of fluids that occurs when there are insufficient stabilizing viscous forces. In FLOW-3D, there are six turbulence models available: the Prandtl mixing length model, the one-equation turbulent energy model, the two-equation (k – ε) model, the Renormalization-Group (RNG) model, the two-equation (k – ω) models, and a large eddy simulation (LES) model. For simulating flow motion, the RNG model is adopted to simulate the motion behavior better than the k – ε and k – ω.

models [30]. The RNG model consists of two main equations for the turbulent kinetic energy KT and its dissipation.εT(5)∂kT∂t+1VFuAx∂kT∂x+vAy∂kT∂y+wAz∂kT∂z=PT+GT+DiffKT-εT(6)∂εT∂t+1VFuAx∂εT∂x+vAy∂εT∂y+wAz∂εT∂z=C1.εTKTPT+c3.GT+Diffε-c2εT2kT

where KT is the turbulent kinetic energy, PT is the turbulent kinetic energy production, GT is the buoyancy turbulence energy, εT is the turbulent energy dissipation rate, DiffKT and Diffε are terms of diffusion, c1, c2 and c3 are dimensionless parameters, in which c1 and c3 have a constant value of 1.42 and 0.2, respectively, c2 is computed from the turbulent kinetic energy (KT) and turbulent production (PT) terms.

2.6. Sediment scour model

The sediment scour model available in FLOW-3D can calculate all the sediment transport processes including Entrainment transport, Bedload transport, Suspended transport, and Deposition. The erosion process starts once the water flows remove the grains from the packed bed and carry them into suspension. It happens when the applied shear stress by water flows exceeds critical shear stress. This process is represented by entrainment transport in the numerical model. After entrained, the grains carried by water flow are represented by suspended load transport. After that, some suspended grains resort to settling because of the combined effect of gravity, buoyancy, and friction. This process is described through a deposition. Finally, the grains sliding motions are represented by bedload transport in the model. For the entrainment process, the shear stress applied by the fluid motion on the packed bed surface is calculated using the standard wall function as shown in Eq.7.(7)ks,i=Cs,i∗d50

where ks,i is the Nikuradse roughness and Cs,i is a user-defined coefficient. The critical bed shear stress is defined by a dimensionless parameter called the critical shields number as expressed in Eq.8.(8)θcr,i=τcr,i‖g‖diρi-ρf

where θcr,i is the critical shields number, τcr,i is the critical bed shear stress, g is the absolute value of gravity acceleration, di is the diameter of the sediment grain, ρi is the density of the sediment species (i) and ρf is the density of the fluid. The value of the critical shields number is determined according to the Soulsby-Whitehouse equation.(9)θcr,i=0.31+1.2d∗,i+0.0551-exp-0.02d∗,i

where d∗,i is the dimensionless diameter of the sediment, given by Eq.10.(10)d∗,i=diρfρi-ρf‖g‖μf213

where μf is the fluid dynamic viscosity. For the sloping bed interface, the value of the critical shields number is modified according to Eq.11.(11)θ`cr,i=θcr,icosψsinβ+cos2βtan2φi-sin2ψsin2βtanφi

where θ`cr,i is the modified critical shields number, φi is the angle of repose for the sediment, β is the angle of bed slope and ψ is the angle between the flow and the upslope direction. The effects of the rolling, hopping, and sliding motions of grains along the packed bed surface are taken by the bedload transport process. The volumetric bedload transport rate (qb,i) per width of the bed is expressed in Eq.12.(12)qb,i=Φi‖g‖ρi-ρfρfdi312

where Φi is the dimensionless bedload transport rate is calculated by using Meyer Peter and Müller equation.(13)Φi=βMPM,iθi-θ`cr,i1.5cb,i

where βMPM,i is the Meyer Peter and Müller user-defined coefficient and cb,i is the volume fraction of species i in the bed material. The suspended load transport is calculated as shown in Eq.14.(14)∂Cs,i∂t+∇∙Cs,ius,i=∇∙∇DCs,i

where Cs,i is the suspended sediment mass concentration, D is the diffusivity, and us,i is the grain velocity of species i. Entrainment and deposition are two opposing processes that take place at the same time. The lifting and settling velocities for both entrainment and deposition processes are calculated according to Eq.15 and Eq.16, respectively.(15)ulifting,i=αid∗,i0.3θi-θ`cr,igdiρiρf-1(16)usettling,i=υfdi10.362+1.049d∗,i3-10.36

where αi is the entrainment coefficient of species i and υf is the kinematic viscosity of the fluid.

2.7. Grid type

Using simple rectangular orthogonal elements in planes and hexahedral in volumes in the (FLOW-3D) program makes the mesh generation process easier, decreases the required memory, and improves numerical accuracy. Two mesh blocks were used in a joined form with a size ratio of 2:1. The first mesh block is coarser, which contains the reservoir water, and the second mesh block is finer, which contains the dam. For achieving accuracy and efficiency in results, the mesh size is determined by using a grid convergence test. The optimum uniform cell size for the first mesh block is 0.012 m and for the second mesh block is 0.006 m.

2.8. Time step

The maximum time step size is determined by using a Courant number, which controls the distance that the flow will travel during the simulation time step. In this study, the Courant number was taken equal to 0.25 to prevent the flow from traveling through more than one cell in the time step. Based on the Courant number, a maximum time step value of 0.00075 s was determined.

2.9. Numerical model validation

The numerical model accuracy was achieved by comparing the numerical model results with previous experimental results. The experimental study of Schmocker and Hager [7] was based on 31 tests with changes in six parameters (d₅₀, H_o, B_o, L_k, X_D, and Q_in). All experimental tests were conducted in a straight open glass-sided flume. The horizontal flume has a rectangular cross-section with a width of 0.4 m and a height of 0.7 m. The flume was provided with a flow straightener and an intake with a length of 0.66 m. All tested dams were inserted at various distances (X_D) from the intake. Test No.1 from this experimental program was chosen to validate the numerical model. The different parameters used in test No.1 are as follows:

(1) uniform sediment with a mean diameter (d₅₀ = 0.31 mm), (2) H_o = 0.2 m, (3) B_o = 0.2 m, (4) L_k = 0.1 m,

(5) X_D = 1.0 m, (6) Q_in = 6.0 lit/s, (7) S_u and S_d = 2:1, (8) mass density (ρ_s = 2650 kg/m³) (9) Homogenous and non-cohesive embankment dam. As shown in Fig. 2, the simulation is contained within a rectangular grid with dimensions: 3.56 m in the x-direction (where 0.66 m is used as inlet, 0.9 m as dam base width, and 1.0 m as outlet), in y-direction 0.2 m (dam length), and in the z-direction 0.3 m, which represents the dam height (0.2 m) with a free distance (0.1 m) above the dam. There are two main reasons that this experimental program is preferred for the validation process. The first reason is that this program deals with homogenous, non-cohesive soil, which is available in FLOW-3D. The second reason is that this program deals with small-scale models which saves time for numerical simulation. Finally, some important assumptions were considered during the validation process. The flow is assumed to be incompressible, viscous, turbulent, and three-dimensional.

By comparing dam profiles at different time instants for the experimental test with the current numerical model, it appears that the numerical model gives good agreement as shown in Fig. 3 and Fig. 4, with an average error percentage of 9% between the experimental results and the numerical model.

3. Analysis and discussions

The current model is used to study the effects of different parameters such as (initial breach shapes, dimensions, locations, upstream and downstream dam slopes) on the peak outflow discharge, Q_P, time of peak outflow, t_P, and rate of erosion, E.

This study consists of a group of scenarios. The first scenario is changing the shapes of the initial breach according to Singh [1], the most predicted shapes are rectangular and V-notch as shown in Fig. 5. The second scenario is changing the initial breach dimensions (i.e., width and depth). While the third scenario is changing the location of the initial breach. Eventually, the last scenario is changing the upstream and downstream dam slopes.

All scenarios of this study were carried out under the same conditions such as inflow discharge value (Qin=1.0lit/s), dimensions of the tested dam, where dam height (Ho=0.20m), crest width.

(Lk=0.1m), dam length (Bo=0.20m), and homogenous & non-cohesive soil with a mean diameter (d50=0.31mm).

3.1. Dam breaching process evolution

The dam breaching process is a very complex process due to the quick changes in hydrodynamic conditions during dam failure. The dam breaching process starts once water flows reach the downstream face of the dam. During the initial stage of dam breaching, the erosion process is relatively quiet due to low velocities of flow. As water flows continuously, erosion rates increase, especially in two main zones: the crest and the downstream face. As soon as the dam crest is totally eroded, the water levels in the dam reservoir decrease rapidly, accompanied by excessive erosion in the dam body. The erosion process continues until the water levels in the dam reservoir equal the remaining height of the dam.

According to Zhou et al. [11], the breaching process consists of three main stages. The first stage starts with beginning overtopping flow, then ends when the erosion point directed upstream and reached the inflection point at the inflection time (ti). The second stage starts from the end of the stage1 until the occurrence of peak outflow discharge at the peak outflow time (tP). The third stage starts from the end of the stage2 until the value of outflow discharge becomes the same as the value of inflow discharge at the final time (tf). The outflow discharge from the dam breach increases rapidly during stage1 and stage2 because of the large dam storage capacity (i.e., the dam reservoir is totally full of water) and excessive erosion. While at stage3, the outflow values start to decrease slowly because most of the dam’s storage capacity was run out. The end of stage3 indicates that the dam storage capacity was totally run out, so the outflow equalized with the inflow discharge as shown in Fig. 6 and Fig. 7.

3.2. The effect of initial breach shape

To identify the effect of the initial breach shape on the evolution of the dam breaching process. Three tests were carried out with different cross-section areas for each shape. The initial breach is created at the center of the dam crest. Each test had an ID to make the process of arranging data easier. The rectangular shape had an ID (Rec5h & 5b), which means that its depth and width are equal to 5% of the dam height, and the V-notch shape had an ID (V-noch5h & 1:1) which means that its depth is equal to 5% of the dam height and its side slope is equal to 1:1. The comparison between rectangular and V-notch shapes is done by calculating the ratio between maximum dam height at different times (Z_Max) to the initial dam height (H_o), rate of erosion, and hydrograph of outflow discharge for each test. The rectangular shape achieves maximum erosion rate and minimum inflection time, in addition to a rapid decrease in the dam reservoir levels. Therefore, the dam breaching is faster in the case of a rectangular shape than in a V-notch shape, which has the same cross-section area as shown in Fig. 8.

Also, by comparing the hydrograph for each test, the peak outflow discharge value in the case of a rectangular shape is higher than the V-notch shape by 5% and the time of peak outflow for the rectangular shape is shorter than the V-notch shape by 9% as shown in Fig. 9.

3.3. The effect of initial breach dimensions

The results of the comparison between the different initial breach shapes indicate that the worst initial breach shape is rectangular, so the second scenario from this study concentrated on studying the effect of a change in the initial rectangular breach dimensions. Groups of tests were carried out with different depths and widths for the rectangular initial breach. The first group had a depth of 5% from the dam height and with three different widths of 5,10, and 15% from the dam height, the second group had a depth of 10% with three different widths of 5,10, and 15%, the third group had a depth of 15% with three different widths of 5,10, and 15% and the final group had a width of 15% with three different heights of 5, 10, and 15% for a rectangular breach shape. The comparison was made as in the previous section to determine the worst case that leads to the quick dam failure as shown in Fig. 10.

The results show that the (Rec 5 h&15b) test achieves a maximum erosion rate for a shorter period of time and a minimum ratio for (Z_max / H_o) as shown in Fig. 10, which leads to accelerating the dam failure process. The dam breaching process is faster with the minimum initial breach depth and maximum initial breach width. In the case of a minimum initial breach depth, the retained head of water in the dam reservoir is high and the crest width at the bottom of the initial breach (L`K) is small, so the erosion point reaches the inflection point rapidly. While in the case of the maximum initial breach width, the erosion perimeter is large.

3.4. The effect of initial breach location

The results of the comparison between the different initial rectangular breach dimensions indicate that the worst initial breach dimension is (Rec 5 h&15b), so the third scenario from this study concentrated on studying the effect of a change in the initial breach location. Three locations were checked to determine the worst case for the dam failure process. The first location is at the center of the dam crest, which was named “Center”, the second location is at mid-distance between the dam center and dam edge, which was named “Mid”, and the third location is at the dam edge, which was named “Edge” as shown in Fig. 11. According to this scenario, the results indicate that the time of peak outflow discharge (tP) is the same in the three cases, but the maximum value of the peak outflow discharge occurs at the center location. The difference in the peak outflow values between the three cases is relatively small as shown in Fig. 12.

The rates of erosion were also studied for the three cases. The results show that the maximum erosion rate occurs at the center location as shown in Fig. 13. By making a comparison between the three cases for the dam storage volume. The results show that the center location had the minimum values for the dam storage volume, which means that a large amount of water has passed to the downstream area as shown in Fig. 14. According to these results, the center location leads to increased erosion rate and accelerated dam failure process compared with the two other cases. Because the erosion occurs on both sides, but in the case of edge location, the erosion occurs on one side.

