Ocean Engineering 158 (2018) 78–98

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Numerical modeling of non-breaking, impulsive breaking, and broken wave

interaction with elevated coastal structures: Laboratory validation and
inter-model comparisons
Hyoungsu Park a, *, Trung Do b, Tori Tomiczek c, Daniel T. Cox a, John W. van de Lindt b
a
School of Civil and Construction Engineering, 101 Kearney Hall, Oregon State University, Corvallis, OR 97331, United States
b
Department of Civil and Environmental Engineering, Colorado State University, Fort Collins, CO 80523, United States
c
Department of Naval Architecture and Ocean Engineering United States Naval Academy, Annapolis, MD 21402, United States

A R T I C L E I N F O A B S T R A C T

Keywords: Quantitative CFD model validation and inter-model comparisons between IHFOAM and ANSYS-FLUENT were
Experiment performed for pressures and forces on an elevated structure using a 1:10 physical model. Non-breaking, impulsive
IHFOAM breaking, and broken wave conditions at the structure's location were simulated in IHFOAM and FLUENT. The
ANSYS FLUENT calculated time series of water surface elevation and horizontal and vertical pressures and forces were compared
Validation with the measured data. We introduced the impulse of residual to quantify the variation of the force and pressure
Force
time series. Results indicated that the numerical models performed differently depending on the wave conditions,
Pressure
even for the same initial set up. Non-breaking wave simulations showed the best agreement with experimental
data for both models, while broken wave trials showed the largest deviations. Bottom pressures and vertical forces
were less sensitive to wave breaking conditions. Results indicate that future benchmarking tests for an elevated
structure must consider both horizontal and vertical forces due to various wave breaking conditions. The accuracy
of simulated wave shoaling and breaking processes played a key role in precisely calculating the forces and
pressures on the structure, and it was difficult for the CFD models to simulate the exact wave breaking conditions
as the measurements.

1. Introduction were built with outdated codes and standards, especially as coastal re-
gions are threatened by increasing storm intensity and global sea level
Hurricanes and typhoons generate elevated surge levels and strong rise (e.g. Mori et al., 2013). Retrofitting a structure is one option to
waves that can cause extensive damage to buildings and other coastal mitigate damage during future storm events and thus increase the resil-
infrastructure, especially those located in low-lying coastal regions. The ience of coastal communities. However, to effectively mitigate damage,
history of recorded damage on buildings near the shoreline from past these techniques require precise predictions of the wave climate and the
storms indicates that the intensity of storms and resulting damage has corresponding wave loads under various storm wave conditions.
increased over the past 30 years (Emanuel, 2005). For example, the Estimation of wave forces on elevated structures is available through
United States has been impacted by recent events such as Hurricanes analytical solutions or empirical solutions based on physical experi-
Katrina (2005), Ike (2008), and Sandy (2012). The 2017 Atlantic hurri- mental results. Previous studies have mostly focused on coastal infra-
cane season was one of the most active and costliest seasons in recorded structure such as bridges and jetties rather than residential buildings.
history. To withstand the high surge levels and waves induced from Kaplan (1992) and Kaplan et al. (1995) predicted the time history of
hurricanes and typhoons, structures are commonly elevated above grade; impact loadings on offshore platforms and the wave impact force from
this structural design is a common building type in regions of low large incident waves based on momentum flux. Cuomo et al. (2007)
elevation such as barrier islands on the East Coast and Gulf Coast of the conducted a 1:25 scale model test of wave forces on exposed jetties and
United States. However, the magnitude of damage is exacerbated in re- developed new dimensionless predictive solutions. Cuomo et al. (2009)
gions that are characterized by aging infrastructure and buildings that also performed large scale (1:10) experiments on coastal highway bridges

* Corresponding author.
E-mail address: [email protected] (H. Park).

https://doi.org/10.1016/j.oceaneng.2018.03.088
Received 1 September 2017; Received in revised form 22 January 2018; Accepted 31 March 2018
Available online 4 April 2018
0029-8018/© 2018 Elsevier Ltd. All rights reserved.
H. Park et al. Ocean Engineering 158 (2018) 78–98

and determined the dynamics of wave loadings and the effects of open- force on the elevated structure (e.g. Wiebe et al., 2014; Park et al., 2017).
ings in bridge decks. Based on this work, they derived predictive methods Impulsive breaking waves (wave slamming) induce the highest hori-
for both quasi-static and impulsive wave loads. Bradner et al. (2011) zontal forces on vertical walls (e.g. Bea et al., 1999; Linton et al., 2012);
investigated a 1:5 scale model of a reinforced concrete type coastal however, it is reported that the breaking wave itself could be separated
bridge superstructure. They investigated the vertical and horizontal force into three phases such as early breaking, late breaking, and perfect
for an array of wave heights, periods, and water levels and found that the breaking (Kirkg€ oz, 1995). Each phase creates a wave impact force of
vertical force could be greater than the horizontal force by a factor of different magnitude. Therefore, it is still a challenging task to model the
four. various types of wave deformations over a sloping beach and to calculate
Recently, ASCE/SEI (2016) published minimum design standards for the consequent wave pressure distributions and forces on elevated
buildings as a design guideline for elevated structures, including meth- structures for various types of wave breaking conditions.
odologies to calculate three wave loading types: non-breaking, breaking The goals of the present paper are to: (1) validate two CFD models
and broken (Sections 5.4.2 to 5.4.5). FEMA (2011) also published a (IHFOAM and FLUENT) with a scaled (1:10) experimental dataset of
design manual for residential coastal dwellings, which includes guidance waves impacting an elevated structure (Park et al., 2017) with a range of
on both horizontal and vertical wave forces, similar to those presented in surge levels, wave conditions, and air gaps; and (2) quantify the perfor-
the ASCE standard. However, those theoretical approaches are limited in mance of the two models for the wave induced horizontal and vertical
their applicability to real, complex, storm wave climates and the corre- forces (pressures) on the elevated structure. We compare the perfor-
sponding wave forces. Wiebe et al. (2014) proposed an analytical solu- mance and sensitivity of the two CFD model results conditioned on the
tion to estimate the wave-induced force on an elevated structure by three different wave impact conditions: non-breaking, breaking, and
modifying Goda's pressure formulae for a caisson breakwater (Goda, broken. In particular, we compare the time-series of the water surface
1974, 2010) to include the effects of various freeboard conditions (air-- elevations over the compound slope, as well as the front and bottom
gap) and wave climates. They concluded that three types of breaking pressures and vertical and horizontal force components on the elevated
conditions yielded different aspects of horizontal wave force as function structure, which was positioned on a flat region slightly inland of the
of wave height, period and freeboard. compound slope. Section 2 introduces the overall setup of the experiment
As an alternative method, Computational Fluid Dynamics (CFD) and dataset utilized in model validation including detailed instrumen-
models have been widely developed and applied to estimate wave tation and test conditions. Section 3 introduces the two CFD models
induced pressures and forces; advances in recent years have been sup- (IHFOAM and FLUENT) and details of each CFD model setup. Section 4
ported by an increase in computation power, which allows more detailed presents an overview of the two CFD model results, including the detailed
calculations of the complex hydrodynamics associated with wave action. validation process. Section 5 discusses the quantitative comparison of the
However, the performance of CFD models must be validated or verified two CFD models and the sensitivity of each model to different mesh size
through detailed comparisons with benchmark tests (e.g. analytic solu- conditions in calculations of the water surface elevation, pressure, and
tions or physical experiments), and difficulties of pressure and force force. Finally, Section 6 discusses the major conclusions of this work.
calculations and their sensitivity to wave conditions were well reported
by previous studies (e.g. Mokrani and Abadie, 2016). In case of the wave 2. Experimental design
pressure (forces) on elevated structure, many recent numerical studies
have been performed with bridge decks (Xiao and Huang, 2008; Jin and The physical model tests were conducted at the Large Wave Flume
Meng, 2011; Hayatdavoodi et al., 2014, 2015; Seiffert et al., 2015; Wu (LWF) at Oregon State University's Hinsdale Wave Research Laboratory
et al., 2016) and validated through the scaled physical experimental re- (HWRL) and are described in detail in Park et al. (2017). The tests are
sults. They mostly utilized solitary or cnoidal waves with constant depth summarized here to provide a context for the numerical modeling in this
conditions for simple wave climate and focused on analyzing wave loads work. The experiment was designed to measure wave-induced pressures,
on coastal bridge decks. Recently, Do et al. (2016) used the results of a horizontal forces, and vertical forces separately on an elevated coastal
1:5 scale bridge experiment (Linton et al., 2012) to validate the ANSYS structure. Experiments were performed with a constant water depth and
FLUENT model (ANSYS, 2013). They then applied the model setup to a bathymetry while varying the significant wave height and peak period
vertical wall using a solitary wave. for each test. Three different wave conditions were tested including
Generally, differences in wave impacts result from the three breaking regular (periodic), irregular (random), and transient (tsunami-like)
types (non-breaking, breaking, and broken waves). Wave breaking is waves, and the specimen was elevated to different air-gap conditions to
highly dependent on the wave characteristics (e.g. wave height, period, measure the horizontal and vertical wave forces on the structure. The
and surge levels), bathymetric and topographic conditions (e.g. cross- model-data validations use the regular waves only.
shore beach profile, foreshore slope, and presence or absence of
offshore sandbars and dunes). Geometric conditions of the infrastructure 2.1. Experimental setup
(e.g. community layout, beach hardening) are also important. In partic-
ular, the air gap, or the distance from the water level to the lowest chord A profile-view sketch of the LWF including bathymetric conditions
of the structure, can significantly affect the magnitude of the wave impact and wave gage locations is presented in Fig. 1. The length, width and

Fig. 1. Profile view of the Large Wave Flume (LWF).

