The 1st International Conference on Water Resources and Coastal Engineering

NUMERICAL SIMULATION OF WAVE LOADS ON A SUBMERGED

RECTANGULAR BRIDGE DECK
Em Le Huynh1)*, Chia-Ren Chu 1)
1)
Department of Civil Engineering, National Central University, Taiwan, ROC
E-mail: [email protected], [email protected]

Abstract - Cross-river structures such as bridges in the experiments and demonstrated that the drag coefficient of
coastal areas could be partially or entirely submerged in the rectangular cylinder, due to the blockage effect, is
the water during storm surges and attacked by the currents larger than that in an unbounded domain. The drag
and waves. The failure of cross-river structures usually coefficient of the cylinder in trans-critical flows is greater
occurs in the event of extreme floods and storm surges than that in the subcritical flows under the same blockage
which deviate tremendously from the steady assumption. ratio, owing to the wave-induced drag. Furthermore, the
There have been several incidents around the world where separated shear flow on the upper side of the cylinder is
bridge decks have been fully submerged in torrential constrained by the water surface when the cylinder is
flows. For example, Hurricane Katrina damaged 45 close to the water surface, and the resulting asymmetric
bridges in the gulf area of the U.S. in 2005. Nonetheless, pressure distribution generates a downward force on the
the bridge engineers usually neglect the hydrodynamic cylinder, and the lift coefficient is a function of the
loading on the bridge structures, due to the difficulty to Froude number and the submergence ratio.
predict the dynamic process of turbulent flows and wave The interaction between submerged decks and the
loads on the cross-river structures. This study integrates a wave loads is an interesting topic of a number of scientific
two-dimensional Large Eddy Simulation (LES) model and engineering problems. Wave loads on submerged
and the Volume of Fluid (VOF) method to investigate the decks is essential to the design and analysis of tsunami
flow fields around a submerged rectangular bridge deck. and storm wave loads on coastal bridge decks. Water
The validated numerical model then was utilized to study height may rise to a higher elevation due to the effect of
the effects of submergence, deck thickness (blockage storm surge during flooding events. Bridge decks can be
effect) on the dimensionless force coefficient of the partially or entirely submerged in the river flows when the
submerged deck. The simulation results indicated that the design discharge is exceeded. There have been several
wave height downstream of the bridge deck was smaller incidents around the world where bridge decks have been
than that of upstream of the deck. Also, the blockage ratio entirely submerged in torrential flows. When bridge decks
did not significantly affect the simulated wave heights are submerged, the hydrodynamic loading on these
downstream of the bridge . bridges is an essential parameter for evaluating their
safety. For example, coastal bridges were entirely
submerged under such extreme storm surges or tsunamis,
Key words - Large Eddy Simulation; Wave loads; Bridge
as they were reported during the 2004 Sumatra earthquake
deck; Volume of Fluid; Blockage ratio; Submergence
and the subsequent tsunami in the Indian Ocean. In
ratio.
addition, Hurricane Katrina damaged 45 bridges in the
Gulf area of the U.S. in 2005, and 2011 Tohoku
1. Introduction earthquake and tsunami destroyed many coastal structures
Catalano et al. [2] studied and compared the in Japan.
simulation results of a Large Eddy Simulation (LES) The effect of wave loads on submerged decks has
model and a Reynolds-averaged Navier–Stokes (RANS) been studied and validated by a large number of
for flows around a circular cylinder without water surface theoretical and experimental results. There are some
under high Reynolds numbers (Re = 5 × 105 – 2 × 106). previous studies on the responses of wave forces on
The Reynolds number is computed by using the relation submerged decks. The results became completely
between the diameter of the cylinder and the free stream valuable by their application in the offshore industry.
velocity. Their results concluded that the drag coefficient Linear solutions and relations, similar to that suggested by
determined by the LES model is better than that by the Morison et al. (1950), were applied to approximate the
RANS, as compared to the experimental laboratory loads. Brater et al. (1958), Herbich and Shank (1970), and
results. Besides, the pressure drag is much larger than the Durgin and Shiau (1975), and more lately Rey and
skin friction drag on the cylinder surface. Touboul (2011), did experimental results on the
Chu et al. [8] studied the hydrodynamic forces on a interaction between sinusoidal waves and submerged
fully submerged rectangular cylinder by employing a horizontal decks in the intermediate and deep water.
three-dimensional Large Eddy Simulation (LES) model. Hayatdavoodi et al. [1] used the Green–Naghdi (GN)
Their simulation results were validated by flume equation to compute the non-linear wave load on a
 Em Le Huynh, Chia-Ren Chu 2