3.5. The effect of upstream and downstream dam slopes

The results of the comparison between the different initial rectangular breach locations indicate that the worst initial breach location is the center location, so the fourth scenario from this study concentrated on studying the effect of a change in the upstream (S_u) and downstream (S_d) dam slopes. Three slopes were checked individually for both upstream and downstream slopes to determine the worst case for the dam failure process. The first slope value is (2H:1V), the second slope value is (2.5H:1V), and the third slope value is (3H:1V). According to this scenario, the results show that the decreasing downstream slope angle leads to increasing time of peak outflow discharge (tP) and decreasing value of peak outflow discharge. The difference in the peak outflow values between the three cases for the downstream slope is 2%, as shown in Fig. 15, but changing the upstream slope has a negligible impact on the peak outflow discharge and its time as shown in Fig. 16.

The rates of erosion were also studied in the three cases for both upstream and downstream slopes. The results show that the maximum erosion rate increases by 6.0% with an increasing downstream slope angle by 4°, as shown in Fig. 17. The results also indicate that the erosion rates aren’t affected by increasing or decreasing the upstream slope angle, as shown in Fig. 18. According to these results, increasing the downstream slope angle leads to increased erosion rate and accelerated dam failure process compared with the upstream slope angle. Because of increasing shear stress applied by water flows in case of increasing downstream slope.

According to all previous scenarios, the dimensionless peak outflow discharge QPQin is presented for a fixed dam height (H_o) and inflow discharge (Q_in). Fig. 19 illustrates the relationship between QP∗=QPQin and.

Lr=ho2/3∗bo2/3Ho. The deduced relationship achieves R2=0.96.(17)QP∗=2.2807exp-2.804∗Lr

4. Conclusions

A spatial dam breaching process was simulated by using FLOW-3D Software. The validation process was performed by making a comparison between the simulated results of dam profiles and the dam profiles obtained by Schmocker and Hager [7] in their experimental study. And also, the peak outflow value recorded an error percentage of 12% between the numerical model and the experimental study. This model was used to study the effect of initial breach shape, dimensions, location, and dam slopes on peak outflow discharge, time of peak outflow, and the erosion process. By using the parameters obtained from the validation process, the results of this study can be summarized in eight points as follows.1.

The rectangular initial breach shape leads to an accelerating dam failure process compared with the V-notch.2.

The value of peak outflow discharge in the case of a rectangular initial breach is higher than the V-notch shape by 5%.3.

The time of peak outflow discharge for a rectangular initial breach is shorter than the V-notch shape by 9%.4.

The minimum depth and maximum width for the initial breach achieve maximum erosion rates (increasing breach width, b0, or decreasing breach depth, h0, by 5% from the dam height leads to an increase in the maximum rate of erosion by 11% and 15%, respectively), so the dam failure is rapid.5.

The center location of the initial breach leads to an accelerating dam failure compared with the edge location.6.

The initial breach location has a negligible effect on the peak outflow discharge value and its time.7.

Increasing the downstream slope angle by 4° leads to an increase in both peak outflow discharge and maximum rate of erosion by 2.0% and 6.0%, respectively.8.

The upstream slope has a negligible effect on the dam breaching process.

References

[1]V. SinghDam breach modeling technologySpringer Science & Business Media (1996)Google Scholar [2]Wahl TL. Prediction of embankment dam breach parameters: a literature review and needs assessment. 1998.Google Scholar [3]Z. Alhasan, J. Jandora, J. ŘíhaStudy of dam-break due to overtopping of four small dams in the Czech RepublicActa Universitatis Agriculturae et Silviculturae Mendelianae Brunensis, 63 (3) (2015), pp. 717-729 View PDF CrossRef View Record in Scopus Google Scholar [4]D. FreadBREACH, an erosion model for earthen dam failures: Hydrologic Research LaboratoryNOAA, National Weather Service (1988)Google Scholar [5]J. Říha, S. Kotaška, L. PetrulaDam Break Modeling in a Cascade of Small Earthen Dams: Case Study of the Čižina River in the Czech RepublicWater, 12 (8) (2020), p. 2309, 10.3390/w12082309 View PDF View Record in Scopus Google Scholar [6]E. Goodarzi, L. Teang Shui, M. ZiaeiDam overtopping risk using probabilistic concepts–Case study: The Meijaran DamIran Ain Shams Eng J, 4 (2) (2013), pp. 185-197Article Download PDF View Record in Scopus Google Scholar [7]L. Schmocker, W.H. HagerPlane dike-breach due to overtopping: effects of sediment, dike height and dischargeJ Hydraul Res, 50 (6) (2012), pp. 576-586 View PDF CrossRef View Record in Scopus Google Scholar [8]J.S. Walder, R.M. Iverson, J.W. Godt, M. Logan, S.A. SolovitzControls on the breach geometry and flood hydrograph during overtopping of noncohesive earthen damsWater Resour Res, 51 (8) (2015), pp. 6701-6724View Record in Scopus Google Scholar [9]H. Wei, M. Yu, D. Wang, Y. LiOvertopping breaching of river levees constructed with cohesive sedimentsNat Hazards Earth Syst Sci, 16 (7) (2016), pp. 1541-1551 View PDF CrossRef View Record in Scopus Google Scholar [10]Y. Yang, S.-Y. Cao, K.-J. Yang, W.-P. LiYang K-j, Li W-p. Experimental study of breach process of landslide dams by overtopping and its initiation mechanismsJ Hydrodynamics, 27 (6) (2015), pp. 872-883Article Download PDF CrossRef View Record in Scopus Google Scholar [11]G.G.D. Zhou, M. Zhou, M.S. Shrestha, D. Song, C.E. Choi, K.F.E. Cui, et al.Experimental investigation on the longitudinal evolution of landslide dam breaching and outburst floodsGeomorphology, 334 (2019), pp. 29-43Article Download PDF View Record in Scopus Google Scholar [12]J. Zhang, Z.-x. Guo, S.-y. CaoYang F-g. Experimental study on scour and erosion of blocked damWater Sci Eng, 5 (2012), pp. 219-229Article Download PDF View Record in Scopus Google Scholar [13]K. Höeg, A. Løvoll, K. VaskinnStability and breaching of embankment dams: Field tests on 6 m high damsInt J Hydropower Dams, 11 (2004), pp. 88-92View Record in Scopus Google Scholar [14]H. Hakimzadeh, V. Nourani, A.B. AminiGenetic programming simulation of dam breach hydrograph and peak outflow dischargeJ Hydrol Eng, 19 (4) (2014), pp. 757-768View Record in Scopus Google Scholar [15]A.R. Refaiy, N.M. AboulAtta, N.Y. Saad, D.A. El-MollaModeling the effect of downstream drain geometry on seepage through earth damsAin Shams Eng J, 12 (3) (2021), pp. 2511-2531Article Download PDF View Record in Scopus Google Scholar [16]Y. Zhu, P.J. Visser, J.K. Vrijling, G. WangExperimental investigation on breaching of embankmentsScience China Technological Sci, 54 (1) (2011), pp. 148-155 View PDF CrossRef View Record in Scopus Google Scholar [17]M.-H. Yu, H.-Y. Wei, Y.-J. Liang, Y. ZhaoInvestigation of non-cohesive levee breach by overtopping flowJ Hydrodyn, 25 (4) (2013), pp. 572-579Article Download PDF CrossRef View Record in Scopus Google Scholar [18]S. Wu, M. Yu, H. Wei, Y. Liang, J. ZengNon-symmetrical levee breaching processes in a channel bend due to overtoppingInt J Sedim Res, 33 (2) (2018), pp. 208-215Article Download PDF View Record in Scopus Google Scholar [19]O. Saberi, G. ZenzNumerical investigation on 1D and 2D embankment dams failure due to overtopping flowInt J Hydraulic Engineering, 5 (2016), pp. 9-18View Record in Scopus Google Scholar [20]M. Guan, N.G. Wright, P.A. Sleigh2D Process-Based Morphodynamic Model for Flooding by Noncohesive Dyke BreachJ Hydraul Eng, 140 (7) (2014), p. 04014022, 10.1061/(ASCE)HY.1943-7900.0000861 View PDF View Record in Scopus Google Scholar [21]W. Wu, R. Marsooli, Z. HeDepth-Averaged Two-Dimensional Model of Unsteady Flow and Sediment Transport due to Noncohesive Embankment Break/BreachingJ Hydraul Eng, 138 (6) (2012), pp. 503-516View Record in Scopus Google Scholar [22]Z. Wang, D.S. BowlesThree-dimensional non-cohesive earthen dam breach model. Part 1: Theory and methodologyAdv Water Resour, 29 (10) (2006), pp. 1528-1545Article Download PDF View Record in Scopus Google Scholar [23]Říha J, Duchan D, Zachoval Z, Erpicum S, Archambeau P, Pirotton M, et al. Performance of a shallow-water model for simulating flow over trapezoidal broad-crested weirs. J Hydrology Hydromechanics. 2019;67:322-8.Google Scholar [24]C.B. VreugdenhilNumerical methods for shallow-water flowSpringer Science & Business Media (1994)Google Scholar [25]L.A. Larocque, J. Imran, M.H. Chaudhry3D numerical simulation of partial breach dam-break flow using the LES and k–∊ turbulence modelsJ Hydraul Res, 51 (2) (2013), pp. 145-157 View PDF CrossRef View Record in Scopus Google Scholar [26]C. Yang, B. Lin, C. Jiang, Y. LiuPredicting near-field dam-break flow and impact force using a 3D modelJ Hydraul Res, 48 (6) (2010), pp. 784-792 View PDF CrossRef View Record in Scopus Google Scholar [27]FLOW-3D. Version 11.1.1 Flow Science, Inc., Santa Fe, NM. https://wwwflow3dcom.Google Scholar [28]C.W. Hirt, B.D. NicholsVolume of fluid (VOF) method for the dynamics of free boundariesJ Comput Phys, 39 (1) (1981), pp. 201-225Article Download PDF Google Scholar [29]S.V. PatankarNumerical heat transfer and fluid flow, Hemisphere PublCorp, New York, 58 (1980), p. 288View Record in Scopus Google Scholar [30]M. Alemi, R. MaiaNumerical simulation of the flow and local scour process around single and complex bridge piersInt J Civil Eng, 16 (5) (2018), pp. 475-487 View PDF CrossRef View Record in Scopus Google Scholar

PDF 전체 원문 보기

FLOW-3D 모형을 이용한 해상풍력기초 세굴현상 분석

박영진¹, 김태원^2*1 서일대학교 토목공학과, ² (주)지티이

Analysis of Scour Phenomenon around Offshore Wind Foundation using Flow-3D Mode

Abstract

국내․외에서 다양한 형태의 석유 대체에너지는 온실효과 가스를 배출하지 않는 청정에너지로 개발되고 있으며, 특히 해상풍력은 풍력 자원이 풍부하고 육상보다 풍력 감소가 상대적으로 작아 다양하게 연구되고 있다. 본 연구에서는 해상 풍력기초의 세굴현상을 분석하기 위해서 Flow-3D 모형을 이용하여 모노 파일과 삼각대 파일 기초에 대하여 수치모의를 수행 하였다. 직경이 다른(D=5.0 m, d=1.69 m) 모노 파일 형식과 직경이 동일한(D=5.0 m) 모노파일에 대하여 세굴현상을 평가하 였다. 수치해석 결과, 동일한 직경을 가진 모노파일에서 하강류가 증가되었으며, 최대세굴심은 약 1.7배 이상 발생하였다. 삼각대 파일에 대하여 관측유속과 극치파랑 조건을 상류경계조건으로 각각 적용한 후 세굴현상을 평가하였다. 극치파랑조건 을 적용한 경우 최대 세굴심은 약 1.3배 정도 깊게 발생하였다. LES 모형을 적용하였을 경우 세굴심은 평형상태에 도달한 반면, RNG  모형은 해석영역 내 전반적으로 세굴현상이 발생하였으며, 세굴심은 평형상태에 도달하지 않았다. 해상풍 력기초에 대하여 세굴현상을 평가하기 위해서 수치모형 적용시 파랑조건 및 LES 난류모형을 적용하는 것이 타당할 것으로 판단된다.

Various types of alternative energy sources to petroleum are being developed both domestically and internationally as clean energy that does not emit greenhouse gases. In particular, offshore wind power has been studied because the wind resources are relatively limitless and the wind power is relatively smaller than onshore. In this study, to analyze the scour phenomenon around offshore wind foundations, mono pile and tripod pile foundations were simulated using a FLOW-3D model. The scour phenomenon was evaluated for mono piles: one is a pile with a 5 m diameter and d=1.69 m and the other is a pile with a 5 m diameter. Numerical analysis showed that in the latter, the falling-flow increased and the maximum scour depth occurred more than 1.7 times. For a tripod pile foundation, the measured velocity and the maximum wave condition were applied to the upstream boundary condition, respectively, and the scour phenomenon was evaluated. When the maximum wave condition was applied, the maximum scour depth occurred more than about 1.3 times. When the LES model was applied, the scour depth reached equilibrium, whereas the numerical results of the RNG model show that the scour phenomenon occurred in the entire boundary area and the scour depth did not reach equilibrium. To evaluate the scour phenomenon around offshore wind foundations, it is reasonable to apply the wave condition and the LES turbulence model to numerical model applications.