79
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 2. Diagrams of experimental specimen and instrumentation. (a) Top view of zoomed in wave flume with the locations of string (circles) and ultrasonic (triangle)
wave gages, (b) Specimen with frame, (c) Front, back, bottom and top views of specimen.

depth of the flume were 104 m, 3.66 m, and 4.6 m, respectively, with a 2.2. Experimental wave conditions
piston–type hydraulic actuator assembly wavemaker. A compound
slope was installed in the flume beginning 14.07 m from the wave- Regular wave trials consisted of 10 different combinations of sig-
maker. The first section was composed of a horizontal, 3.64 m long slab nificant wave heights and peak periods to model storm wave condi-
with a 0.15 m height, followed by a 10.98 m long section with a 1:11 tions. Table 1 shows regular wave conditions measured at wg1. Based
slope, giving a vertical rise of 0.99 m. This steeper slope was followed on observations and video analysis of the wave impact on the struc-
by a 14.64 m long section with a 1:24 slope, rising an additional 0.61 m, ture, we classify three types of wave impact conditions: non-breaking,
and finally followed by a 36.60 m long flat section starting at breaking, and broken. The non-breaking impact is generally observed
x ¼ 43.33 m with a constant 1.75 m elevation. The specimen, which was for relatively small wave heights and long wave period conditions. The
based on a 1:10 length scale model of a residential coastal structure, surface of sinusoidal wave transforms during the shoaling process,
was designed to measure vertical and horizontal forces and pressures approaching peaky, asymmetric sine curves. However, wave breaking
and was located at x ¼ 43.84 m, 0.51 m landward from the start of the does not occur, and thus the wave generates small run-up at the
flat section. specimen. Trials REG_X1, X2, and X6 were classified as non-breaking
Fig. 2 provides a detailed sketch of the instrumentation installed in wave impact conditions. The breaking wave condition is found when
the wave flume. Fig. 2a is a detailed overhead view of wave flume the shoaling wave height is over the breaking limits of depth or length
showing the experimental specimen and nearby wave gauges. Three (e.g. H/h ¼ 0.78 or H/L ¼ 0.14 tanh (kh)). The wave breaks near the
wire resistance wave gages were positioned at the beginning (wg1), specimen and creates a large jet of splashing water at the time of
middle (wg2), and end (wg3) of the slope to capture the trans- impact. Experimental trials REG_X3, X7, and X8 are classified as
formation as waves propagated along the sloping bathymetry. Fig. 2b
is a side view of the specimen. A frame was utilized to hold the
specimen in place at varying air gaps and to prevent lateral (y-direc- Table 1
tion) motion. The specimen has a box shape (1.02 m by 1.02 m square Regular wave conditions.
bottom with 0.61 m height) and it represents an idealization of an Exp. REG
elevated residential structure. This specimen was designed to measure
H (m) T (s) h (m) Dur. (min)
pressures on the front, bottom and rear sides; the rail connection to the
frame system allowed the simultaneous, independent measurements of X1 0.10 4.10 2.15 4.00
X2 0.21 4.10 2.15 4.00
the horizontal and vertical forces. Fig. 2c depicts the front, back,
X3 0.29 4.10 2.15 4.00
bottom and top views of the specimen, which show the locations of the X4 0.40 4.10 2.15 4.00
14 pressure gages and 5 load cells used in experiments. All pressure X5 0.50 4.10 2.15 4.00
gages and load cells recorded data at a frequency of 500 Hz, and data X6 0.16 2.52 2.15 2.50
was filtered using a low pass filter for frequencies above 50 Hz. More X7 0.23 2.98 2.15 3.00
X8 0.26 3.64 2.15 3.50
details of the coordinate locations of instruments are available in Park
X9 0.35 4.68 2.15 4.50
et al. (2017). X10 0.42 5.04 2.15 5.00

80
H. Park et al. Ocean Engineering 158 (2018) 78–98

breaking wave impact conditions. Lastly, relatively higher wave height 3. Numerical model
conditions result in the onset of wave breaking processes over the
slope and show a decrease of wave height with air entrainment at the 3.1. Model overview: IHFOAM and FLUENT
moment of impact. REG_X4, X5, X9, and X10 are classified as broken
waves in these experiments. The repeatability of wave maker and The first CFD model utilized in this study is IHFOAM 2.0 (Higuera
recording devices was tested, and results are described in the physical et al., 2013), which implemented a new solver, and boundary conditions
model study by Park et al. (2017). For all trials, the mean absolute for wave generation and absorption. IHFOAM was developed in the open
error was ~10% for pressure gages and less than 1% for water surface source platform of OpenFOAM® (Open Field Operation And Manipula-
elevation gages. For numerical model validation, we choose three tion). OpenFOAM solves the 3-D Reynolds Averaged Navier-Stokes
regular wave conditions: REG_X2, X8, and X4, which clearly show (RANS) equations for two incompressible phases using a finite volume
non-breaking, breaking, and broken wave conditions at the time the discretization and the volume of fluid (VOF) method. It also supports
wave reaches the specimen. For each trial, the breaking condition is several turbulence models such as κε, κω SST, and LES. The governing
verified based on both video recordings and pressure measurements on RANS equation includes conservation of mass (Eq. (1)) and momentum
the specimen. (Eq. (2)), which are
Fig. 3 provides snapshots from the video recordings of the wave
interaction on the specimen for (a) REG_X2, (b) REG_X8, and (c) r⋅!
u ¼0 (1)
REG_X4. In this study, we consider the case in which the base of the
structure was located at the still water level (giving an air gap, a, equal ∂ρ!u
 !
þ r⋅ðρ!
u!u Þ  r⋅ μeff ⋅r!
u ¼ rp*  !
g X ⋅rρ þ r!
u ⋅rμeff (2)
to zero), which was shown by Park et al. (2017) to induce the maximum ∂t
vertical force.
!
where each vector ! u, !
g , and X is velocity, acceleration of gravity and
position, respectively, and each scalar p* and μeff is pseudo-dynamic
pressure and efficient dynamic viscosity, which account for the molec-
ular dynamic viscosity plus turbulent effects (IHFOAM, 2017). The
density (ρ) is calculated as ρ ¼ αρwaterþ(1-α) ρair, where α indicates the
phase function, given by

∂α
þ r⋅! !αð1  αÞ
u α þ ru (3)
∂t c

In Eq. (3), !
uc is an artificial compression vector to preserve a sharp
interface between air and water (Maza et al., 2015). IHFOAM utilizes a
mixture of the PISO (Pressure Implicit with Splitting of Operators) and
SIMPLE (Semi-Implicit Method for Pressure-Linked Equations) algo-
rithms, which is called PIMPLE. The main structure of PIMPLE is
inherited from the original PISO, but it allows equation under-relaxation,
as in SIMPLE, to ensure the convergence of all the equations at each time
step (IHFOAM, 2017). The performance of IHFOAM was validated
through several benchmark studies such as run-up, wave breaking
(Higuera et al., 2013), coastal structures (Higuera et al., 2014) and 3-D
oscillating water columns (Iturrioz et al., 2015).
The second CFD model is FLUENT, a module in the ANSYS workbench
to solve for the fluid flow. FLUENT can interact with the mechanical
module in ANSYS to solve the two-way fluid-structure interaction;
however, this paper only focuses on one-way fluid-structure interaction
by using the FLUENT module only. Therefore, the structure can be
assumed to be a rigid boundary condition during the analysis. (Fluent
Inc., 2005).
The governing equations for Volume of Fluid (VOF) model for multi-
phase problem solving in Fluent are as below (Eqs. (4) and (5)) with the
assumption that temperature remains unchanged during the analysis
(ANSYS, 2009). The mass conservation equation is given as
   X
1 ∂
 n


αq ρq þ r: αq ρq !
vq ¼ m_ sq  m_ qs (4)
ρq ∂t s¼1

where ρq is the mass per unit volume of the phase q, t is time, r is the
differential operator, ! vq is velocity vector of phase q, αq is the volume
P
fraction of phase q with ns¼1 αq ¼ 1, m_ sq is mass transfer from phase s to
phase q, m_ qs is mass transfer from phase q to phase s, and n is the number
of phases.
The momentum conservation equation is used throughout the domain
as

∂ ! !
ðρ v Þ þ r⋅ðρ!
v!v Þ ¼ rp þ r:½μðr!
v þ r!
v Þ þ ρ!
T
g þ F (5)
Fig. 3. Photographs of wave impact conditioned on three different waves: (a) ∂t
non breaking, REG_X2, (b) breaking, REG_X8, and (c) broken, REG_X4.