submerged bridge deck in shallow waters. They assumed the quantity is the spatially filtered value [3,4,5]. The
the bridge deck is a thin horizontal plate and blockage effective viscosity, μeff, is defined as:
ratio (the deck thickness over the average water depth) of eff    SGS (3)
the deck did not affect the dimensionless force
coefficients of the bridge deck. where μ is the dynamic viscosity of the water; and μSGS is
the viscosity of the sub-grid scale turbulence [10]. The
sub-grid scale turbulence was modeled by the
2. Numerical Model Smagorinsky [9] model:
This study utilized the same Large Eddy Simulation
SGS    CS  S 
2
(LES) model used by Chu et al. [8] to compute the wake 2S ij S ij (4)
flow of a circular cylinder underneath the water surface. where Cs is the Smagorinsky coefficient; Sij is the filtered
The governing equations are the filtered continuity rate of strain tensor:
equation and Navier–Stokes equations:
1  u u j 
ui S ij   i   (5)
0 (1) 2  x j xi 
xi 

 and s is the characteristic length of the spatial filter and

  ui  ui u j P  
 ui u j
    g i 3   eff    is calculated as:
t x j xi x j   x j xi

 
S   xyz 
1/3
(6)
(2)
where x, y, z are the three components of the grid
where u is the fluid velocity; subscripts i, j =1, 2 and 3 lengths. In this study, the value of the Smagorinsky
represent the x, y and z directions, respectively; t is the coefficient was set as Cs = 0.30 for two-dimensional
time; P is the pressure; ρ is the density of the fluid; g is flows. The computational domain was discretized by the
the gravitational acceleration; and the over-bar represents Finite Volume Method (FVM).The forward difference
 The 1st International Conference on Water Resources and Coastal Engineering
scheme was used to calculate the time derivative term. became smaller when the deck came close to the water
The Poisson Pressure Equation (PPE) was solved by the surface. The value of the wave height decreased as the
projection method [11]. The convergence criterion for the submergence ratio decreased.
momentum equations was set as 10-6.
The kinematics of the water surface were solved by
the Volume of Fluid (VOF) method [7]. The volume
fraction fm occupied by the water in a grid cell can be
described by:
f m 

t x j

fm u j  0  (7)

The computational cell is full of water when the

volume fraction fm=1; and the cell partially occupied by
water when 0 ≤ fm ≤ 1. The wall function, suggested by
Cabot and Moin [6], was used to calculate the effective
viscosity near the channel bed:
SGS
 
 2
  z  1  e Z w /A
(8)

Fig. 3: Water surface variation for the case without deck
 zu* and deck with S*=1 and S*=2 of the deck with D/H=0.1
z  (9)

where z is the vertical distance from the cell center to the
wall; κ (= 0.41) is the von Karman constant; u∗ is the
shear velocity; and the parameter A = 19. Further details
of the numerical model can be found in Chu et al. [8].
The deck surface and channel floor were set as a no-
slip boundary condition; the lateral boundaries were set as
a free-slip boundary condition. The pressure above the
water surface was set as zero. The computational domain
was discretized by the finite-volume method (FVM). The
grid arrangement was shown in Fig. 2. The entire
computational domain was divided into three zones. Zone
II (-10 < x/D < 15), where the deck was located, adopted a
uniform grid with the smallest grid size. The grid size in
(a)
Zones I and III was twice the smallest grid size. The
Courant number was set as Cr = 0.85, and the average
time step was t = 5 × 10-3 s.

3. Results and discussion

3.1. Submergence Effect
The submergence ratio is defined as:
S
S*  (10)
D
where S is the distance between averaged water surface
and the deck; D is thickness of the bridge deck.
The blockage ratio is defined as:
D
Br  (11) (b)
H Fig. 4: Simulated surface displacement in the case S*= 1 at
where H is the average water depth. the upstream and downstream of the deck with (a) D/H=0.2 and
(b) D/H=0.4
Fig. 3 shows the water surface variation at the location
(-20 < x/L < 80) for the case without deck and the deck
with the submergence ratio S*= 1 and S*= 2, respectively.
As the wave went over the deck, the wave surface was
changed, became unsteady. The submerged horizontal
deck can also be used as wave breakers. The wave height
 Em Le Huynh, Chia-Ren Chu 4