Keywords : Flow-3D, LES model, Mono pile, Offshore wind foundation, RNG k-e model, Scour phenomenon, Tripod pile

서론

지구환경문제에 대한 관심이 증가되고 있는 현실에 서, 풍력발전은 석유 대체에너지로서 뿐만 아니라, 이산 화탄소 등 온실효과 가스를 배출하지 않는 청청에너지의 발전방식으로 국내․외에서 개발이 증가되고 있다. 특 히, 해상풍력은 풍력 자원이 풍부하고, 육상보다 풍력 감 소가 상대적으로 작아 전기 출력량이 크기 때문에 신재 생에너지원 확보 차원에서 국내․외 해상풍력단지 사업 계획이 수립되어 추진되고 있는 실정이다. Fig. 1은 세계 최대 네델란드 해상풍력단지인 Nysted Offshore Wind Farm의 사진이다.

하천 내 교각 주변에서 세굴 현상은 발생하며 교각의 안정성 측면에서 세굴보호공을 설치한다. 해양에서 해상 풍력발전 기초를 설치할 경우 구조물로 인해 교란된 흐 름은 세굴을 유발시킨다. 따라서 해상풍력기초를 계획할 경우 안정성 측면에서 세굴현상을 검토할 필요가 있다. 특히 하천의 경우 교각 세굴보호공에 대하여 다양한 공 법들이 설계에 반영되고 있으나, 해양구조물 기초에 대 한 연구는 미흡한 상태이다.

이에 본 연구에서는 수치모 형을 이용하여 해상풍력기초에 대한 세굴현상을 분석하 였다. 수치모형을 이용하여 세굴현상을 예측함에 있어서 본 연구와 연관된 연구동향으로는 양원준과 최성욱(2002) 은 FLOW-3D 모형을 이용하여 세굴영향 평가를 함에 있어서 난류모형을 비교․분석 하였다. 전반적으로 수리 모형실험 자료와 좀 더 잘 일치하는 난류모형은 LES 모 형으로 분석되었다[1]. 여창건 등(2010)은 세굴영향 평 가를 위해 FLOW-3D 모형을 이용할 경우 세굴에 미치 는 중요한 인자에 대하여 매개변수 민감도분석을 수행하 였다.

검토결과, 세굴에 민감한 변수는 유사의 입경, 세 굴조절계수, 안식각 등의 순서로 민감한 것으로 검토되 었다[2]. 오명학 등(2012)은 해상풍력발전기초 시설 주 변에서 FLOW-3D 모형을 이용하여 세굴영향 검토를 수 행하였다. 오명학 등이 검토한 지역은 본 연구 지역과 동 일한 지역이나 경계조건 및 세굴평가에서 가장 중요한 평균입경이 다르다. 세굴검토를 위해 수치모형에 입력한 경계조건은 대조기 창조 최강유속 1.0 m/s을 상류경계조 건으로, 평균입경은 0.0353 mm를 적용하였다. 이와 같은 조건에서 모노파일에서 발생하는 최대세굴심은 약 5.24 m로 분석되었다[3].

Stahlmann과 Schlurmann(2010)은 본 과업에서 적용할 해상풍력기초와 유사한 기초를 가진 구조물에 대하여 수리모형실험을 수행하였다. 연구대상 지역은 독일 해안가에 의한 해상풍력단지에 대하여 삼각 대 형식의 해상풍력기초에 대하여 1/40과 1/12 축척으로 각각 수리모형실험을 수행하였다. 1/40과 1/12 축척에 따라서 세굴분포양상 및 최대세굴심의 위치가 다르게 관 측되었다[4].

본 연구에서는 3차원 수치모형인 Flow-3D를 이용하 여 세굴현상을 평가함에 있어서, 파일 형상 변화, 경계조 건이 다른 경우 및 서로 다른 난류모형을 적용하였을 경 우에 대하여 수치해석이 국부세굴 현상에 미치는 영향을 검토하였다. 이와 같은 연구는 향후 수치모형을 이용하 여 해상풍력발전 기초에 대하여 세굴현상을 평가함에 있 어서 기초 자료로 활용될 수 있을 것으로 판단된다.

Fig. 3. Boundary Area and Grid of Flow-3D

Fig. 5. Velocity Development around Monopile

Fig. 6. Flow Phenomenon and Scour around Tripod Pile Foundation

Fig. 7. Scour according to Turbulence Models(RNG k-e & LES Model)

결론

본 연구에서는 해상풍력기초 형식이 모노파일과 삼각 대 파일일 경우 세굴현상을 평가하기 위해서 3차원 수치 모형인 Flow-3D를 이용하였다. 직경이 서로 다른(D=5.0 m, d=1.69 m) 모노파일과 직경이 동일한(D=5.0 m) 모노파일에 대하여 LES 모형 을 적용하여 세굴현상을 평가하였다. 서로 다른 직경을 가진 모노파일 주변에서 최대 세굴심은 4.13 m, 동일한 직경을 가진 모노파일 주변에서는 7.13 m의 최대 세굴 심이 발생하였다. 또한 동일한 직경을 가진 파일에서 하 강류가 증가되어 최대세굴심이 증가된 것으로 분석되었 다. 수치해석 결과, 세굴에 대한 기초의 안정성 측면에서 서로 다른 직경을 가진 기초 형식이 유리한 것으로 분석 되었다. 수치모형을 이용하여 세굴현상을 평가함에 있어서 경 계조건 및 난류모형의 선정은 중요하다. 본 연구에서는 서로 다른 직경을 가진 삼각대 형식의 해상풍력기초에 대하여 상류경계조건으로 관측유속과 극치파랑조건을 각각 적용하였을 경우 세굴현상을 평가하였다. 극치파랑 조건을 적용하였을 경우가 최대세굴심이 약 1.3배 정도 깊게 발생하였다. 또한 극치파랑조건에서 RNG 과 LES 모형을 적용하여 세굴현상을 평가하였다. LES 모 형을 적용하였을 경우 파일 주변에서 세굴현상이 발생하 였으며, 세굴심은 일정시간이 경과된 후에는 증가되지 않는 평형상태에 도달하였다. 그러나 RNG 모형을 적용한 경우는 평형상태에 도달하지 않고 계속해서 세굴 이 진행되어 세굴심을 평가할 수 없었다. 현재 해양구조 물 기초에 대한 세굴현상 연구는 미흡한 상태로 하천에 서 교각 세굴현상을 검토하기 위해서 적용되는 경계조건 을 적용하기보다는 해상 조건인 파랑조건을 적용하여 검 토하는 것이 기초의 안정성 측면에서 유리할 것으로 판 단된다. 또한 정확한 세굴현상을 예측하기 위해서는 RNG 모형보다는 LES 모형을 적용하는 것이 타당 할 것으로 판단된다. 향후 해상풍력기초에 대한 세굴관측을 수행하여 수치 모의 결과와 비교․분석이 필요하며, 또한 다양한 파랑 조건에서 난류모형에 대한 비교․분석이 필요할 것으로 생각된다.

References

[1] W. J. Yang, S. U. Choi. “Three- Dimensional Numerical
Simulation of Local Scour around the Bridge Pier using
Large Eddy Simulation”, Journal of KWRA, vol. 22, no.
4-B, pp. 437-446, 2002.
[2] C. G. Yeo, J. E. Lee, S. O. Lee, J. W. Song. “Sensitivity
Analysis of Sediment Scour Model in Flow-3D”,
Proceedings of KWRA, pp. 1750-1754, 2010.
[3] M. H. Oh, O. S. Kwon, W. M. Jeong, K. S. Lee.
“FLOW-3D Analysis on Scouring around Offshore Wind
Foundation”, Journal of KAIS, vol. 13, no. 3, pp.
1346-1351, 2012.
DOI: http://dx.doi.org/10.5762/KAIS.2012.13.3.1346

[4] A. Stahlmann, T. Schlurmann, “Physical Modeling of
Scour around Tripod Foundation Structures for Offshore
Wind Energy Converters”, Proceedings of 32nd
Conference on Coastal Engineering, Shanghai, China,
no. 32, pp. 1-12, 2010.
[5] Flow Science. Flow-3D User’s Manual. Los Alamos,
NM, USA, 2016.
[6] KEPRI. 『Test Bed for 2.5GW Offshore Wind Farm at
Yellow Sea』 Interim Design Report(in Korea), 2014.
[7] Germanischer Lloyd. Guideline for the Certification of
Offshore Wind Turbines. Hamburg, Germany, 2005.
[8] B. M. Sumer, J. Fredsøe, The Mechanics of Scour in the
Marine Environment. World Scientific Publishing Co.
Pte. Ltd. 2002.
[9] S. J. Ahn, U. Y. Kim, J. K. Lee. “Experimental Study
for Scour Protection around Bridge Pier by Falling-Flow
Interruption”, Journal of KSCE, vol. 19, no. II-1, pp.
57-65, 1999.
[10] V. Yakhot, S. A. Orszag, S. Thangam, T. B. Gatski, C.
G. Speziale, “Development of turbulence models for
shear flows by a double expansion technique”, Physics
of Fluids, vol. 4, no. 7, pp. 1510-1520, 1992.
DOI: https://doi.org/10.1063/1.858424

PDF View

FLOW-3D를 이용한 해상 자켓구조물 주변의 세굴 수치모의 실험

Numerical Simulation Test of Scour around Offshore Jacket Structure using FLOW-3D

J Korean Soc Coast Ocean Eng. 2015;27(6):373-381Publication date (electronic) : 2015 December 31doi : https://doi.org/10.9765/KSCOE.2015.27.6.373 Dong Hui Ko ^*, Shin Taek Jeong^,^**, Nam Sun Oh ^***^*Hae Poong Engineering Inc.^**Department of Civil and Environmental Engineering, Wonkwang University^***Ocean·Plant Construction Engineering, Mokpo Maritime National University
고동휘^*, 정신택^,^**, 오남선^***

^*(주)해풍기술^**원광대학교 토목환경공학과^***목포해양대학교 해양·플랜트건설공학과

Abstract

해상풍력 기기, 해상 플랫폼과 같은 구조물이 해상에서 빈번하게 설치되면서 세굴에 관한 영향도 중요시되고 있다. 이러한 세굴 영향을 검토하기 위해 세굴 수치모의 실험을 수행한다. 일반적으로 수치모의 조건은 일방향 흐름에 대해서만 검토가 이뤄지고 있으며 서해안과 같은 왕복성 조류 흐름에 대해서는 검토되지 않는다. 본 연구에서는 서해안에 설치된 HeMOSU-1호 해상 자켓구조물 주변에서 발생하는 세굴 현상을 FLOW-3D를 이용하여 수치모의하였다. 해석 조건으로는 일방향 흐름과 조석현상을 고려한 왕복성 흐름을 고려하였으며, 이를 현장 관측값과 비교하였다. 10,000초 동안의 수치모의 결과, 일방향의 흐름 조건에서는 1.32 m의 최대 세굴심이 발생하였으며, 양방향 흐름 조건에서는 1.44 m의 최대 세굴심이 발생하였다. 한편, 현장 관측값의 경우 약 1.5~2.0 m의 세굴심이 발생하여 양방향의 흐름에 대한 해석 결과와 근사한 값을 보였다.

Keywords : 세굴, 일방향 흐름, 왕복성 조류 흐름, 해상 자켓구조물, FLOW-3D, 최대 세굴심, scour, uni-directional flow, bi-directional tidal current flow, offshore jacket substructure, Flow-3D, maximum scour depth

As offshore structures such as offshore wind and offshore platforms have been installed frequently in ocean, scour effects are considered important. To test the scour effect, numerical simulation of scour has been carried out. However, the test was usually conducted under the uni-directional flow without bi-directional current flow in western sea of Korea. Thus, in this paper, numerical simulations of scour around offshore jacket substructure of HeMOSU-1 installed in western sea of Korea are conducted using FLOW-3D. The conditions are uni-directional and bi-directional flow considering tidal current. And these results are compared to measured data. The analysis results for 10,000 sec show that under uni-directional conditions, maximum scour depth was about 1.32 m and under bi-directional conditions, about 1.44 m maximum scour depth occurred around the structure. Meanwhile, about 1.5~2.0 m scour depths occurred in field observation and the result of field test is similar to result under bi-directional conditions.