81
H. Park et al. Ocean Engineering 158 (2018) 78–98

! the physical model. Both the water and air phases were simulated with
where p is the static pressure, and each ρ! g and F are the gravitational
density (ρ) set to 998 kg/m3 and 1.20 kg/m3, respectively, and kinematic
and external body forces, respectively, and μ is the molecular viscosity. In
viscosity (ν) set to 1.0  106 m2/s and 1.48  105 m2/s, respectively.
order to account for the turbulent effects, realizable k-ε model (ANSYS,
We utilized the default solver setups, described in the IHFOAM example
2009) was applied with the other two variables (k-turbulent kinetic en-
(IHFOAM, 2017) and both the κε and the κω SST turbulence equations
ergy, and ε-turbulent kinetic energy dissipation rate). In the next step,
were considered. The default parameters of kε (e.g. Launder and
implicit scheme is selected for discretize those governing equations, and
Sharma, 1974) and κω SST turbulence models (Wilcox, 2008) were
the pressure-based coupled algorithm is obtained to solve the discretized
utilized in this study, and we utilize kε equations in the final validation
equations for pressures, velocities, and other scalar equations.
results.
The sketch of the numerically modeled 3-D wave flume is presented in
3.2. IHFOAM model setup Fig. 4a. We treat the specimen and frame as located at the center of the
flume and cut the flume in half in the y-direction to reduce computation
There are two mesh tools in OpenFOAM: ‘blockMesh’ and ‘snappy- time. Three different sized meshes, coarse (M1), medium (M2), and fine
HexMesh’. The blockMesh tool creates the geometry of each block, which (M3) were utilized in this study, each of which had a variable mesh size
is generally defined by 8 vertices and 6 boundary faces, one at each depending on the location in the flume. Fig. 4b and c show the sketches of
corner of a hexahedron (cube). It requires information of the mesh the front and top views of the flume, giving detailed information for the
vertices, faces, and correct numbering to generate the mesh; thus, rela- fine mesh size (M3) and showing the variation in the mesh size in the
tively simple geometry conditions are preferred when using this tool. flume. The finest mesh (0.02 m) on the x-axis was concentrated near the
However, the snappyHexMesh tool generates 3-D meshes, creating specimen (L1), and the finest mesh (0.02–0.025 m) on z-axis was around
hexahedra and split-hexahedra automatically from triangulated surface the region of the free surface elevation (Z2 and Z3) to capture the non-
geometries given in StereoLithography (STL) format, so it supports the linear wave deformation including shoaling, reflection, and breaking
generation of mesh with complex geometry conditions. In this study, we processes during propagation over the slope. We set the mesh size at
utilize the blockMesh tool and generate a 1:1 scale model of the flume 0.02 m at the specimen on the y-axis, and gradually increased the mesh
used for the physical model study, including the specimen (painted or- size to 0.05 m at the lateral wall of the flume.
ange as shown in Fig. 3), but not the gray supporting frame, with set of The smallest mesh sizes for M1 and M2 were 0.08 and 0.04 m,
hexahedrons (cubes). We utilize a prism-shaped block, which is respectively, and the maximum mesh size for both configurations was
composed of 6 vertices and 4 boundary faces to reproduce the sloped limited to 0.2 m. The total number of grid elements increased by 2.6
bathymetry. The piston-type wavemaker is not directly modeled. Instead times from M1 to M2 and by 6.2 times from M1 to M3. Table 2 sum-
IHFOAM allows users to choose various wave input conditions based on marizes the mesh types and lists the total number of elements utilized in
wave theory (e.g. Stokes, solitary, and irregular) and wavemaker theory each of the mesh types. Regular wave conditions were simulated for a
(piston-type, flap-type). Here we utilize the piston-type wavemaker duration of 50 s through a single desktop with i7-4770 (3.4 GHz) and 64G
boundary condition (inlet) to replicate the wave input using the RAM with 4-core parallel. Since we have set the maximum Courant
measured time-series of displacement and free surface elevation at the Number (Co) to 0.5 for all model runs, different computation times were
wavemaker. We assign the no-slip wall boundary condition for both required for the different wave hydrodynamic conditions. The average
bottom slope and around the specimen, and leave as an empty boundary computation times for a 4-core computer are 1.1, 15.2, and 127.2 h for
the lateral walls. The bathymetry of the flume was designed to dissipate the M1, M2, and M3 meshes. Generally, the computation time increased
waves through the natural beach at the end of the flume. Therefore, we by roughly 1.3 times for the breaking condition (REG_X8), and 1.9 times
did not implement a radiation boundary condition or sponge layer at the for the broken wave condition (REG_X4) compared to the non-breaking
end of the flume, so that our simulation would replicate the conditions in wave calculations (REG_X2).

Fig. 4. (a) Sketch of wave flume in the IHFOAM model, and the mesh configuration for (b) front view and (c) top view.

82
H. Park et al. Ocean Engineering 158 (2018) 78–98

Table 2 because there is essentially no noise in the IHFOAM pressure and force
Mesh configuration for IHFOAM. results due to extremely small frequency components.
Mesh Types

M1 M2 M3 (fine) 3.3. FLUENT model setup

(coarse) (medium)