Fig. 5: Simulated velocity vectors on the central plane (y/W Fig. 6: Simulated velocity vectors on the central plane (y/W
= 0.5) of the deck with S*=1 . (a) t = To + T/4; (b)t = To + T/2; = 0.5) of the deck with S*=2 . (a) t = To + T/4; (b)t = To + T/2;
(c) t = To + 3T/4; (d) t = To + T. (c) t = To + 3T/4; (d) t = To + T.
 The 1st International Conference on Water Resources and Coastal Engineering
Fig. 4 shows the simulated surface displacement in the height downstream of the bridge deck is smaller than that
case S* = 1 at the upstream and downstream of the deck of upstream. Furthermore, the effect of the blockage ratio
with D/H = 0.2 and D/H = 0.4. The wave surface was on the simulated wave height is insignificant. The future
simulated in 5 seconds. The simulation results indicated study will focus on pressure and wave load of the
that the wave height upstream (x/L = -5) of the bridge submerged deck, which are important parameters for the
deck is larger than that of downstream (x/L = 10) of the design of coastal bridges.
deck.
Figs. 5 and 6 show the simulated velocity vectors on
the central plane (y/W = 0.50) of the tank S*= 1 and S*=
2, respectively, at four different phases of one cycle. The
period T = 0.5 s, and the reference time To = 15 s. The
interaction between the deck and the vortices became
more complicated as the submergence ratio increased.
Furthermore, the wave height and wave-induced velocity
in the case S*= 1 was larger than that of S* = 2. The
vortices originated by the deck became weaker with
increasing the submergence ratio S*.
3.2. Blockage effect
Fig. 7 compares the water surface variation for the
case with the deck and without the deck, under the
Fig. 8: Simulated surface displacement in the case Br=0.2,
blockage ratio Br = 0.20, 0.30 and 0.40 at the location (-
Br=0.3 and Br=0.4 at the upstream and downstream of the deck.
20 < x/L < 80). As the waves went over the deck, the
wave surface was changed, became more complicated.
Fig. 8 shows the simulated surface displacement in the References
case of the deck with the blockage ratio Br = 0.20, 0.30 [1] Hayatdavoodi, M., Ertekin, R.C., 2015c. Wave forces on a
and 0.40 at the location (x/L = 10). The simulation results submerged horizontal plate. Part Ii: solitary and cnoidal waves. J.
Fluids and Structures 54, 580–596. http://dx.doi.org/
indicated that the blockage ratio did not significantly
10.1016/j.jfluidstructs.2014.12.009.
affect the wave heights. [2] Catalano P, Wang M, Iaccarino G, Moin P. Numerical simulation
of the flow around a circular cylinder at high Reynolds numbers.
Int J Heat Fluid Flow 2003;24(4):463–9
[3] Deardorff, JW. A numerical study of three dimensional turbulent
channel flow at large Reynolds numbers. J Fluid Mech
1970;41:453–80.
[4] O’Neil J, Meneveau, C. Subgrid-scale stresses and their
modelling in a turbulent plane wake. J Fluid Mech
1997;349:253–93.
[5] Pope SB. Turbulent flows. Cambridge, U.K.: Cambridge
University Press; 2000
[6] DeLong M. Two examples of the impact of partitioning with
Chaco and Metis on the convergence of additive-Schwarz
preconditioned FGMRES, New Mexico, U.S.A: Los Alamos
National Laboratory; 1997. Technical report LA-UR-97-4181
[7] Hirt CW, Nichols BD. Volume of fluid (VOF) method for the
dynamics of free boundaries. J Comput Phys 1981;39(1):201–25
[8] Chu C-R, Chung C-H, Wu T-R, Wang C-Y. Numerical analysis
of free surface flow over a submerged rectangular bridge deck. J
Fig. 7: Water surface variation in the case without deck and Hydraulic Eng. 2016; 142(12):1–11. doi:10.1061/(ASCE)HY.
deck with Br=0.2, Br=0.3, Br=0.4 1943-7900.0001177.
[9] Smagorinsky J. General circulation experiments with the
primitive equations: I. The basic experiment. Mon Weather Rev
4. Conclusion 1963;91:99–164
[10] Wu T-R, Chu C-R, Huang C-J, Wang C-Y, Chien S-Y, Chen M-
This study used a Large Eddy Simulation (LES) Z. A two-way coupled simulation of moving solids in free-
model and the Volume of Fluid (VOF) method to surface flows. Comput Fluids 2014;100:347–55.
investigate the wave over a submerged rectangular bridge [11] Cabot W, Moin P. Approximate wall boundary conditions in the
deck. The effects of submergence ratio and blockage ratio large eddy simulation of high Reynolds number flow. Flow
Turbul Combust 2000;63:269–91
of the deck were studied systematically by the numerical
model. The simulation results indicated that the wave

(Receive on: …/…/2019, Review on: …/…/2019)

Em Le Huynh, Chia-Ren Chu 6

Author information

Full name of author 1: Em Le Huynh

- Department of Civil Engineering, Ho Chi Minh University of Technology, Master Degree.
- Ph. D student in Department of Civil Engineering, National Central University
- Research topics: Hydraulic engineering, wind engineering
- Phone number: +886 910 824 414

Full name of author 2: Chia-Ren Chu

- Civil and Environmental Engineering, Cornell University, Ph.D.
- Professor in Department of Civil Engineering, National Central University
- Research topics: Fluid mechanics, hydraulic engineering
- Phone number: 886 3 4227151 #34138