1. 서 론

최근 해상풍력기기, 해상플랫폼과 같은 해상구조물 설치가 빈번해지면서 해상구조물의 안정성을 저하시키는 요인에 대한 대응 연구가 필요하다. 특히 해상에서의 구조물 설치는 육상과 달리 수력학적 하중이 작용하게 되기 때문에 파랑에 의한 구조물과의 진동, 세굴 현상에 대하여 철저한 사전 검토가 요구된다. 특히, 해상 기초에서 발생하는 세굴은 조류 및 파랑 등 유체 흐름과 구조물 사이의 상호작용으로 인해 해저 입자가 유실되는 현상으로 정의할 수 있으며 해상 외력 조건에 포함되어 설계시 고려하도록 제안하고 있다(IEC, 2009).구조물을 해상에 설치하게 되면 구조물이 흐름을 방해하는 장애요인으로 작용하여 구조물 주위에 부분적으로 더 빠른 유속이 발생하게 된다. 이러한 유속 변화는 압력 분포 변화에 기인하게 되어 해양구조물 주위에 아래로 흐르는 유속(downflow), 말굽형 와류(horseshoe vortex) 그리고 후류 와류(wake vortex)가 나타난다. 결국, 유속과 흐름의 변화를 야기하고 하상전단응력과 유사이동 능력을 증가시켜 해저 입자를 유실시키며 구조물의 안정성을 위협하는 요인으로 작용하게 된다. 이러한 세굴 현상이 계속 진행되면 해상풍력 지지구조물 기초의 지지력이 감소하게 될 뿐만 아니라 지지면의 유실로 상부반력 작용에 편심을 유발하여 기초의 전도를 초래한다. 또한 세굴에 의한 기초의 부등 침하가 크게 발생하면 상부 해상풍력 지지구조물에 보다 큰 단면력이 작용하므로 세굴에 의한 붕괴가 발생할 수 있다. 이처럼 세굴은 기초지지구조물을 붕괴하고, 침하와 얕은 기초의 변형을 초래하며, 구조물의 동적 성능을 변화시키기 때문에 설계 및 시공 유지관리시 사전에 세굴심도 산정, 세굴 완화 대책 등을 고려하여야 한다.또한 각종 설계 기준서에서는 세굴에 대해 다양하게 제시하고 있다. IEC(2009), ABS(2013), BSH(2007), MMAF(2005)에서는 세굴에 대한 영향을 검토할 것을 주문하지만 심도 산정 등 세굴에 대한 구체적인 내용은 언급하지 않고 전반적인 내용만 수록하고 있다. 그러나 DNV(2010), CEM(2006)에서는 경험 공식을 이용한 세굴 심도 산정 등 구체적인 내용을 광범위하게 수록하고 있어 세굴에 대한 영향 검토시 활용가능하다. 그 외의 기준서에서는 수치 모델 등을 통한 세굴 검토를 주문하고 있어 사용자들이 직접 판단하도록 제안하고 있다.그러나 세굴은 유속, 수심, 구조물 폭, 형상, 해저입자 등에 의해 결정되기 때문에 세굴의 영향 정도를 정확하게 예측하기란 쉽지 않지만 수리 모형 실험 또는 CFD(Computational Fluid Dynamics)를 이용한 수치 해석을 통해 지반 침식 및 퇴적으로 인한 지형변화를 예측할 수 있다. 한편, 침식과 퇴적 등 구조물 설치로 인한 해저 지형 변화를 예측하는 모델은 다양하지만, 본 연구에서는 Flowscience의 3차원 유동해석모델인 Flow-3D 모델을 사용하였다.해상 구조물은 목적에 따라 비교적 수심이 낮은 지역에 설치가 용이하다. 국내의 경우, 서남해안과 같이 비교적 연안역이 넓고 수심이 낮은 지역에 구조물을 설치하는 것이 비용 및 유지관리 측면에서 유리할 수 있다. 그러나 국내 서남해안 지역은 왕복성 흐름, 즉 조류가 발생하는 지역으로 흐름의 방향이 시간에 따라 변화하게 된다. 따라서, 세굴 수치 모의시 이러한 왕복성 흐름을 고려해야한다. 그러나 대부분의 수치 모델 적용시 조류가 우세한 지역에서도 일방향의 흐름에 대해서만 검토하며 왕복성 흐름에 의한 지층의 침식과 퇴적작용으로 인해 발생하는 해저 입자의 상호 보충 효과는 배제되게 된다. 또한 이로 인해 수치모델 결과에 많은 의구심이 발생하게 되며 현실성이 결여된 해석으로 보여질 수 있다. 이러한 왕복흐름의 영향을 검토하기 위해 Kim and Gang(2011)은 조류의 왕복류 흐름을 고려하여 지반의 수리 저항 성능 실험을 수행하였으며, 양방향이 일방향 흐름보다 세굴이 크게 발생하는 것을 발표하였다. 또한 Kim et al.(2012)은 흐름의 입사각에 따른 수리저항 실험을 수행하였으며 입사각이 커짐에 따라 세굴률이 증가하는 것으로 나타났다.본 연구에서는 단일방향 고정유속 그리고 양방향 변동유속조건에서 발생하는 지형 변화와 세굴 현상을 수치 모의하였으며, 이러한 비선형성 흐름변화에 따른 세굴 영향 정도를 검토하였다. 더불어 현장 관측 자료와의 비교를 통해 서남해안과 같은 왕복성 흐름이 발생하는 지역에서의 세굴 예측시 적절한 모델 수립 방안을 제안하고자 한다.

2. 수치해석 모형

본 연구에서는 Autodesk의 3D max 프로그램을 이용하여 지지구조물 형상을 제작하였으며, 수치해석은 미국 Flowscience가 개발한 범용 유동해석 프로그램인 FLOW-3D(Ver. 11.0.4.5)를 사용하였다. 좌표계는 직교 좌표계를 사용하였으며 복잡한 3차원 형상의 표현을 위하여 FAVOR 기법(Fractional Area/Volume Obstacle Representation Method)을 사용하였다. 또한 유한차분법에 FAVOR 기법을 도입한 유한체적법의 접근법을 사용하였으며 직교좌표계 에서 비압축성 유체의 3차원 흐름을 해석하기 위한 지배방정식으로는 연속방정식과 운동방정식이 사용되었다. 난류모형으로는 RNG(renormalized group)모델을 사용하였다.

2.1 FLOW-3D의 지배방정식

수식은 MathML 표현문제로 본 문서의 하단부의 원문바로가기 링크를 통해 원문을 참고하시기 바랍니다.

2.1.1 연속방정식

직교좌표계 (x,y,z)에서 비압축성 유체는 압축성 유체의 연속방정식에서 유도될 수 있으며 다음 식 (1)과 같다.

(1)

∂∂x(uAx)+∂∂y(vAy)+∂∂z(wAz)=RSORρ∂∂x(uAx)+∂∂y(vAy)+∂∂z(wAz)=RSORρ
여기서, u, v, w는 (x,y,z) 방향별 유체속도, A_x, A_y, A_z는 각 방향별 유체 흐름을 위해 확보된 면적비 (Area fraction), ρ는 유체 밀도, RSOR은 질량생성/소멸(Mass source/sink)항이다.

2.1.2 운동방정식

본 모형은 3차원 난류모형이므로 각각의 방향에 따른 운동량 방정식은 다음 식(2)~(4)와 같다.

(2)

∂u∂t+1VF(uAx∂u∂x+vAy∂u∂y+wAz∂u∂z) =−1ρ∂p∂x+Gx+fx−bx−RSORρVFu∂u∂t+1VF(uAx∂u∂x+vAy∂u∂y+wAz∂u∂z) =−1ρ∂p∂x+Gx+fx−bx−RSORρVFu

(3)

∂v∂t+1VF(uAx∂v∂x+vAy∂v∂y+wAz∂v∂z) =−1ρ∂p∂y+Gy+fy−by−RSORρVFv∂v∂t+1VF(uAx∂v∂x+vAy∂v∂y+wAz∂v∂z) =−1ρ∂p∂y+Gy+fy−by−RSORρVFv

(4)

∂w∂t+1VF(uAx∂w∂x+vAy∂w∂y+wAz∂w∂z) =−1ρ∂p∂z+Gz+fz−bz−RSORρVFw∂w∂t+1VF(uAx∂w∂x+vAy∂w∂y+wAz∂w∂z) =−1ρ∂p∂z+Gz+fz−bz−RSORρVFw여기서, RSOR은 질량생성/소멸(Mass source/sink)항, V_F는 체적비 (Volume fraction), p는 압력, G_x, G_y, G_z는 방향별 체적력항, f_x, f_y, f_z는 방향별 점성력항, b_x, b_y, b_z는 다공질 매체에서 방향별 흐름 손실이다.그리고 점성계수 µ에 대하여 점성력항은 다음 식 (5)~(7)과 같다.

(5)

ρVffx=wsx−{∂∂x(Axτxx)+R∂∂y(Ayτxy)+∂∂z(Azτxz)+ζx(Axτxx−Ayτyy)}ρVffx=wsx−{∂∂x(Axτxx)+R∂∂y(Ayτxy)+∂∂z(Azτxz)+ζx(Axτxx−Ayτyy)}

(6)

ρVffy=wsy−{∂∂x(Axτxy)+R∂∂y(Ayτyy)+∂∂z(Azτyz)+ζx(Axτxx−Ayτxy)}ρVffy=wsy−{∂∂x(Axτxy)+R∂∂y(Ayτyy)+∂∂z(Azτyz)+ζx(Axτxx−Ayτxy)}

(7)

ρVffz=wsz−{∂∂x(Axτxz)+R∂∂y(Ayτyz)+∂∂z(Azτzz)+ζx(Axτzz)}ρVffz=wsz−{∂∂x(Axτxz)+R∂∂y(Ayτyz)+∂∂z(Azτzz)+ζx(Axτzz)}여기서, ws_x, ws_y, ws_z는 벽전단응력이며, 벽전단응력은 벽 근처에서 벽 법칙 (law of the wall)을 따르며, 식 (8)~(13)에 의해 표현되어진다.

(8)

τxx=−2μ{∂u∂x−13(∂u∂x+R∂v∂y+∂w∂z+ζux)}τxx=−2μ{∂u∂x−13(∂u∂x+R∂v∂y+∂w∂z+ζux)}

(9)

τyy=−2μ{R∂v∂y+ζux−13(∂u∂x+R∂v∂y+∂w∂z+ζux)}τyy=−2μ{R∂v∂y+ζux−13(∂u∂x+R∂v∂y+∂w∂z+ζux)}

(10)

τzz=−2μ{R∂w∂y−13(∂u∂x+R∂v∂y+∂w∂z+ζux)}τzz=−2μ{R∂w∂y−13(∂u∂x+R∂v∂y+∂w∂z+ζux)}

(11)

τxy=−μ{∂v∂x+R∂u∂y−ζvx}τxy=−μ{∂v∂x+R∂u∂y−ζvx}

(12)

τxz=−μ{∂u∂y+∂w∂x}τxz=−μ{∂u∂y+∂w∂x}

(13)

τyz=−μ{∂v∂z+R∂w∂y}τyz=−μ{∂v∂z+R∂w∂y}

2.1.3 Sediment scour model

Flow-3D 모델에서 사용하는 sediment scour model은 해저입자의 특성에 따라 해저 입자의 침식, 이송, 전단과 흐름 변화로 인한 퇴적물의 교란 그리고 하상 이동을 계산한다.

2.1.3.1 The critical Shields parameter

무차원 한계소류력(the dimensionless critical Shields parameter)은 Soulsby-Whitehouse 식에 의해 다음 식 (14)와 같이 나타낼 수 있다(Soulsby, 1997).

(14)

θcr,i=0.31+1.2R∗i+0.055[1−exp(−0.02R∗i)]θcr,i=0.31+1.2Ri*+0.055[1−exp(−0.02Ri*)]여기서 무차원 상수, R∗iRi*는 다음 식 (15)와 같다.

(15)

R∗i=ds,i0.1(ρs,i−ρf)ρf∥g∥ds,i−−−−−−−−−−−−−−−−−−−√μfRi*=ds,i0.1(ρs,i−ρf)ρf‖g‖ds,iμf여기서 ρ_{s, i}는 해저 입자의 밀도, ρ_f는 유체 밀도, d_{s, i}는 해저입자 직경, g는 중력가속도이다.한편, 안식각에 따라 한계소류력은 다음 식 (16)과 같이 표현될 수 있다.

(16)

θ′cr,i=θcr,icosψsinβ+cos2βtan2ψi−sin2ψsin2β−−−−−−−−−−−−−−−−−−−−√tanψiθcr,i′=θcr,icosψsinβ+cos2βtan2ψi−sin2ψsin2βtanψi여기서, β는 하상 경사각, ψ_i는 해저입자의 안식각, ψ는 유체와 해저경사의 사잇각이다.또한 local Shields number는 국부 전단응력, τ에 기초하여 다음 식 (17)과 같이 계산할 수 있다.

(17)

θi=τ∥g∥ds,i(ρs,i−ρf)θi=τ‖g‖ds,i(ρs,i−ρf)여기서, ||g||g 는 중력 벡터의 크기이며, τ는 식 (8)~(13)의 벽 법칙을 이용하여 계산할 수 있다.

2.1.3.2 동반이행(Entrainment)과 퇴적

다음 식은 해저 지반과 부유사 사이의 교란을 나타내는 동반이행과 퇴적 현상을 계산한다. 해저입자의 동반이행 속도의 계산식은 다음 식 (18)과 같으며 부유사로 전환되는 해저의 양을 계산한다.

(18)

ulift,i=αinsd0.3∗(θi−θ′cr,i)1.5∥g∥ds,i(ρs,i−ρf)ρf−−−−−−−−−−−−−−√ulift,i=αinsd*0.3(θi−θcr,i′)1.5‖g‖ds,i(ρs,i−ρf)ρf여기서, α_i는 동반이행 매개변수이며, n_s는 the packed bed interface에서의 법선벡터, µ는 유체의 동점성계수 그리고 d_*은 무차원 입자 직경으로 다음 식 (19)와 같다.

(19)

d∗=ds,i[ρf(ρs,i−ρf)∥g∥μ2]1/3d*=ds,i[ρf(ρs,i−ρf)‖g‖μ2]1/3또한 퇴적 모델에서 사용하는 침강 속도 식은 다음 식 (20)같이 나타낼 수 있다.