dx (m) 0.08 ̶ 0.2 0.04 ̶ 0.2 0.02 ̶ 0.1

Because of symmetry, one half of the wave flume and the specimen
dz (m) 0.08 ̶ 0.2 0.04 ̶ 0.2 0.02 ̶ 0.1 were modeled in FLUENT as was done in IHFOAM. The specimen was
dy (m) 0.08 ̶ 0.16 0.04 ̶ 0.08 0.02 ̶ 0.05 modeled as fixed walls, meaning no dynamic effects were considered for
Number of elements 505,125 1,306,971 3,149,370 the box during analysis. All wall boundaries were assumed split free. A
Co 0.5 0.5 0.5
multiphase volume of fluid model was employed to model water and air
Computation time of REG_X2 (hr) 0.8 9.9 82.3
Computation time of REG_X8 (hr) 1.0 16.2 131.5 phases. The realizable k  ε viscous model was selected for the solver. By
Computation time of REG_X4 (hr) 1.3 19.5 167.8 adding a new turbulent viscosity formulation and dissipation rate
Average computation time of three wave 1.1 15.2 127.2 transport equation, this realizable kε model is more accurately pre-
conditions (hr) dicted for both planar and round jets compared to the standard kε
model (ANSYS, 2009). An enhanced wall treatment function was intro-
duced for the near-wall region by blending the laminar and turbulent
The sampling rate and filtering of output results (surface elevation, law, and combining with the reliable kε model to produce better results
pressure, or forces) for model validation is also a critical issue for the for this type of problem (ANSYS, 2009). Output time steps were
pressure and force comparison. We utilize the function ‘Probes,’ which is controlled by a Courant number of 0.5. The convergence criteria for
added at the end of the ‘controlDict’ text file that sets the simulation time continuity and velocity components was 1E-6. Calculations for κ, ε and
and data input/output control in OpenFOAM. The Probes function allows the volume fraction were defined to converge at 1E-3 or repeated for a
us to extract any calculated values (e.g. velocity, pressure, k, and ε) at the maximum 50 iterations, at which point the analysis continued for
specific x, y, and z location with a varying frequency. We apply the computational efficiency. An advantage of FLUENT is that the wave-
Probes function for the 14 pressure gages and 9 wave gages, and extract maker can be treated as the moving boundary at the inlet so that the
each datum with the same time of model calculation. To set up the wave “numerical flume” has the same boundary conditions as the physical
gages in IHFOAM, we define a series of x, y, and z points along the z-axis wave flume with the piston-type wavemaker. The displacement of the
with 0.02 m resolution at each wave gage location. To obtain the surface physical wavemaker was recorded during the experiments and was used
elevation result, we take the sum of all α values along the z-axis at each as the input for the numerical wave flume. Remeshing every time step
wave gage location and multiply by the resolution (0.02 m). Here, α is was necessary for the area close to the inlet boundary to remain consis-
defined as the quantity of water per unit of volume at each cell; thus if tent with the initial element size, with a split and collapse factor of 0.4.
α ¼ 1, the cell is full of water and if α ¼ 0, the cell is full of air. To estimate The wave gauges were modeled by creating flat column with a height
the horizontal and vertical forces acting on the front and bottom faces of equal to that of the wave flume and width of 0.05 m at the wave gauge
the specimen, respectively, we set up a different patch for each the front locations as shown in Fig. 5a. Wave height was measured by integrating
and the bottom face in the blockMeshDict, and then applied ‘libforces.so,’ the water phase then dividing by the width of the column. This technique
which is the intrinsic function implemented in OpenFOAM to extract works well with a solid wave but typically underestimates the splash
corresponding forces at a given patch wall with three Cartesian di- compared to regular string wave gauges (wg) or ultrasonic wave gauges
rections. To validate the performance of ‘libforces.so’, we compared the (uswg). Pressures were collected by creating points at pressure gauge
results of ‘libforces.so’ to the integrated force, interpolated from multi- (PG) locations along the sides of the box as shown in Fig. 5b. PG1 to PG9
pressure points (36) at the front and bottom with normal direction, represent pressure on the front side of the box, and PG11 to PG14
and they showed good agreement. represent the pressure at the bottom of the box. The total horizontal and
Fig. A1 in the appendix provides the computation frequencies of M1, vertical forces were computed from the drag and lift coefficient at all
M2, and M3 during the 20–50 s simulation. As we mentioned earlier, the faces of the box divided by the reference values.
extraction time of all surface elevation, pressure and force data is the For comparison with IHFOAM, the mesh configurations also should
same as the computation time step (Fig. A1), and the computation time is be consistent. The cut-cell mesh method was used to obtain the maximum
governed by designating a maximum Courant Number (Co ¼ 0.5), which number of “square blocks,” which have the best element quality. The
varies depending on the velocity and size of mesh. The computation time overall minimum and maximum element size was set to be
step is small enough to compare with the measurements, and ranges from 0.02 m–0.16 m. Element size increased the block edge by a factor of two
300 Hz to 1000 Hz at the moment of wave impact for the non-breaking from fine resolution (M3) to medium resolution (M2) and from medium
wave condition (M3). Although the measurement results were pro- resolution to coarse resolution (M3), as shown in Table 3. The total
cessed using a low pass filter for frequencies more than 50 Hz (Park et al., number of elements increased by a factor of 2.7 from M1 to M2 and by a
2017), the IHFOAM results do not include any filtering in this study factor of 6.3 from M1 to M3. To evaluate the element quality of the M3

Fig. 5. (a) The mesh sketch of wave flume (M2), and (b) instrument locations on the specimen.

83
H. Park et al. Ocean Engineering 158 (2018) 78–98

Table 3 4. Validations and comparisons of numerical models

Mesh configuration for FLUENT.
Mesh Types In Section 4.1, we will present 3-D snapshots of modeling results for
the three wave conditions. In Sections 4.2, 4.3 and 4.4, we provide a
M1 M2 M3 (fine)
(coarse) (medium) detailed comparison of the numerically modelled water surface elevation
and pressure gauge time series to the experimental data at six wave gages
dx (m) 0.08 ̶ 0.16 0.04 ̶ 0.16 0.02 ̶ 0.16
dz (m) 0.08 ̶ 0.16 0.04 ̶ 0.16 0.02 ̶ 0.16
(wg2, wg3, uswg2, uswg4, uswg5, and uswg6) and six pressure gages
dy (m) 0.08 ̶ 0.16 0.04 ̶ 0.16 0.02 ̶ 0.16 (press1, press3, press6, press11, press12, and press14) for all wave
Number of elements 544,880 1,458,243 3,419,530 conditions. We also compare the time-series of the horizontal and vertical
Courant number 0.5 0.5 0.5 forces measured from five load cells on the specimen to the numerically
Computation time of REG_X2 (hr) 16.1 40.7 120.2
modelled results. Note that wg2 and wg3 were located at the middle and
Computation time of REG_X8 (hr) 17.5 62.4 178.2
Computation time of REG_ X4 (hr) 16.7 81.5 230.4 end of the slope, respectively, to capture wave shoaling and breaking,
Average computation time of three wave 16.7 61.5 176.3 and uswg 2, 4, 5, and 6 were positioned on the flat region to capture wave
conditions (hr) deformation around the specimen (Fig. 2a). The three ultrasonic wave
gages (uswg) were offset 0.53 m from the wall, with uswg2 uswg6, and
uswg5 positioned in front, at the middle, and behind the specimen,
mesh, the average element orthogonal quality (ranging from 0 to 1, closer respectively; uswg4 was also located behind the specimen but 1.83 m
to 1 is better) was determined as 0.99, with a minimum of 0.18. The offset from the wall (Fig. 2a). For comparisons of pressure time series,
average skewness (ranging from 0 to 1, closer to 0 is better) is 1.5E-3 with press6, press3, and press1 were located on the front face of the specimen,
maximum of 0.81. The average computation times for the three cases run and press11, press12 and press14 were located on the bottom of the
on a 12-core computer node in the Linux system (two Intel Xeon E5-2640, specimen (Fig. 2c).
2.5 GHz) are 16.7, 61.5, and 176.3 h for the M1, M2, and M3 meshes,
respectively. 4.1. Overview results of IHFOAM and FLUENT
By comparing Tables 2 and 3, we can see that the number of mesh
elements for IHFOAM and FLUENT were similar and vary by less than 5% Figs. 6 and 7 provide snapshots of the water surface field given by
for each of the mesh configurations. However, each model shows IHFAOM and FLUENT modeling at four time steps, including at the time
significantly different times of total computation. Those differences are of wave impact. Panels a, b, c, and d in each figure show the non-breaking
somewhat expected, because we could not perform both IHFOAM and wave condition; Panels e, f, g, and h show the breaking wave condition;
FLUENT runs under the exact same conditions, such as solvers, numerical and Panels i, j, k, and l show the broken wave condition.
schemes, parallelization tool, and computational resources. In this study, For IHFOAM, Fig. 6 a, e, and i show the wave surface about 1.5 s
IHFOAM has a relatively shorter computation time than FLUENT for M2 before the impact on the specimen, and each wave is a different shape
and M3, and a much shorter computation time for the coarse mesh depending on input conditions. Fig. 6 c, g, and k include the detailed
condition (M1). figures for non-breaking, breaking, and broken wave impacts on the front

Fig. 6. Numerical water surface field snapshots of IHFOAM for three wave conditions: REG_X2, nonbreaking, (a, b, c, and d), REG_X8, impulsive breaking (e, f, g, and
h), and REG_X4, broken (i, j, k, and l).

84
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 7. Numerical water surface elevation from FLUENT for the three wave conditions: REG_X2, nonbreaking, (a, b, c, and d), REG_X8, impulsive breaking (e, f, g, and
h), and REG_X4, broken (i, j, k, and l).

of the specimen. The frontal distributions of α for both non-breaking and (Fig. 7g) shows the strongest water splash on the structure, followed by
breaking waves are uniform, while the broken wave impact shows an the non-breaking and then the broken wave condition, respectively.
irregular distribution of α due to air entrapment. When the comparing
model output with the recorded videos of experiments, we observed that 4.2. Nonbreaking wave (REG_X2)
the IHFOAM results for the breaking and broken wave conditions tend to
be less developed than waves generated in experiments (i.e. numerical Fig. 8 shows the model validation of the water surface elevation for
waves were not quite impulsively breaking or broke closer to the struc- the non-breaking wave (REG_X2) condition. We compared the time series
ture than experimental breaking and broken waves, respectively). Thus, of water surface elevation at six wave gages from 20 to 50 s of the
the numerical model results showed relatively less splash for the experimental simulation time, which includes the first regular wave
breaking wave conditions and more splash for the broken wave impacting the specimen and 7 continuous regular waves. As waves
conditions. propagate toward the specimen, their shapes transform and steepen from
Fig. 7 shows the same time range as Fig. 6 for the three wave con- wg2 to uswg2. Overall, the wave heights increase due to shoaling effects,
ditions modeled by FLUENT in 2D contours. We can observe the highest and they are almost near breaking at uswg2. After waves pass the spec-
wave height for the breaking wave condition (Fig. 7e), a smaller wave imen, they keep the steep waveform, but in general, wave heights
height for the non-breaking wave condition (Fig. 7a) and the lowest wave decrease. Generally, both IHFOAM and FLUENT results show excellent
height for the broken wave condition (Fig. 7i). By inspecting Fig. 7 c, g, agreement to the measured surface elevation, but both models slightly
and k, we can see that at the time of impact, the breaking condition underestimate the positive surface elevation and show relatively high

Fig. 8. Comparisons of water surface elevation for the nonbreaking wave case, REG_X2: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue
dash dot line). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

85
H. Park et al. Ocean Engineering 158 (2018) 78–98

deviations at uswg4 and uswg5, which are located behind of the spec- waves (t > 33 s). The maximum reduction of peak pressure is observed at
imen. One interesting feature is that the first four waves of both IHFOAM press6, decreasing from 2.3 kPa to 1.0 kPa. This reduction highlights that
and FLUENT show excellent agreement at most of the wave gages, but the even minor changes of water surface elevation resulted in large peak
models begin to show larger deviations for subsequent waves, especially pressure changes (in this case, by over a factor 2) for both calculated and
at wg3, uswg4, and uswg6. These disagreements could come from the measured data.
effects of wave reflection from the specimen and sidewalls or from wave- Fig. 10 shows the model validation of the non-breaking wave con-
structure interaction, which the numerical model cannot capture well dition (REG_X2) for the bottom pressure gages. We compare the time
due to the rigid body assumption. series of pressure at press11 (a), press12 (c), and press14 (e) and show
Fig. 9 shows the model validation of the non-breaking wave condition the corresponding detailed time series of pressure for one wave (b, d, and
for the front pressure gages. We compare the time series of pressure at f). Both IHFOAM and FLUENT results could not capture the large peaky
press6 (a), press3 (c), and press1 (e) and show the corresponding detailed pressures at the time of impact, but both models captured the pattern of
time series (t ¼ 32–35 s) at each pressure gage (b, d, and f). Generally, pressure distributions well. The short peaky pressures at the time of wave
both IHFOAM and FLUENT match well with the measured peak pres- impact had a relatively short frequency (~20 Hz), and may require much
sures, but IHFOAM slightly underestimates the peak at press6, while finer mesh for the numerical model simulation to capture. In addition,
FLUENT underestimates the peak at press3 and press1. The overall negative pressures are observed in both experimentally measured and
pattern of pressure distribution at the time of impact shows good numerically modeled results. Relatively large negative pressures are
agreement, except for instances where the experimental gages recorded observed at press11, with smaller negative pressures at press12. Non-
negative pressures, which are found at press3 and press6. It is interesting negative pressures are seen at press14 for both measured data and
to note that only press3 and press6 recorded negative pressures among IHFOAM, but they are not clearly found in FLUENT. Each pressure gage
whole time span of the three wave gages. Negative pressures were also recorded a uniquely shaped time series of pressure depending on its
observed press3 for the breaking wave condition (REG_X8), but were not location, and both models capture this variation well for each pressure
observed at any pressure gage for the broken wave condition (REG_X4), gage. The major difference between the two models when compared with
as seen in Figs. 12 and 16. This phenomenon could be caused by the large the measurement is that IHFOAM results calculate a similar peak pressure
jet of splashing water that is projected upward at the time of wave distribution after t ¼ 30 s of simulation time for all gages, but FLUENT
impact, but the current mesh size (0.02 m) is not small enough to capture either could not capture the peak pressure or modeled a relatively smaller
it. The largest peak pressure is recorded at t ¼ 33 s for both the mea- peak, especially at press12 (Fig. 10d) and press14 (Fig. 10e). Generally,
surement and calculations, which is the time of impact of the first wave the FLUENT results underestimated the peak pressure compared to the
whose wave height is equivalent to our input condition (~0.2 m at wg1). IHFOAM results.
Fig. 9b, d, and f show the detailed time-series of pressure near the time of Fig. 11 shows the model validation of the horizontal force and vertical
the largest peak pressure. A reduction in peak pressures at all gages was force for the REG_X2 wave condition. Overall, the IHFOAM outputs show
observed for subsequent wave impacts (t > 33 s) for both calculated and good agreement with the measurements for both the peak and pattern of
measured data. This pressure reduction occurs despite the observed wave forces, and FLUENT outputs show a slight overestimation of the
surface elevation at uswg2 recording similar wave heights for subsequent horizontal force. The measurement results show a clear damping of forces

Fig. 9. Comparison of front pressure time series for press6, 3 and 1 for t¼20–50 s (a, c, and e) and detailed comparison for one wave, t ¼ 32–35 s (b, d, and f) for
nonbreaking wave case, REG_X2. Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to
color in this figure legend, the reader is referred to the Web version of this article.)

86
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 10. Comparison of bottom pressure time series for press11, 12 and 13 for t¼20–50 s (a, c, and e) and detailed comparison for one wave, t ¼ 32–35 s (b, d, and f)
for nonbreaking wave case, REG_X2. Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to
color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 11. Comparison of (a) horizontal and (c) vertical forces for nonbreaking wave case, Reg_X2 and the respective cases shown in detail for t¼32–35 s (b and d):
Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to color in this figure legend, the
reader is referred to the Web version of this article.)

after wave impacts, especially for higher peak forces (t ¼ 33 and 37 s), bottom pressure gages (Fig. 10), these deviations did not result in a
but those damping effects decrease as the magnitude of the peak force significant difference in the vertical force comparison for both IHFOAM
decreases (t ¼ 41, 45, and 49 s). In contrast, IHFOAM and FLUENT results and FLUENT.
do not predict any damping effects on forces because of the rigid body
assumption applied in the calculation. Comparing modelled and 4.3. Breaking wave (REG_X8)
measured vertical forces, we observe that the overall shape of forces and
the peaks generally match, while IHFOAM shows some deviation at the Fig. 12 shows the model validation for the breaking wave condition
tail of each wave impact, especially for negative forces after wave impact. (REG_X8) for water surface elevation. Both IHFOAM and FLUENT results
Even though we could not capture the peaky pressures recorded by match well with the measured data at wg2 (Fig. 11a), but calculated data

87
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 12. Comparisons of water surface elevation for the breaking wave case, REG_X8: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue
dash dot line). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