(20)

usettling,i=νfds,i[(10.362+1.049d3∗)0.5−10.36]usettling,i=νfds,i[(10.362+1.049d*3)0.5−10.36]여기서, ν_f는 유체의 운동점성계수이다.

2.1.3.3 하상이동 모델(Bedload transport)

하상이동 모델은 해저면에 대한 단위 폭당 침전물의 체적흐름을 예측하는데 사용되며 다음 식 (21)과 같이 표현되어진다.

(21)

Φi=βi(θi−θ′cr,i)1.5Φi=βi(θi−θcr,i′)1.5여기서 Φ_i는 무차원 하상이동률이며 β_i는 일반적으로 8.0의 값을 사용한다(van Rijn, 1984).단위 폭당 체적 하상이동률, q_i는 다음 식 (22)와 같이 나타낼 수 있다.

(22)

qb,i=fb,i Φi[∥g∥(ρs,i−ρfρf)d3s,i]1/2qb,i=fb,i Φi[‖g‖(ρs,i−ρfρf)ds,i3]1/2여기서, f_{b, i}는 해저층의 입자별 체적률이다.또한 하상이동 속도를 계산하기 위해 다음 식 (23)에 의해 해저면층 두께를 계산할 수 있다.

(23)

δi=0.3ds,id0.7∗(θiθ′cr,i−1)0.5δi=0.3ds,id*0.7(θiθcr,i′−1)0.5그리고 하상이동 속도 식은 다음 식 (24)와 같이 계산되어진다.

(24)

ubedload,i=qb,iδifb,iubedload,i=qb,iδifb,i

2.2 모델 구성 및 해역 조건

2.2.1 해역 조건 및 적용 구조물

본 수치해석은 위도와 안마도 사이의 해양 조건을 적용하였으며 지점은 Fig. 1과 같다.

본 해석 대상 해역은 서해안의 조석 현상이 뚜렷한 지역으로 조류 흐름이 지배적이며 위도의 조화분석의 결과를 보면 조석형태수가 0.21로서 반일주조 형태를 취한다. 또한 북동류의 창조류와 남서류의 낙조류의 특성을 보이며 조류의 크기는 대상 영역에서 0.7~1 m/s의 최강유속 분포를 보이는 것으로 발표된 바 있다. 또한 대상 해역의 시추조사 결과를 바탕으로 해저조건은 0.0353 mm 로 설정하였고(KORDI, 2011), 수위는 등수심도를 바탕으로 15 m로 하였다.한편, 풍황자원 분석을 통한 단지 세부설계 기초자료 제공, 유속, 조류 등 해양 환경변화 계측을 통한 환경영향평가 기초자료 제공을 목적으로 Fig. 2와 같이 해상기상탑(HeMOSU-1호)을 설치하여 운영하고 있다. HeMOSU-1호는 평균해수면 기준 100 m 높이이며, 중량은 100 톤의 자켓구조물로 2010년 설치되었다. 본 연구에서는 HeMOSU-1호의 제원을 활용하여 수치 모의하였으며, 2013년 7월(설치 후 약 3년 경과) 현장 관측을 수행하였다.

2.2.2 모델 구성

본 연구에서는 왕복성 조류의 영향을 살펴보기 위해 2 case에 대하여 해석하였다. 먼저, Case 1은 1 m/s의 고정 유속을 가진 일방향 흐름에 대한 해석이며, Case 2는 -1~1 m/s의 유속분포를 가진 양방향 흐름에 대한 해석이다. 여기서 (-)부호는 방향을 의미한다. Fig. 3은 시간대별 유속 분포를 나타낸 것이다.

2.2.3 구조물 형상 및 격자

HeMOSU-1호 기상 타워 자켓 구조물 형상은 Fig. 4, 격자 정보는 Table 1과 같으며, 본 연구에서는 총 2,883,000 개의 직교 가변 격자체계를 구성하였다.

Table 1.

Grid information of jacket structure

X_min/X_max(m)	Y_min/Y_max(m)	Z_min/Z_max(m)	No. of x grid	No. of y grid	No. of z grid
−100/100	−40/40	−9/20	310	155	60

한편, 계산영역의 격자 형상은 Fig. 5와 같다.

2.3 계산 조건

계산영역의 경계 조건으로, Case 1의 경우, 유입부는 유속 조건을 주었으며 유출부는 outflow 조건을 적용하였다. 그리고 Case 2의 경우, 왕복성 흐름을 표현하기 위해 유입부와 유출부 조건을 유속 조건으로 설정하였다. 또한 2가지 경우 모두 상부는 자유수면을 표현하기 위해 pressure로 하였으며 하부는 지반 조건의 특성을 가진 wall 조건을 적용하였다. 양측면은 Symmetry 조건으로 대칭면으로 정의하여 대칭면에 수직한 방향의 에너지와 질량의 유출입이 없고 대칭면에 평행한 방향의 유동저항이 없는 경우로 조건을 설정하였다. 본 연구에서 케이스별 입력 조건을 다음 Table 2에 정리하였다.

Table 2.

Basic information of two scour simulation tests

Case	Structure type	Velocity	Direction	Analysis time
Case 1	Jacket	1 m/s	Unidirectional	10,000 sec
Case 2	−1~1 m/s	Bidirectional

FLOW-3D는 자유표면을 가진 유동장의 계산에서 정상상태 해석이 불가능하므로 비정상유동 난류해석을 수행하게 되는데 정지 상태의 조건은 조위를 설정하였다. 또한 유속의 초기 흐름은 난류상태의 비정상흐름이 되므로 본 해석에서는 정상상태의 해석 수행을 위해 1,000초의 유동 해석을 수행하였으며 그 후에 10,000초의 sediment scour 모델을 수행하였다. 해수의 밀도는 1,025 kg/m³의 점성유체로 설정하였으며 RNG(renormalized group) 난류 모델을 적용하였다.Go to :

3. 수치모형 실험 결과

3.1 Case 1

본 케이스에서는 1 m/s의 유속을 가진 흐름이 구조물 주변을 흐를 때, 발생하는 세굴에 대해서 수치 모의하였다. Fig. 6은 X-Z 평면의 유속 분포도이고 Fig. 7은 X-Y 평면의 유속 분포이다. 구조물 주변에서 약간의 유속 변화가 발생했지만 전체적으로 1 m/s의 정상 유동 상태를 띄고 있다.

이러한 흐름과 구조물과의 상호 작용에 의한 세굴 현상이 발생되며 Fig. 8에 구조물 주변 지형 변화를 나타내었다. 유속이 발생하는 구조물의 전면부는 대체로 침식이 일어나 해저지반이 초기 상태보다 낮아진 것을 확인할 수 있으며, 또한 전면부의 지반이 유실되어 구조물 후면부에 최대 0.13 m까지 퇴적된 것을 확인할 수 있다.

일방향 흐름인 Case 1의 경우에는 Fig. 9와 같이 10,000초 후 구조물 주변에 최대 1.32 m의 세굴이 발생하는 것으로 나타났다. 또한 구조물 뒤쪽으로는 퇴적이 일어났으며, 구조물 전면부에는 침식작용이 일어나고 있다.

3.2 Case 2

서해안은 조석현상으로 인해 왕복성 조류 흐름이 나타나고 있으며 대상해역은 -1~1 m/s의 유속분포를 가지고 있다. 본 연구에서는 이러한 특성을 고려한 왕복성 흐름에 대해서 수치모의하였다.다음 Fig. 10은 X-Z 평면의 유속 분포도이며 Fig. 11은 X-Y 평면의 유속 분포도이다.

양방향 흐름인 Case 2의 경우에는 Fig. 12와 같이 10,000초후 구조물 주변에 최대 1.44 m의 세굴이 발생하는 것으로 나타났다. 특히 구조물 내부에 조류 흐름 방향으로 침식 작용이 일어나고 있는 것으로 나타났다.

Fig. 13은 3차원 수치해석 모의 결과이다.

3.3 현장 관측

본 연구에서는 수치모의 실험의 검증을 위해 HeMOSU-1호 기상 타워를 대상으로 하여 2013년 7월 1일 수심 측량을 실시하였다.HeMOSU-1호 주변의 수심측량은 Knudsen sounder 1620과 미국 Trimble사의 DGPS를 이용하여 실시하였다. 매 작업시 Bar-Check를 실시하고, 수중 음파속도는 1,500 m/s로 결정하여 조위 보정을 통해 수심을 측량하였다. 측량선의 해상위치자료는 DGPS를 사용하여 UTM 좌표계로 변환을 실시하였다. 한편, 수심측량은 해면이 정온할 때 실시하였으며 관측 자료의 변동성을 제거하기 위해 2013년 7월 1일 10시~13시에 걸쳐 수심 측량한 자료를 동시간대에 국립해양조사원에서 제공한 위도 자료를 활용해 조위 보정하였다. 다음 Fig. 14는 위도 조위 관측소의 현장관측시간대 조위 시계열 그래프이다.

2013년 7월 1일 오전 10시부터 오후 1시에 걸쳐 수심측량한 결과를 이용하여 0.5 m 간격으로 등수심도를 작성하였으며 그 결과는 Fig. 15와 같다. 기상탑 내부 해역은 선박이 접근할 수 없기 때문에 측량을 실시하지 않고 Blanking 처리하였다.

대상 해역의 수심은 대부분 -15 m이나 4개의 Jacket 구조물 주변에서는 세굴이 발생하여 수심의 변화가 나타났다. 특히 L-3, L-4 주변에서 최대 1.5~2.0 m의 세굴이 발생한 것으로 보였으며, L-4 주변에서는 넓은 범위에 걸쳐 세굴이 발생하였다. 창조류는 북동, 낙조류는 남서 방향으로 흐르는 조류 방향성을 고려하였을 때, L-4 주변은 조류방향과 동일하게 세굴이 발생하고 있었으며, 보다 상세한 세굴형태는 원형 구조물 내부 방향의 세굴 심도를 측정하여 파악하여야 할 것으로 판단된다.관측결과 최대 1.5~2.0 m인 점을 고려하면 양방향 흐름을 대상으로 장기간에 걸쳐 모의실험을 진행하는 경우, 실제 현상에 더 근접하는 결과를 얻을 수 있을 것으로 사료된다.Go to :

4. 결론 및 토의

본 연구에서는 자켓구조물인 해상기상탑 HeMOSU-1 주변에서 발생하는 세굴현상을 검토하기 위하여 2013년 7월 1일 현장 관측을 수행하고, FLOW-3D를 이용하여 수치모의 실험을 수행하였다. 실험 조건으로는 먼저 1 m/s의 유속을 가진 일방향 흐름과 -1~1 m/s의 흐름 분포를 가진 왕복성 흐름에 대해서 수치모의를 수행하였다. 그 결과 일방향 흐름의 경우, 10,000 초에 이르렀을 때 1.32 m, 왕복성 흐름의 경우 동일 시간에서 1.44 m의 최대 세굴심도가 발생하였다. 동일한 구조물에 대해서 현장 관측 결과는 1.5~2.0 m로 관측되어 일방향 흐름보다 왕복성 흐름의 경우 실제 현상에 더 근사한 것으로 판단되었다. 이는 일방향 흐름의 경우, Fig. 8에서 보는 바와 같이 구조물 후면에 퇴적과 함께 해저입자의 맞물림이 견고해져 해저 지반의 저항력이 커지는 현상에 기인한 것으로 판단된다. 반면 양방향 흐름의 경우, 흐름의 변화로 인해 맞물림이 약해지고 이로 인해 지반의 저항력이 일방향 흐름보다 약해져 세굴이 더 크게 발생하는 것으로 판단되었다.또한 장시간에 걸쳐 모델링을 수행하는 경우, 보다 근사한 결과를 얻을 수 있을 것을 사료되며, 신형식 기초 구조물을 개발하여 세굴을 저감할 수 있는 지 여부를 판단하는 등의 추가 연구가 필요하다.Go to : International Electrotechnical Commission (IEC). (2009). IEC 61400-3: Wind turbines – Part 3: Design Requirements for Offshore Wind Turbines, Edition 1.0, IEC.

감사의 글

본 연구는 지식경제 기술혁신사업인 “승강식 해상플랫폼을 가진 수직 진자운동형 30kW급 파력발전기 개발(과제번호 :20133010071570)”와 첨단항만건설기술개발사업인 “해상풍력 지지구조 설계기준 및 콘크리트 지지구조물 기술 개발(과제번호:20120093)”의 일환으로 수행되었습니다.Go to :

References

American Bureau of Shipping (ABS). (2013). Guide for Building and Classing Bottom-Founded Offshore Wind turbine Installations.

API RP 2A WSD. (2005). Recommended Practice for Planning, Designing and Constructing Fixed Offshore Platforms-Working Stress Design, API.

Det Norske Veritas (DNV). (2010). OS-J101 Design of Offshore Wind Turbine Structures.

Federal Maritime and Hydrographic Agency (BSH). (2007). Standard. Design of Offshore Wind Turbines.

FLOW SCIENCE. (2014). FLOW-3D User’s Manual, Version 11.0.4.5.