show some deviations to the measurements at the other wave gages. structure interaction and wall or specimen-induced reflection effects.
Generally, both IHFOAM and FLUENT results underestimate wave height Fig. 13 shows the model validation for the breaking wave (REG_X8)
at wg3 and uswg2. To be specific, IHFOAM results underestimate wave condition for the front pressure gages. We compare the time series of
heights at wg3 after t ¼ 35 s, and at uswg2 after t ¼ 40 s. These differ- pressure at press6 (a), press3 (c), and press1 (e) and show the detailed
ences propagate to the other wave gages located behind the specimen. time series of pressure for t¼33–36 s (b, d, and f). The overall patterns of
FLUENT outputs show relatively better agreement at wg3. However, pressure impact times at each gage show good agreement for both
FLUENT also underestimates wave height at uswg2 after t ¼ 37 s. These models. However, the IHFOAM results underestimate the front pressure
patterns are similar to what we observed for the non-breaking wave compared to FLUENT, and both models show larger deviations from the
condition in that the numerical results showed good agreement to the measurement for the upper pressure gage, especially at press6 (Fig. 13a).
measurements until the wave height reached the target wave height The underestimations of pressure in IHFOAM and FLUENT are somewhat
(0.35 m at wg1). However, once the experimental waves developed to expected based on the underestimation of wave height at wg3 and uswg2
fully breaking (e.g. after t ¼ 35 s at uswg3), both model results match seen in Fig. 12. Generally, IHFOAM results indicate a weaker plunging
experimental results relatively poorly. As a given wave passes the spec- type of wave breaking at the specimen compared to the experimental
imen, the wave height generally decreases due to energy loss from wave measurements. Consequently, the numerical model significantly un-
breaking and wave impact. IHFOAM results show better agreement to the derestimates the peak pressure at press6. FLUENT results also underes-
measured data, while FLUENT results underestimate the wave height at timate the frontal pressure but show relatively better agreement to the
uswg5 and uswg6. Also, a slight phase shift is observed at uswg2, uswg5 measurement for their peak pressure outputs. Still, FLUENT cannot
and uswg6 in IHFOAM, which is not seen for the non-breaking condition capture the peak pressures. In addition, as was seen for the non-breaking
or for FLUENT data. REG_X2 case, negative pressures are measured at press3, and both
The change in wave heights from wg2 to uswg2 indicate the effects of IHFOAM and FLUENT could not capture these negative pressures. The
wave shoaling and breaking processes over the compound slope. We largest peak pressure was recorded at t ¼ 33 s at press6, which is the
observe that the wave height decreases slightly from wg2 to uswg2 in the impact time of the first wave whose wave height was equal to the
measured data. IHFOAM wave heights are almost same or have slightly intended wave height. Both model results predicted the largest peak
increased, while FLUENT wave heights have clearly decreased between pressure for this wave.
the two wave gages. IHFOAM shows better agreement to the measured Fig. 14 shows the model validation of breaking wave condition for
data after the specimen at uswg5 and uswg6, while FLUENT results un- bottom pressure gages. We compare the time series of pressure at press11
derestimate wave height at these locations. Upon careful observation of (a), press12 (c), and press14 (e) and show the detailed time series near
both the experimental and numerical model results including the snap- the maximum peak pressure (b, d, and f). Overall, both IHFOAM and
shots in Figs. 6 and 7, we conclude that IHFOAM waves tend to break FLUENT results show good agreement with the measurement. Frontal
later than the experimental data, while waves output by FLUENT tend to peaky pressures are observed for all measured data. IHFOAM outputs did
break earlier along the compound slope compared with the measured not capture the first short peaks (e.g. at t ¼ 33.9 s in Fig. 14b); however, it
data. Overall, we find that both IHFOAM and FLUENT results differ from captures well the secondary peak pressure (e.g. at t ¼ 34.2 s in Fig. 14b).
the experimentally measured data; however, they show opposite patterns FLUENT results capture relatively well the first peak pressure but could
of deviation, with IHFOAM waves breaking later and FLUENT waves not capture the secondary peak pressure. Overall, the absolute peak
breaking earlier than experimental data. Deviations between modeled pressure for each wave is similar to the measurement, but each model
and measured wave heights are larger for the breaking wave condition shows different pressure distributions for the vertical pressures. As the
than for the non-breaking wave condition. Errors may be induced due to pressure gages move from the leading edge (press11) to the trailing edge
the limitations of the current numerical models' setups (e.g. mesh, default (press14) of the specimen, the peak pressure tends to decrease, and the
solver, or turbulence model) and the complexity in simulating wave- expanse of the positive pressure region increases. In addition, IHFOAM

88
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 13. Comparison of frontal pressure time series for t¼20–50 s (press6, 3 and 1) and detailed time series of pressure for t¼33–36 s (b, d, and f) for the breaking wave
case, REG_X8: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to color in this figure
legend, the reader is referred to the Web version of this article.)

results show a phase shift at press14, similar to the phase shift for the breaking wave shape; hence, numerical outputs of vertical force show
water surface elevation that we observed at uswg5 and uswg6 (Fig. 11e good agreement to the measurement.
and f).
Fig. 15 shows the model validation of the horizontal force and vertical 4.4. Broken wave (REG_X4_a5)
force for the REG_X8 wave condition. Overall, IHFOAM results under-
estimate the experimentally measured horizontal force but show better Fig. 16 shows the model validation of the free surface for the broken
agreement to the measured vertical force. FLUENT shows better agree- waves. Generally, IHFOAM results show good agreement with mea-
ment to the measurement for the horizontal forces but greater deviation surement, capturing the shoaling and breaking processes recorded in the
from the measured vertical forces. Generally, the characteristics of the experiment for first four waves (before 35 s at uswg2). However, it
wave at the moment of impact significantly affect the magnitude of the overestimates the water surface elevation at uswg2, 5, and 6 for subse-
horizontal force. To specify, IHFOAM results show good agreement to the quent waves (after 35 s at uswg2), and we observe a slight phase shift in
measurement for both horizontal and vertical forces at t ¼ 31 s. However, IHFOAM water surface outputs. FLUENT tends to underestimate the
IHFOAM significantly underestimates the horizontal force for the water surface elevation at most of wave gages except at uswg5, and
following waves, while the vertical force shows slightly uniform results FLUENT results show a similar phase shift compared with the experi-
after t ¼ 31 s for both IHFOAM and measurement. Upon careful obser- mental measurements. The different aspects of the water surface eleva-
vation of both the experimental and IHFOAM model results, we find that tion between two models, especially at wg3 and uswg2, indicate that
the plunging shape of the wave as it impacted the front face of the IHFOAM and FLUENT each undergo different wave shoaling and
specimen in IHFOAM was not same as the shape of the experimental breaking calculations. These calculations are dependent on each model's
wave. The measurement shows clear plunging impact on the specimen inherent differences or set up preferences, which we cannot define in
while IHFOAM did not a show fully developed plunging wave impact. exactly the same way for this study (e.g. default model setups, numerical
This difference in wave shape causes the underestimation of the front solvers, and schemes). IHFOAM waves tended to break later compared
pressure and horizontal force after t ¼ 34 s. In addition, both model re- with measured results, in that the numerically modelled results indicated
sults were unable to capture the horizontal force measured at the impact that the wave broke closer to the front of the specimen compared to the
including the inertial force on the specimen, which consequently effected measurement. In contrast, waves modelled by FLUENT tended to break
the underestimation of the horizontal force. We also observe clear sooner than the measured data. In summary, IHFOAM overestimate the
damping effects in the measured horizontal force after the wave impact, water surface elevations while FLUENT underestimates the water surface
while there are no oscillations in either model result. Relatively small elevations associated with broken waves. These trends propagate
damping effects are observed for the vertical force, and each vertical through in comparisons between the experimental and numerical pres-
force is generally uniform. In summary, we can conclude that the hori- sure and force data.
zontal force is sensitive to the shape of the breaking wave in the nu- Fig. 17 shows the model validation of the front pressure gages for the
merical model results, while the vertical force is less sensitive to the broken waves. We compare the time series of pressure at press6 (a),

89
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 14. Comparison of bottom pressure time series for t¼20–50 s (press11, 12 and 14) and detailed time series of pressure for t¼33–36 s (b, d, and f) for breaking
wave case REG_X8: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to color in this
figure legend, the reader is referred to the Web version of this article.)

Fig. 15. Comparison of (a) horizontal and (c) vertical forces for breaking wave case, Reg_X8 and the detailed time series for t¼33–36 s (b and d): Experiment (solid
black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to color in this figure legend, the reader is referred to the
Web version of this article.)

press3 (c), and press1 (e) and show the detailed time series of pressure comparison of water surface elevation results at wg3 and uswg2 (Fig. 16b
between t¼31–34 s (b, d, and f). In general, IHFOAM results overestimate and c). IHFOAM results show that the wave broke closer to the front of
the frontal pressure after t ¼ 36 s, while FLUENT results tend to under- the specimen compared to the measurement, while FLUENT results show
estimate the pressures for the entire time series. The deviations of both that the wave broke further offshore of the specimen compared to the
model results from measurements were somewhat expected based on the measurement. Thus, IHFOAM results produced higher wave heights at

90
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 16. Comparison of water surface elevation for broken wave case REG_X4: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot
line). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 17. Comparison of frontal pressure time series for t¼20–50 s (press6, 3 and 1) and detailed time series of pressure for t¼31–34 s (b, d, and f) for the broken wave
case REG_X4: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to color in this figure
legend, the reader is referred to the Web version of this article.)

uswg2 and higher front pressures than the measurement, while FLUENT broken waves. We compare the time series of pressure at press11 (a),
results produced lower wave heights at uswg2 and smaller front press12 (c), and press14 (e) and detailed time series of pressure between
pressures. t¼31–34 s (b, d, and f). Generally, bottom pressure results given by the
Fig. 18 shows the model validation of the bottom pressure gages for models show better agreement to measurements than front pressures.

91
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 18. Comparison of frontal pressure time series for t¼20–50 s (press11, 12 and 14) and detailed time series of pressure for t¼31–34 s (b, d, and f) for the broken
wave case, REG_X4: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT (blue dash dot line). (For interpretation of the references to color in this
figure legend, the reader is referred to the Web version of this article.)