International Electrotechnical Commission (IEC). (2009). IEC 61400-3: Wind turbines – Part 3: Design Requirements for Offshore Wind Turbines, Edition 1.0, IEC.

International Organization for Standardization (ISO). (2007). ISO 19902: Petroleum and Natural Gas Industries – Fixed Steel Offshore Structures.

Kim, YS, Kang, GO. (2011). Experimental Study on Hydraulic Resistance of Sea Ground Considering Tidal Current Flow, Journal of Korean Society of Coastal and Ocean Engineers. 23(1):118-125 (in Korean).

Kim, YS, Han, BD, Kang, GO. (2012). Effect of Incidence Angle of Current on the Hydraulic Resistance Capacity of Clayey Soil, Journal of Korean Society of Coastal and Ocean Engineers. 24(1):26-35 (in Korean).

KORDI. (2011). BSPN64710-2275-2. An Analysis on the Marine Characteristics and Design Supporting for Offshore Wind Power Plant (in Korean).

Ministry of Maritime Affairs and Fisheries. (2005). Harbor and fishery design criteria (in Korean).

Soulsby, R. (1997). Dynamics of marine sands. Thomas Telford Publications, London.

U.S. Army Corps of Engineers. (2006). Coastal Engineering Manual, Part II : Coastal Hydrodynamics, Chapter II–2, Meteorology and Wave Climate.

van Rijn, L. (1984). Sediment transport, Part II:bed load transport, Journal of Hydraulic Engineering, 110(10):1431-1456.

원문 바로 가기

FLOW-3D What’s New Ver.12.0

FLOW-3D_v12_새기능 다운로드

FLOW-3D v12는 그래픽 사용자 인터페이스 (GUI)의 설계 및 기능에서 매우 큰 변화를 이룬 제품으로 모델 설정을 단순화하고 사용자 워크 플로를 향상시킵니다. 최첨단 Immersed Boundary Method(침수경계 방법)은 FLOW-3D v12 솔루션의 정확성을 높여줍니다. 다른 주요 기능으로는 슬러지 침강 모델, 2-Fluid 2-Temperature 모델 및 Steady State Accelerator가 있으며,이를 통해 사용자는 자유 표면 흐름을 더욱 빠르게 모델링 할 수 있습니다.

Physical and Numerical Model

Immersed boundary method

힘과 에너지 손실에 대한 정확한 예측은 고체 주위의 흐름과 관련된 많은 엔지니어링 문제를 모델링하는 데 중요합니다. 새 릴리스 FLOW-3D v12에는 이러한 문제점 해결을 위해 설계된 새로운 고스트 셀 기반 Immersed Boundary Method (IBM)가 있습니다. IBM은 내 외부 흐름 해석을 위해, 벽 근처에서 보다 정확한 해를 제공하여 드래그 앤 리프트 힘의 계산을 향상시킵니다.힘과 에너지 손실의 정확한 예측은 고체 주위의 흐름을 포함하는 많은 공학적 문제를 모델링 하는데 중요합니다.

Two-field temperature for the two-fluid model

2 유체 열전달 모델은 각 유체에 대한 에너지 전달 방정식을 분리하기 위해 확장되었습니다. 각 유체는 이제 자체 온도 변수를 가지므로 인터페이스 근처의 열 및 물질 전달 솔루션의 정확도가 향상됩니다. 인터페이스에서의 열전달은 이제 시간의 표 함수가 될 수 있는 사용자 정의 열전달 계수에 의해 제어됩니다.

블로그 보기

Sludge settling model

새로운 슬러지 정착 모델은 수처리 애플리케이션에 부가되어 사용자들이 수 처리 탱크와 클래리퍼의 고형 폐기물 역학을 모델링 할 수 있게 해 줍니다. 침전 속도가 분산상의 액적 크기의 함수 인 드리프트-플럭스 모델과 달리, 침전 속도는 슬러지 농도의 함수이며 기능 및 표 형식으로 입력 할 수 있습니다.

개발노트 읽기

Steady-state accelerator for free surface flows

이름에서 알 수 있듯이 정상 상태 가속기는 정상 상태 솔루션에 대한 접근을 빠르게합니다.
이것은 작은 진폭 중력과 모세관 표면파를 감쇠시킴으로써 달성되며 자유 표면 흐름에만 적용 할 수 있습니다.

개발노트 읽기

Void particles

Void particles 가 기포 및 상 변화 모델에 추가되었습니다. Void particles는 붕괴 된 Void 영역을 나타내며, 항력 및 압력을 통해 유체와 상호 작용하는 작은 기포로 작용합니다. 주변 유체 압력에 따라 크기가 변하고 시뮬레이션이 끝날 때의 최종 위치는 공기 유입 가능성을 나타냅니다.

Sediment scour model

퇴적물 수송 및 침식 모델은 정확성과 안정성을 향상시키기 위해 정비되었습니다. 특히 퇴적물 종의 질량 보존이 크게 개선되었습니다.

개발 노트 읽기>

Outflow pressure boundary condition

고정 압력 경계 조건에는 압력 및 유체 분율을 제외한 모든 유량이 해당 경계의 상류의 유량 조건을 반영하는 ‘유출’옵션이 포함됩니다. 유출 압력 경계 조건은 고정 압력 및 연속 경계 조건의 하이브리드입니다.

Moving particle sources

시뮬레이션 중에 입자 소스를 이동할 수 있습니다. 시간에 따른 병진 및 회전 속도는 표 형식으로 정의됩니다. 입자 소스의 운동은 소스에서 방출 된 입자의 초기 속도에 추가됩니다.

Variable center of gravity

기변 무게중심은 중력 및 비관 성 기준 프레임 모델에서, 시간의 함수로서 무게 중심의 위치는 외부 파일에서 테이블로서 정의 될 수있다. 이 기능은 연료를 소비하고 분리 단계를 수행하는 로켓과 같은 모형을 모델링 할 때 유용합니다.

공기 유입 모델

가장 간단한 부피 기반 공기 유입 모델 옵션이 기존 질량 기반 모델로 대체되었습니다. 질량 기반 모델은 부피와 달리 주변 유체 압력에 따라 부피가 변화하는 동안 흡입된 공기량이 보존되기 때문에 물리학적 모델입니다.

Tracer diffusion

유동 표면에서 생성된 추적 물질은 분자 및 난류 확산 과정에 의해 확산될 수 있으며, 예를 들어 실제 오염 물질의 동작을 모방한다.

Model Setup

Simulation units

온도를 포함하여 단위 시스템은 완전히 정의해야하는데 표준 단위 시스템이 제공됩니다. 또한 사용자는 다양한 옵션 중에서 질량, 시간 및 길이 단위를 정의 할 수 있으므로 사용자 정의가 가능한 편리한 단위를 사용할 수 있습니다. 사용자는 압력이 게이지 또는 절대 단위로 정의되는지 여부도 지정해야합니다. 기본 시뮬레이션 단위는 기본 설정에서 설정할 수 있습니다. 단위를 완전히 정의하면 FLOW-3D 가 물리량의 기본값을 정의하고 범용 상수를 설정하여 사용자가 요구하는 작업량을 최소화 할 수 있습니다.

Shallow water model

Manning’s roughness in shallow water model

Manning의 거칠기 계수는 지형 표면의 전단 응력 평가를 위해 천수(shallow water) 모델에서 구현되었습니다. 표면 결함의 크기를 기반으로 기존 거칠기 모델을 보완하며 이 모델과 함께 사용할 수 있습니다. 표준 거칠기와 마찬가지로 매닝 계수는 구성 요소 또는 하위 구성 요소의 속성이거나 지형 래스터 데이터 세트에서 가져올 수 있습니다.

Mesh generation

하단 및 상단 경계 좌표의 정의만으로 수직 방향의 메시 설정이 단순화되었습니다.

Component transformations

사용자는 이제 여러 하위 구성 요소로 구성된 구성 요소에 회전, 변환 및 스케일링 변환을 적용하여 복잡한 형상 어셈블리 설정 프로세스를 단순화 할 수 있습니다. GMO (General Moving Object) 구성 요소의 경우, 이러한 변환을 구성 요소의 대칭 축과 정렬되도록 신체에 맞는 좌표계에 적용 할 수 있습니다.

Changing the number of threads at runtime

시뮬레이션 중에 솔버가 사용하는 스레드 수를 변경하는 기능이 런타임 옵션 대화 상자에 추가되어 사용 가능한 스레드를 추가하거나 다른 태스크에 자원이 필요한 경우 스레드 수를 줄일 수 있습니다.

Probe-controlled heat sources

활성 시뮬레이션 제어가 형상 구성 요소와 관련된 heat sources로 확장되었습니다. 히스토리 프로브로 열 방출을 제어 할 수 있습니다.

Time-dependent temperature at sources

질량 및 질량 / 운동량 소스의 유체 온도는 이제 테이블 입력을 사용하여 시간의 함수로 정의 할 수 있습니다.

Emissivity coefficients

공극으로의 복사 열 전달을위한 방사율 계수는 이제 사용자가 방사율과 스테판-볼츠만 상수를 지정하도록 요구하지 않고 직접 정의됩니다. 후자는 이제 단위 시스템을 기반으로 솔버에 의해 자동으로 설정됩니다.

Output

등속 필드 솔버 옵션을 사용할 때 유량 속도를 선택한 데이터 로 출력 할 수 있습니다 .
벽 접착력으로 인한 지오메트리 구성 요소의 토크 는 기존 벽 접착력의 출력 외에도 일반 이력 데이터에 별도의 수량으로 출력됩니다.
난류 모델 출력이 요청 될 때 난류 에너지 및 소산과 함께 전단 속도 및 y +가 선택된 데이터로 자동 출력됩니다 .
공기 유입 모델 출력에 몇 가지 수량이 추가되었습니다. 자유 표면을 포함하는 모든 셀에서 혼입 된 공기 및 빠져 나가는 공기의 체적 플럭스가 재시작 및 선택된 데이터로 출력되어 사용자에게 공기가 혼입 및 탈선되는 위치 및 시간에 대한 자세한 정보를 제공합니다. 전체 계산 영역 및 각 샘플링 볼륨 에 대해이 두 수량의 시간 및 공간 통합 등가물 이 일반 히스토리 로 출력됩니다.
솔버의 출력 파일 flsgrf 의 최종 크기 는 시뮬레이션이 끝날 때보 고됩니다.
2 유체 시뮬레이션의 경우, 기존의 출력 수량 유체 체류 시간 및 유체 가 이동 한 거리는 이제 유체 # 1 및 # 2와 유체의 혼합물에 대해 별도로 계산됩니다.
질량 입자의 경우 각 종의 총 부피와 질량이 계산되어 전체 계산 영역, 샘플링 볼륨 및 플럭스 표면에 대한 일반 히스토리 로 출력되어 입자 종 수에 대한 현재 출력을 보완합니다.
예를 들어 사용자가 가스 미순환을 식별하고 연료 탱크의 환기 시스템을 설계하는 데 도움이 되도록 마지막 국부적 가스 압력이 옵션 출력량으로 추가되었습니다. 이 양은 유체가 채워지기 전에 셀의 마지막 간극 압력을 기록하며, 단열 버블 모델과 함께 사용됩니다.

New Customizable Source Routines

사용자 정의 가능한 새로운 소스 루틴이 추가되었으며 사용자의 개발 환경에서 액세스 할 수 있습니다.

소스 루틴 이름	설명
cav_prod_cal	캐비 테이션 생산 및 확산 속도
sldg_uset	슬러지 정착 속도
phchg_mass_flux	증발 및 응축에 의한 질량 흐름
flhtccl	유체#1과#2사이의 열 전달 계수
dsize_cal	2상 유동에서의 동적 낙하 크기 모델의 충돌 및 이탈율
elstc_custom.	점탄성 유체에 대한 응력 방정식의 소스 용어

Brand New User Interface

FLOW-3D의 사용자 인터페이스가 완전히 재설계되어 사용자의 작업 흐름을 획기적으로 간소화하는 최신의 타일 구조를 제공합니다.

Dock widgets 설정

Physics, Fluids, Mesh 및 FAVOR ™를 포함한 모든 설정 작업이 형상 창 주위의 dock widgets으로 변환되어 모델 설정을 단일 탭으로 압축 할 수 있습니다. 이 전환을 통해 이전 버전의 복잡한 트리가 훨씬 깔끔하고 효율적인 메뉴 표시로 바뀌어 모델 설정 탭을 떠나지 않고도 모든 매개 변수에 쉽게 액세스 할 수 있습니다.

New Model Setup icons
With our new Model Setup design comes new icons, representing each step of the setup process.

New Physics icons
Our Physics icons are designed to be easily differentiated from one another at a glance, while providing clear visual representation of each model’s purpose and use.

RSS feed

새 RSS 피드부터 FLOW-3D v12.0 의 시뮬레이션 관리자 탭이 개선되었습니다 . FLOW-3D 를 시작하면 사용자에게 Flow Science의 최신 뉴스, 이벤트 및 블로그 게시물이 표시됩니다.

Configurable simulation monitor

시뮬레이션을 실행할 때 중요한 작업은 모니터링입니다. FLOW-3Dv12.0에서는 사용자가 시뮬레이션을 더 잘 모니터링할 수 있도록 Simulation Manager의 플로팅 기능이 향상되었습니다. 사용자는 시뮬레이션 런타임 그래프를 통해 모니터링할 사용 가능한 모든 일반 기록 데이터 변수를 선택하고 각 그래프에 여러 변수를 추가할 수 있습니다. 이제 런타임에서 사용할 수 있는 일반 기록 데이터는 다음과 같습니다.