Still, we observe a similar pattern to the horizontal pressure validation in force for the REG_X4 wave condition. Generally, IHFOAM overestimates
that IHFOAM results overestimate pressure at press11 and FLUENT re- both the horizontal and vertical forces, as was observed in comparisons of
sults underestimate pressure at press 11, 12, and 14. Fewer short peaky both the front and bottom pressures, while FLUENT shows good agree-
pressures were observed in the measurement for REG_X4, and pressures ment to the measurements. One explanation for this large difference in
generally decrease as the wave passes by the specimen, similar to the pressure but relatively small difference in the force is that the bore type
REG_X8 condition. In addition, both IHFOAM and FLUENT results show wave randomly impacts the front of the structure, creating wet and dry
phase shifts at press12 and 14; these are related to the phase shifts of conditions that are not the same in the physical and numerical models.
water surface elevations that we observed for both models at uswg5 and Consequently, peak pressures do not all occur at the same time. There-
6 (Fig. 16e and f). fore, the magnitude of the peak pressure could vary significantly
Fig. 19 shows the model validation of the horizontal force and vertical depending on location of interest on the structure, and any peak pressure

Fig. 19. Comparison of (a) horizontal and (c) vertical forces for broken wave case Reg_X4: Experiment (solid black line), IHFOAM (red dashed line), and FLUENT
(blue dash dot line). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

92
H. Park et al. Ocean Engineering 158 (2018) 78–98

at a point provides a relatively small contribution to the total force. For where JR is the impulse of absolute residual at the specific time range
example, the maximum horizontal force in IHFOAM is greater than the from t1 to t2. We choose t1 and t2 as the simulation time of numerical
measurement by a factor of 2 at t ¼ 40 s, while the numerically output model (50 s), but we exclude the initial three waves for ramp up which do
front pressure is greater than the measured by a factor of 5 at press6. The not satisfy our input condition. Thus, t1 is the time at which the fourth
observed points considered in comparisons with measurements (press1, wave reaches uswg2, and t2 is the time at which the next four waves have
3, and 6) could not capture the complete pattern of the pressure distri- propagated past uswg2. For example, we observe the fourth wave of
bution and the corresponding horizontal force of the bore type wave. REG_X2 at uswg2 at t ¼ 32 s. Therefore, t1 is 32 s and t2 is 48 s (32 s þ 4
waves  4 s (wave period)). Note that the wave period in both REG_X2
5. CFD model residual comparison and sensitivity test and REG_X4 is 4 s, and the wave period in trial REG_X8 is 3.5 s.
We normalize JR with the impulse of absolute force, or force per unit
5.1. Residual comparison of two CFD models area (pressure), given by

Time series comparisons between IHFOAM, FLUENT, and measured t

JX ¼ ∫ t21 jXjdt (7)
free surface elevations, pressure time series, and force time series were
performed in previous section. Note that both the peak value and dura- where JX is the impulse of force or force per unit area (pressure) at the
tion of wave-induced pressures (forces) are important factors that can specific range of time. Finally, the normalized form of ratio of impulse
affect structural damage (e.g. Bullock et al., 2007). Therefore, it is (RJ) is represented as
necessary to determine the uncertainty not only in the peak pressure, but
also throughout the time series of a given wave impact. To quantify the JR
RJ ¼ (8)
uncertainty of time series data, we determine the residual between the JX
calculations and measurements, Residual (R) ¼ Measurement – Calcula- Fig. 21 shows the IHFOAM results of the ratio of impulse for all front
tion. Fig. 20 shows an example of the measured and calculated pressure and bottom pressure gages and horizontal and vertical forces for the
and the resulting R using IHFOAM with 0.01 s resolution for the three three wave conditions considered. Each filled and void symbol indicates
wave conditions. A negative residual indicates overestimation by the the results of pressure and force, and each red circle, blue square, and
model. black triangle indicates the results of non-breaking (REG_X2), breaking
To quantify the residuals over a given time duration, we integrate the (REG_X8) and broken wave conditions (REG_X4), respectively. Generally,
absolute value of residual over the time domain: the highest RJ is observed for the broken wave condition, and the
t smallest RJ is observed for the non-breaking wave condition. As observed
JR ¼ ∫ t21 jRjdt (6)
in Section 4, the best agreement between measured and calculated
pressure, FH, and FV is found for the non-breaking wave condition. In
addition, a relatively larger RJ is observed at lower frontal pressure gages
(e.g. at press1) for all three wave conditions. Non-breaking wave con-
ditions also show the smallest RJ for bottom pressures; however, for this
wave condition, the ratio of impulse is larger for bottom pressures than
for front pressures. The calculated RJ values for the bottom pressures in
the breaking and broken wave conditions were relatively more consistent
than those calculated for frontal pressures (except at press11). We can
conclude that the horizontal force and frontal pressures are more sensi-
tive to the wave breaking condition than the vertical force and bottom
pressures. The residual of surface elevation is also evaluated in a similar
manner, but we find that generally the RJ for surface elevation is a factor
103 smaller than the pressure and force results. Note that because of the

Fig. 21. Quantitative IHFOAM comparison of the ratio residual impulse for
Fig. 20. Examples of the residual of pressure at press2 for (a) REG_X2, (b) pressures and forces with the three wave breaking conditions. Each red circle,
REG_X8, (c) REG_X4. Each solid line, red dash line, and dotted indicates the blue square, and black triangle indicates the results of non-breaking, breaking
measurement, calculation, and corresponding residual with 0.01 s resolution. and broken wave conditions, respectively. (For interpretation of the references
(For interpretation of the references to color in this figure legend, the reader is to color in this figure legend, the reader is referred to the Web version of
referred to the Web version of this article.) this article.)

93
H. Park et al. Ocean Engineering 158 (2018) 78–98

definitions used in Eq. (6)–(8), particularly the time intervals used, the wg2, and larger deviations are observed at the peak and tail of the water
resultant RJ is not a direct measure of the absolute error in the model, so surface elevation time series at uswg2, which was located closer to the
RJ ¼ 0.01 should not be interpreted as an error of 1%. However, since the specimen.
time intervals are the same for all comparisons in this section, the values The detailed comparisons of the front and bottom pressures are
can be used to show the relative errors between the computed outputs of shown in Fig. 24. We focus on the same time duration of the fourth wave
pressure and force and between the two numerical models (e.g. impacting the structure that was used for the pressure comparisons in
comparing Figs. 21 and 22). Section 4. Fig. 24a and b show the comparison of results for the non-
In a similar manner, we perform the quantitative comparison for breaking wave at press2 and press12. Fig. 24c and d show the compari-
FLUENT. Fig. 22 shows the ratio of impulse for FLUENT results of all son of results for the breaking wave at press2 and press12, and Fig. 24e
pressure gages (except for the rear gage, press10) and horizontal and and f show the comparison of results for the broken wave at press2 and
vertical forces for the three wave conditions. Similar to IHFOAM re- press12. All mesh sizes M1, M2, and M3 show a similarly shaped non-
sults, the highest RJ is, in general, observed for the broken wave con- breaking wave-induced pressure distribution for front and bottom gages.
dition, and the smallest RJ is observed for the non-breaking wave The finest mesh produces relatively higher pressure outputs over the
condition. Also, a relatively larger RJ is observed at lower frontal duration of the wave and also shows slightly shorter duration of pressure
pressure gages (e.g. at press1) for all three wave conditions. However, compared to the two coarser mesh conditions. For the breaking wave
FLUENT results show the highest RJ for both horizontal and vertical condition, M1 clearly underestimates the pressure for both front and
forces at the breaking wave condition. In contrast, the RJ for both non- bottom pressures compared with the other two mesh conditions. For the
breaking and broken wave conditions are similar and significantly broken wave condition, both and M2 and M3 capture the frontal peak
smaller than that calculated for the breaking wave condition. This pressure, while M1 could not capture it. Generally, all mesh resolutions
pattern follows our observation in Fig. 19 that FLUENT results showed tested show similar results for bottom pressure.
good agreement to the measurement for the both horizontal and ver- The detailed comparisons of pressures at representative gages provide
tical forces for the broken wave condition. Even though the frontal the complete temporal pattern of pressure distributions and indicate the
pressures show a higher RJ at the bottommost pressure gages for broken sensitivity of pressure outputs to different mesh sizes. However, it is
wave condition, both IHFOAM and FLUENT results show a relatively difficult to compare outputs at all pressure gages in the experiment.
smaller RJ for the total horizontal force. Therefore, the conclusion we have made here may not represent all
pressure characteristics of mesh dependence, and they could generate
biased results due to systematic errors, for example, at pressure gages
5.2. Sensitivity of mesh size (IHFOAM) located lower on the structure. For better quantitative comparison of the
sensitivity of mesh sizes, we apply the ratio of residual impulse (RJ) to
The three mesh resolutions (M1, M2, and M3) described in Section 3 evaluate mesh sensitivity. Fig. 25 shows RJ of the three mesh conditions
were tested with IHFOAM for the three wave conditions to evaluate the determined for (a) non-breaking, (b) breaking, and (c) broken wave
sensitivity of outputs of water surface elevation, pressure and forces to conditions. Each red circle, blue triangle, and black square represents the
mesh size. Fig. 23 shows the comparison of water surface elevation re- RJ from M1, M2, and M3.
sults at two representative wave gages. Fig. 23a and b show the com- For the non-breaking wave condition (Fig. 25a), RJ for all pressure
parison of results at wg2 and uswg2 for the non-breaking wave condition, gages is less than 0.01. Overall, there are minor differences among three
Fig. 23c and d show the comparison of results at wg2 and uswg2 for the mesh conditions, with M3 showing a slightly smaller RJ for all pressures
breaking wave condition, and Fig. 23e and f show the comparison of and horizontal force. M3 results clearly show the lowest RJ for all
results at wg2 and uswg2 for the broken wave condition. Line colors bottom pressures, but not for the vertical force. For the breaking wave
indicate results of the measurement (solid black), M1 (red dash dot), M2 condition, the overall RJ is less than 0.015. M2 and M3 results show
(blue dot), and M3 (black dash) results. Generally, the peak and shape of similar RJ values while M1 results show higher RJ for front pressures.
each mesh condition show similar results among the three mesh condi- All mesh conditions show a similar RJ for the horizontal force. As in the
tions, while M1 shows relatively smaller wave heights compared with the non-breaking condition, M3 results clearly show a smaller RJ for bottom
other two. Relatively small differences between modelled and measured pressures, but not for vertical force. Broken wave trials resulted in a
water surface elevations are observed for all three mesh conditions at much higher RJ than other wave conditions at pressure gages but not at
the load cell. To be specific, the difference between the M1 and M3
condition is a factor of 1.7 at press2. We can also observe a decrease of
RJ with elevation above the base of the structure (from press1 to press9)
for all mesh conditions. This reduction makes sense in the context of the
major contact area of bore type waves on the specimen. The deviation
observed for the broken wave condition is especially large for coarser
mesh (M1) simulations, which could result from the difficulty in
simulating the same conditions of air entrainment and water and air
density at the time of wave impact. Although there are relatively higher
RJ values for front pressures, the ratios of residual impulses for the
horizontal force for all three meshes are quite small. Bottom pressures
show a similar pattern to front pressures. Finer mesh results in a smaller
RJ, and a higher RJ is observed at press11, with a sharp decrease moving
inland (from press11 to press14). In general, the horizontal and vertical
forces are less sensitive to mesh size for all wave conditions, while front
and bottom pressures show the significant sensitivity to mesh size,
especially for the broken wave condition. This result is meaningful
Fig. 22. Quantitative FLUENT comparison of the ratio residual impulse for
when considering CFD modeling for the structural design, because the
pressures and forces for the three wave breaking conditions. Each red circle,
blue square, and black triangle indicates the results of non-breaking, breaking total forces are less sensitive to mesh and wave conditions than pressure
and broken wave conditions, respectively. (For interpretation of the references gages. Thus, significant computational expense may be saved by using a
to color in this figure legend, the reader is referred to the Web version of coarser mesh resolution in a CFD model while still achieving a reliable
this article.) estimate of the total force.