최소/최대 유체 온도
프로브 위치의 온도
유동 표면 위치에서의 유량
시뮬레이션 진단(예:시간 단계, 안정성 한계)

*Runtime plots of the flow rate at the gates of the large dam* / *Large dam with flux surfaces at the gates*

Conforming mesh visualization

사용자는 이제 새로운 FAVOR ™ 독 위젯을 통해 적합한 메쉬 블록을 시각화 할 수 있습니다 .

Large raster and STL data

데이터를 처리하는 데 걸리는 시간으로 인해 큰 형상 데이터를 처리하는 것은 어려울 수 있습니다. 대형 지오메트리 데이터를 처리하는 데 여전히 상당한 시간이 소요될 수 있지만 FLOW-3D는 이제 이러한 대형 데이터 세트를 백그라운드 작업으로로드하여 사용자가 데이터를 처리하는 동안 완벽하게 응답하고 중단없는 인터페이스에서 계속 작업 할 수 있습니다.

FLOW-3D 제품소개

About FLOW-3D

FLOW-3D 개발 회사


Industry	Computational Fluid Dynamics Software
Founded	1980
Founder	Dr. C.W. “Tony” Hirt
Headquarters	Santa Fe, New Mexico, USA United States
Key people	Dr. Amir Isfahani, President & CEO
Products	FLOW-3D, FLOW-3D CAST, FLOW-3D AM, FLOW-3D CLOUD, FlowSight
Services	CFD consultation and services

FLOW-3D 개요

FLOW-3D는 미국 뉴멕시코주(New Mexico) 로스알라모스(Los Alamos)에 있는 Flow Scicence, Inc에서 개발한 범용 전산유체역학(Computational Fluid Dynamics) 프로그램입니다. 로스알라모스 국립연구소의 수치유체역학 연구실에서 F.Harlow, B. Nichols 및 T.Hirt 등에 의해 개발된 MAC(Marker and Cell) 방법과 SOLA-VOF 방식을 기초로 하여, Hirt 박사가 1980년에 Flow Science, Inc사를 설립하여 계속 프로그램을 발전시켰으며 1985년부터 FLOW-3D를 전세계에 배포하였습니다.

유체의 3차원 거동 해석을 수행하는데 사용되는 CFD모형은 몇몇 있으나, 유동해석에 적용할 물리모델 선정은 해석의 정밀도와 밀접한 관계가 있으므로, 해석하고자 하는 대상의 유동 특성을 분석하여 신중하게 결정하여야 합니다.

FLOW-3D는 자유표면(Free Surface) 해석에 있어서 매우 정확한 해석 결과를 제공합니다. 해석방법은 자유표면을 포함한 비정상 유동 상태를 기본으로 하며, 연속방정식, 3차원 운동량 보전방정식(Navier-Stokes eq.) 및 에너지 보존방정식 등을 적용할 수 있습니다.

FLOW-3D는 유한차분법을 사용하고 있으며, 유한요소법(FEM, Finite Element Method), 경계요소법(Boundary Element Method)등을 포함하여 자유표면을 포함하는 유동장 해석(Fluid Flow Analysis)에서 공기와 액체의 경계면을 정밀하게 표현 가능합니다.

유체의 난류 해석에 대해서는 혼합길이 모형, 난류 에너지 모형, RNG(Renormalized Group Theory) k-ε 모형, k-ω 모형, LES 모형 등 6개 모형을 적용할 수 있으며, 자유표면 해석을 위하여 VOF(Volume of Fluid) 방정식을 사용하고, 격자 생성시 사용자가 가장 쉽게 만들 수 있는 직각형상격자는 형상을 더욱 정확하게 표현하기 위해 FAVOR(Fractional Area Volume Obstacle Representation) 기법을 각 방정식에 적용하고 있습니다.

FLOW-3D는 비압축성(Incompressible Fluid Flow), 압축성 유체(Compressible Fluid Flow)의 유동현상 뿐만 아니라 고체와의 열전달 현상을 해석할 수 있으며, 비정상 상태의 해석을 기본으로 합니다.

FLOW-3D v12.0은 모델 설정을 간소화하고 사용자 워크 플로우를 개선하는 GUI(그래픽 사용자 인터페이스)의 설계 및 기능에 있어 중요한 변화를 가져왔습니다. 최첨단 Immersed Boundary Method는 FLOW-3Dv12.0솔루션의 정확도를 높여 줍니다. 다른 특징적인 주요 개발에는 슬러지 안착 모델, 2-유체 2-온도 모델, 사용자가 자유 표면 흐름을 훨씬 더 빠르게 모델링 할 수 있는 Steady State Accelerator등이 있습니다.

물리 및 수치 모델

Immersed Boundary Method

힘과 에너지 손실에 대한 정확한 예측은 솔리드 바디 주변의 흐름과 관련된 많은 엔지니어링 문제를 모델링하는 데 중요합니다. FLOW-3D v12.0의 릴리스에는 이러한 문제 해결을 위해 설계된 새로운 고스트 셀 기반 Immersed Boundary Method (IBM)가 포함되어 있습니다. IBM은 내부 및 외부 흐름을 위해 벽 근처 해석을 위해 보다 정확한 솔루션을 제공하여 드래그 앤 리프트 힘의 계산을 개선합니다.

Two-field temperature for the two-fluid model

2유체 열 전달 모델은 각 유체에 대한 에너지 전달 공식을 분리하도록 확장되었습니다. 이제 각 유체에는 고유한 온도 변수가 있어 인터페이스 근처의 열 및 물질 전달 솔루션의 정확도를 향상시킵니다. 인터페이스에서의 열 전달은 시간의 표 함수가 될 수 있는 사용자 정의 열 전달 계수에 의해 제어됩니다.

슬러지 침전 모델 / Sludge settling model

중요 추가 기능인 새로운 슬러지 침전 모델은 도시 수처리 시설물 응용 분야에 사용하면 수처리 탱크 및 정화기의 고형 폐기물 역학을 모델링 할 수 있습니다. 침전 속도가 확산된 위상의 방울 크기에 대한 함수인 드리프트-플럭스 모델과 달리, 침전 속도는 슬러지 농도의 함수이며 기능적인 형태와 표 형태로 모두 입력 할 수 있습니다.

Steady-state accelerator for free surface flows

이름이 암시하듯이, 정상 상태 가속기는 안정된 상태의 솔루션에 대한 접근을 가속화합니다. 이는 작은 진폭의 중력과 모세관 현상을 감쇠하여 이루어지며 자유 표면 흐름에만 적용됩니다.

Void particles

보이드 입자가 버블 및 위상 변경 모델에 추가되었습니다. 보이드 입자는 항력과 압력 힘을 통해 유체와 상호 작용하는 작은 기포의 역할을 하는 붕괴된 보이드 영역을 나타냅니다. 주변 유체 압력에 따라 크기가 변경되고 시뮬레이션이 끝난 후 최종 위치는 공기 침투 가능성을 나타냅니다.

Sediment scour model

침전물의 정확성과 안정성을 향상시키기 위해 침전물의 운반과 침식 모델을 정밀 조사하였다. 특히, 침전물 종에 대한 질량 보존이 크게 개선되었습니다.

Outflow pressure boundary condition

고정 압력 경계 조건에는 이제 압력 및 유체 비율을 제외한 모든 유량이 해당 경계의 상류에 있는 흐름 조건을 반영하는 ‘유출’ 옵션이 포함됩니다. 유출 압력 경계 조건은 고정 압력 및 연속성 경계 조건의 혼합입니다.

Moving particle sources

시뮬레이션 중에 입자 소스는 이동할 수 있습니다. 시간에 따른 변환 및 회전 속도는 표 형식으로 정의됩니다. 입자 소스의 운동은 소스에서 방출 된 입자의 초기 속도에 추가됩니다.

Variable center of gravity

중력 및 비 관성 기준 프레임 모델에서 시간 함수로서의 무게 중심의 위치는 외부 파일의 표로 정의할 수 있습니다. 이 기능은 연료를 소모하는 로켓을 모델링하고 단계를 분리할 때 유용합니다.

공기 유입 모델

Tracer diffusion / 트레이서 확산

유동 표면에서 생성된 추적 물질은 분자 및 난류 확산 과정에 의해 확산될 수 있으며, 예를 들어 실제 오염 물질의 거동을 모방합니다.

모델 설정

시뮬레이션 단위

이제 온도를 포함하여 단위계 시스템을 완전히 정의해야 합니다. 표준 단위 시스템이 제공됩니다. 또한 사용자는 선택한 옵션에서 질량, 시간 및 길이 단위를 정의하여 편리하며, 사용자 정의된 단위를 사용할 수 있습니다. 사용자는 또한 압력이 게이지 단위로 정의되는지 절대 단위로 정의되는지 여부를 지정해야 합니다. 기본 시뮬레이션 단위는 Preferences(기본 설정)에서 설정할 수 있습니다. 단위를 완벽하게 정의하면 FLOW-3D는 물리적 수량에 대한 기본 값을 정의하고 범용 상수를 설정할 수 있으므로 사용자가 필요로 하는 작업량을 최소화할 수 있습니다.

Shallow water model

천수(shallow water) 모델에서 매닝의 거칠기

Manning의 거칠기 계수는 지형 표면의 전단 응력 평가를 위해 천수(shallow water) 모델에서 구현되었습니다. 표면 결함의 크기를 기반으로 기존 거칠기 모델을 보완하며이 모델과 함께 사용할 수 있습니다. 표준 거칠기와 마찬가지로 매닝 계수는 구성 요소 또는 하위 구성 요소의 속성이거나 지형 래스터 데이터 세트에서 가져올 수 있습니다.

메시 생성

하단 및 상단 경계 좌표의 정의만으로 수직 방향의 메시 설정이 단순화되었습니다.

구성 요소 변환

런타임시 스레드 수 변경

프로브 제어 열원

활성 시뮬레이션 제어가 형상 구성 요소와 관련된 heat sources로 확장되었습니다. history probes로 열 방출을 제어 할 수 있습니다.

소스에서 시간에 따른 온도

질량 및 질량/모멘트 소스의 유체 온도는 이제 테이블 입력을 사용하여 시간의 함수로 정의 할 수 있습니다.

방사율 계수

Output

등속 필드 솔버 옵션을 사용할 때 유량 속도를 선택한 데이터로 출력 할 수 있습니다.
벽 접착력으로 인한 지오메트리 구성 요소의 토크는 기존 벽 접착력 출력과 함께 별도의 수량으로 일반 이력 데이터에 출력됩니다.
난류 모델 출력이 요청 될 때 난류 에너지 및 소산과 함께 전단 속도 및 y +가 선택된 데이터로 자동 출력됩니다.
공기 유입 모델 출력에 몇 가지 수량이 추가되었습니다. 자유 표면을 포함하는 모든 셀에서 혼입 된 공기 및 빠져 나가는 공기의 체적 플럭스가 재시작 및 선택된 데이터로 출력되어 사용자에게 공기가 혼입 및 탈선되는 위치 및 시간에 대한 자세한 정보를 제공합니다. 전체 계산 영역 및 각 샘플링 볼륨 에 대해이 두 수량의 시간 및 공간 통합 등가물이 일반 히스토리 로 출력됩니다.
솔버의 출력 파일 flsgrf 의 최종 크기는 시뮬레이션이 끝날 때 보고됩니다.
2 유체 시뮬레이션의 경우, 기존의 출력 수량 유체 체류 시간 및 유체 가 이동 한 거리는 이제 유체 # 1 및 # 2와 유체의 혼합물에 대해 별도로 계산됩니다.
질량 입자의 경우, 각 종의 총 부피 및 질량이 계산되어 전체 계산 영역, 샘플링 볼륨 및 플럭스 표면에 대한 일반 히스토리 로 출력되어 입자 종 수에 대한 현재 출력을 보완합니다.
최종 로컬 가스 압력 은 사용자가 가스 포획을 식별하고 연료 탱크의 배기 시스템 설계를 지원하는 데 도움이되는 선택적 출력량으로 추가되었습니다. 이 양은 유체로 채워지기 전에 셀의 마지막 공극 압력을 기록하며 단열 버블 모델과 함께 사용됩니다.

새로운 맞춤형 소스 루틴

새로운 사용자 정의 가능 소스 루틴이 추가되었으며 사용자의 개발 환경에서 액세스 할 수 있습니다.

소스 루틴 이름	기술
cav_prod_cal	Cavitation 생성과 소산 비율
sldg_uset	슬러지 침전 속도
phchg_mass_flux	증발 및 응축으로 인한 질량 플럭스
flhtccl	유체 # 1과 # 2 사이의 열전달 계수
dsize_cal	2 상 흐름에서 동적 액적 크기 모델의 응집 및 분해 속도
elstc_custom	점탄성 유체에 대한 응력 방정식의 Source Terms

새로운 사용자 인터페이스

FLOW-3D 사용자 인터페이스는 완전히 새롭게 디자인되어 현대적이고 평평한 구조로 사용자의 작업 흐름을 획기적으로 간소화합니다.

Setup dock widgets

Physics, Fluids, Mesh 및 FAVOR ™를 포함한 모든 설정 작업이 지오 메트리 윈도우 주변에서 독 위젯으로 변환되어 모델 설정을 단일 탭으로 요약할 수 있습니다. 이러한 전환으로 인해 이전 버전의 복잡한 접이식 트리가 훨씬 깨끗하고 효율적인 메뉴 프레젠테이션으로 대체되어 사용자는 ModelSetup탭을 떠나지 않고도 모든 매개 변수에 쉽게 액세스 할 수 있습니다.

New Model Setup icons

새로운 모델 설정 디자인에는 설정 프로세스의 각 단계를 나타내는 새로운 아이콘이 있습니다.

New Physics icons

RSS feed

새 RSS 피드부터 FLOW-3D v12.0의 시뮬레이션 관리자 탭이 개선되었습니다. FLOW-3D 를 시작하면 사용자에게 Flow Science의 최신 뉴스, 이벤트 및 블로그 게시물이 표시됩니다.

Configurable simulation monitor

시뮬레이션을 실행할 때 중요한 작업은 모니터링입니다. FLOW-3Dv1.0에서는 사용자가 시뮬레이션을 더 잘 모니터링할 수 있도록 SimulationManager의 플로팅 기능이 향상되었습니다. 사용자는 시뮬레이션 런타임 그래프를 통해 모니터링할 사용 가능한 모든 일반 기록 데이터 변수를 선택하고 각 그래프에 여러 변수를 추가할 수 있습니다. 이제 런타임에서 사용할 수 있는 일반 기록 데이터는 다음과 같습니다.

최소/최대 유체 온도
프로브 위치의 온도
유동 표면 위치에서의 유량
시뮬레이션 진단(예:시간 단계, 안정성 한계)

Runtime plots of the flow rate at the gates of the large dam

Conforming 메쉬 시각화

용자는 이제 새로운 FAVOR ™ 독 위젯을 통해 적합한 메쉬 블록을 시각화 할 수 있습니다.

Large raster and STL data

데이터를 처리하는 데 걸리는 시간 때문에 큰 지오 메트리 데이터를 처리하는 것은 수고스러울 수 있습니다. 대형 지오 메트리 데이터를 처리하는 데는 여전히 상당한 시간이 걸릴 수 있지만, FLOW-3D는 이제 이러한 대규모 데이터 세트를 백그라운드 작업으로 로드하여 사용자가 데이터를 처리하는 동안 완전히 응답하고 중단 없는 인터페이스에서 작업을 계속할 수 있습니다

퇴적, 세굴(쇄굴) / Sediment Scour

퇴적-세굴(쇄굴) / Sediment Scour

유체 역학과 완벽하게 연계된 FLOW-3D 의 sediment scour model은 침전물 수송, 부유물 운반, 인입 및 퇴적을 포함하여 비 점착성 토양의 모든 퇴적물 이동 과정을 모의 실험합니다 (Wei 등, 2014). 입자 크기, 질량 밀도 및 임계 전단 응력과 같은 다른 성질을 갖는 다중 퇴적물 종을 허용합니다. 예를 들어, 중간 모래, 거친 모래 및 자갈은 시뮬레이션에서 세 가지 종으로 분류 할 수 있습니다. 이 모델은 3D 흐름과 2D 천수(shallow water) 흐름에 모두 적용됩니다.

모델에서, 퇴적물의 충진 층은 퇴적물 종의 상이한 조합을 갖는 다수의 하위 구성 요소로 구성 될 수있는 하나의 기하학적 구성 요소에 의해 정의됩니다. 충전된 베드는 면적 및 부피 분율을 사용하는 FAVOR^TM 기술에 의해 기술된다. 베드 인터페이스를 포함하는 메쉬 셀에서 인터페이스의 위치, 방향 및 면적이 계산되어 베드 전단 응력, 임계 실드 매개 변수, 침식 속도 및 베드로의 전송 속도를 결정합니다. 3 차원 난류 유동에서의 전단 응력은 매체 입자 크기 50 에 비례하는 층 표면 거칠기를 고려한 표준 벽 함수를 사용하여 평가됩니다. 2D 천수(shallow water)의 경우, 층 전단 응력 계산은 항력 계수가 사용자 정의이거나 수심과 층 표면 거칠기를 사용하여 국부적으로 계산 된 2 차 법칙을 따릅니다.

그림 1. t = 8 분에서의 유량
이 모델은 Meyer-Peter와 Muller (1948)의 방정식을 사용하여 베드 인터페이스를 포함하는 각 메쉬 셀에서의 베드로드 이송을 계산합니다. 서브 메쉬 (submesh) 방법은 메쉬 셀에서 이웃에있는 각 메쉬 셀로 이동하는 입자의 양을 결정하는 데 사용됩니다. 부유 퇴적물 농도는 퇴적물 수송 방정식을 풀음으로써 얻어집니다. 침식의 계산은 침전물 유입 및 침전을 동시에 고려합니다. entrainment에서 입자의 리프팅 속도는 Winterwerp et al. (1992). 퇴적시의 침강 속도는 3D 유동에 대한 퇴적물의 표류 속도와 같지만 얕은 수류에 대해서는 현존 방정식을 사용하여 계산됩니다 (Soulsby, 1997). 드리프트 플럭스 이론 (Breitour and Hirt, 2009)은 입자의 드리프트 속도를 계산하는 데 사용됩니다.

그림 2. t = 8 분의 구멍 채취
이 페이지의 예는 3 개의 원통형 교각을 중심으로 한, 맑은 물 정화에 대한 시뮬레이션입니다. 교각의 지름은 1.5m이며, 교각은 2m 간격으로 나란히 배치되어 있습니다. 다가오는 유량은 실린더와 정렬되며 2m/s의 속도를가집니다. 베드 재료는 모래 (직경 5mm), 자갈 (10mm) 및 거친 자갈 (20mm) 인 세 가지 퇴적물 종으로 구성됩니다. 그림 1, 2 및 3은 8 분간 실린더 주변의 흐름, 채취 구멍 및 채취 깊이 분포를 보여줍니다.

그림 3. t = 8min에서의 정련 깊이 (양수 값) 및 침전 높이 (양수 값)
이 모델에 대한 더 자세한 정보는 침전물 퇴적에 관한 Flow Science Report를 다운로드하십시오.

The Sedimentation Scour Model [침전 세굴(쇄굴) 모델]

1. Introduction
The three-dimensional sediment scour model for non-cohesive soils was first introduced to FLOW-3D in Version 8.0 to simulate sediment erosion and deposition (Brethour, 2003). It was coupled with the three-dimensional fluid dynamics and considered entrainment, drifting and settling of sediment grains. In Version 9.4 the model was improved by introducing bedload transport and multiple sediment species (Brethour and Burnham, 2010). Although applications were successfully simulated, a major limitation of the model was the approximate treatment of the interface between the packed and suspended sediments. The packed bed was represented by scalars rather than FAVORTM (Fractional Area Volume Obstacle Representation, the standard treatment for solid components in FLOW-3D). As a result, limited information about the packed bed interface was available. That made accurate calculation of bed shear stress, a critical factor determining the model accuracy, challenging.

In this work, the 3D sediment scour model is mostly redeveloped and rewritten. The model is still fully coupled with fluid flow, allows multiple non-cohesive species and considers entrainment, deposition, bedload transport and suspended load transport. The fundamental difference from the old model is that the packed bed is described by the FAVORTM technique. At each time step, area and volume fractions describing the packed sediments are calculated throughout the domain. In the mesh cells at the bed interface, the location, orientation and area of the interface are calculated and used to determine the bed shear stress, the critical Shields parameter, the erosion rate and the bedload transport rate. Bed shear stress is evaluated using the standard wall function with consideration of bed surface roughness that is related to the median grain size d50. A sub-mesh method is developed and implemented to calculate bedload transport. Computation of erosion considers entrainment and deposition simultaneously in addition to bedload transport.

Furthermore, a shallow-water sediment scour model is developed in this work by adapting the new 3D model. It is coupled with the 2D shallow water flows to calculate depth-averaged properties for both suspended and packed sediments. Its main differences from the 3D model are 1) the settling velocity of grains is calculated using an existing equation instead of the drift-flux approach in the 3D model, and 2) turbulent bed shear stress is calculated using a well-accepted quadratic law rather than the log wall function. The drag coefficient for the bed shear stress is either user-given or locally evaluated using the water depth and the bed surface roughness that is proportional to d50 of the bed material. The following sections present the sediment theory used in the model and application and validation cases.

Sediment Scour [침전 / 세굴(쇄굴)]

Introduction
The sediment scour model predicts the behavior of packed and suspended sediment within the three-dimensional flow capabilities of FLOW-3D®. Potential applications include erosion around bridge piers, weirs, dams and underwater pipelines, and removal and drifting of sand or snow over terrain. The model consists of two basic components: drifting and lifting. Drifting acts on sediment that is suspended in the flow; gravity (along with other body forces) causes the settling of the sediment. This model is based on the drift-flux model already incorporated into FLOW-3D®. Lifting takes place only at the interface between the packed sediment and fluid and occurs where the local shear stress imposed by the liquid on the bed interface exceeds a critical value. The amount of lifting is proportional to the shear stress. In conjunction with the drifting and lifting models, a drag model is used to mimic the solid-like behavior of the sediment in regions where its concentration exceeds a cohesive solid fraction. The viscosity and density are functions of the sediment concentration; they are calculated as a function of the sediment concentration.

Water Rivers Bridge Piers

Bridge Piers

FLOW-3D의 Sediment Scour Model 은 강이나 하천에서 수리학적으로 복잡한 교각과 지형에 따라 여러가지 퇴적물들의 높이 변화를 해석할 수 있습니다. 세굴 모델은 FLOW-3D 난류 모델들로 적분하여 3차원 분석이 가능합니다. FLOW-3D의 Shallow Water Model로 더 넓은 범위의 세굴 분석이 가능합니다.

교각 주위의 세굴 해석

세굴 모델은 deposition, packing, entrainment and drift-flux 메커니즘으로 되어있습니다. FLOW-3D v11 에서는 퇴적층의 형상을 FAVOR 하여 좀 더 정확하게 bed net 높이 변화를 시각화 할 수 있습니다. 시공간적으로 침전물의 변화뿐 아니라 유체의 부유물들, 바닥/유체 계면에서의 전단응력들을 분석할 수 있습니다.

Modeling Hydraulic Control Structures

In addition to the flow rates and detail of hydraulic behaviors associated with the control gate structures and powerhouse operation, FLOW-3D‘s sediment and scour model allows users to identify regions of high scour both near the control structure and further downstream in the vicinity of the bridge piers.

Bridge Pier Simulations

The first video shows a FLOW-3D simulation of the erosion that occurs around a group of three 2.4 m diameter piers as river water flows past at 1.5 m/s. The river depth is 15.8 m and the mean sediment size was presumed to be 0.35 mm.

Tag: Sediment Scour Mode

Abstract

1. Introduction

2. Numerical Method

2.1. Governing Equations of Flow

2.2. Turbulent Model

2.3. Sediment Scour Model

2.3.1. Entrainment and Deposition

2.3.2. Bed Load Transport

2.3.3. Suspended Load Transport

3. Model Setup

3.1. Mesh Geometric Dimensions

3.2. Boundary Conditions

3.3. Wave Parameters

3.4. Mesh Sensitivity

3.5. Model Validation

4. Numerical Results and Discussions

4.1. Scour Evolution

4.2. Scour Mechanism under Random Waves

4.3. Equilibrium Scour Depth

4.4. Parametric Study

4.4.1. Influence of Froude Number

4.4.2. Influence of Euler Number

5. Conclusions

Author Contributions

Funding

Institutional Review Board Statement

Informed Consent Statement

Data Availability Statement

Conflicts of Interest

References

Share and Cite

For more information on the journal statistics, click here.

범람으로 인한 비점착성 흙댐 붕괴에 대한 테일워터 깊이의 영향

Abstract

Keywords

Notation

1. Introduction

2. Material and methods

2.1. Experimental procedures

2.2. Repeatability

3. Numerical model

4. Results of experimental work

5. Discussion

5.1. Side erosion

5.2. Upstream and downstream water depths

5.3. Eroded volume

5.4. The outflow discharge

6. Conclusions

Declaration of Competing Interest

Reference

Cited by (0)

공간적 제방댐 붕괴를 통한 최대 유출량 조사

Abstract

Keywords

1. Introduction

2. Numerical simulation

2.1. Geometric presentations

2.2. Governing equations

2.3. Boundary and initial conditions

2.4. Numerical method

2.5. Turbulent models

2.6. Sediment scour model

2.7. Grid type

2.8. Time step

2.9. Numerical model validation

3. Analysis and discussions

3.1. Dam breaching process evolution

3.2. The effect of initial breach shape

3.3. The effect of initial breach dimensions

3.4. The effect of initial breach location

3.5. The effect of upstream and downstream dam slopes

4. Conclusions

References

Analysis of Scour Phenomenon around Offshore Wind Foundation using Flow-3D Mode

Abstract

서론

결론

References

Numerical Simulation Test of Scour around Offshore Jacket Structure using FLOW-3D