94
H. Park et al. Ocean Engineering 158 (2018) 78–98

Fig. 23. Surface elevation comparison at (a) wg2 and (b) uswg2 for non-breaking wave condition, (c) wg2 and (d) uswg2 for breaking wave condition, and (e) wg2
and (f) uswg2 for broken wave condition. Line colors indicate results of the measurement (solid black), M1 (red dash dot), M2 (blue dot), and M3 (black dash). (For
interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 24. Mesh sensitivity at press2 and press12 for three wave conditions. Each color indicates the results of measurement (solid black), M1 (red dash dot), M2 (blue
dot), and M3 (black dash). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

95
H. Park et al. Ocean Engineering 158 (2018) 78–98

both models showed better agreement to the measured data despite

calculated pressures showing relatively high deviations from the
short peaky pressures recorded in the measurement.
3. Front pressures and horizontal forces were sensitive to wave
breaking types, while bottom pressures and vertical forces were
less sensitive. This result highlights that numerical model bench-
mark tests for wave loads on an elevated structure should include
both horizontal force (frontal pressures) and vertical force (bottom
pressures).
4. In general, as one would anticipate, a finer mesh provided
improved numerical simulation results, and non-breaking and
breaking wave impacts showed minor sensitivity to mesh size.
Broken wave results are significantly sensitive to the mesh condi-
tions (IHFOAM).
5. The horizontal and vertical force were less sensitive to mesh size for
all wave conditions. This result is meaningful when considering CFD
modeling for structural design, because the total force is less sensitive
to mesh and wave conditions than the pressure gages. Significant
computational expense may be saved by using a coarser mesh reso-
lution in a CFD model while still achieving a reliable estimate of the
total force

As noted earlier, we applied the same turbulence coefficient for the

κ-ε turbulence model in both IHFAOM and FLUENT for model consis-
tency. In the IHFOAM model, we also tested the simulation using the
κω SST turbulence model (Wilcox, 2008) and compared the two model
results. Both turbulence models generate slightly different aspects of
wave deformation during propagation over the slope, but yield signif-
icant differences in pressure and force calculations. The difference in
results from the respective turbulence models was similar to that
observed between IHFOAM and FLUENT using the κ-ε model. When
using κω SST in IHFOAM, the simulated wave tends to shoal and break
further offshore over the sloped bathymetry. Consequently, κω SST
results slightly overestimated pressures and forces for non-breaking
waves, and underestimated the measurement for both breaking and
broken wave conditions. However, as observed in Section 4, the results
of the κ-ε model show good agreement with non-breaking wave--
induced pressures and forces while tending to underestimate those
produced by breaking waves and overestimate those produced by
broken waves.
The Courant number (Co) is also a significant variable for model
stability and computation time. We set a constant Co ¼ 0.5 as the
default for both IHFOAM and FLUENT simulations, but evaluate the
sensitivity of Co by applying various Co such as Co ¼ 1.0, 0.5, 0.25, and
Fig. 25. Comparison of IHFOAM with three mesh conditions for pressures and 0.1 for M2 in IHFOAM. The results show that the computation time
forces conditioned on (a) non-breaking, (b) breaking, and broken (c)
increased as Co decreased. This is expected, but there are not even
wave conditions.
minor changes of surface elevation, pressure and forces for all three
wave conditions in IHFOAM. Therefore, the computation time step
6. Conclusions and discussion utilized in the model (Co  1.0) was small enough to achieve the
highest resolution for the current model set up. This result implies that
The validation of two CFD models for the wave-induced force on an other disagreements between modeled and measured data such as the
elevated structure were performed, including quantitative compari- negative pressure (Fig. 9) and short peaky pressures at bottom face of
sons of pressures and forces to experimentally measured data. We the specimen (e.g. Fig. 10) were not induced from computation time
simulated IHFOAM and FLUENT for three regular wave conditions parameters. As a future work, we will evaluate the effects of finer mesh
which generated non-breaking, breaking, and broken wave impacts on (~0.01 m) and different turbulence models (e.g. LES) for model
the structure. The following general conclusions were found from this validations.
study:
Acknowledgment
1. The performance of numerical models varied depending on the wave
condition. For the same CFD model set up, non-breaking wave con- This study is based upon work supported by the National Science
ditions generally showed the best agreement for both FLUENT and Foundation under Award Number 1519679. Any opinions, findings, and
IHFOAM results, and the broken wave conditions showed the worst conclusions or recommendations expressed in this material are those of
agreement. the authors and do not necessarily reflect the views of the National Sci-
2. Generally, the free surface comparisons were excellent for both the ence Foundation. Finally, the authors would like to address appreciation
IHFOAM and FLUENT models, and horizontal and vertical forces of to reviewers.

96
H. Park et al. Ocean Engineering 158 (2018) 78–98

Nomenclature

Symbol Descriptions Unit

Co Courant number –
F External Force MLT2
FH Horizontal force MLT2
FV Vertical force MLT2
g Gravity acceleration LT2
H Wave height L
h Water depth L
JR Impulse of residual force or pressure MLT1 or ML1T1
JX Impulse of force or pressure MLT1 or ML1T1
L Wave length L
P Pressure ML1T2
p* Pseudo pressure ML1T2
R Residual of force or pressure ML1T2 or MLT2
RJ Ratio of Impulse –
T Wave period T
t Simulation time T
u Velocity (in OpenFOAM) LT1
v Velocity (in Fluent) LT1
α Quantity of water per unit of volume –
η Surface elevation L
μ Molecular viscosity ML1T1
μeff Efficient dynamic viscosity ML1T1
ν Kinematic viscosity of water L2T1
ρ Density ML3
ρwater Density of water ML3
ρair Density of air ML3
jj Absolute value –

Appendix

Fig. A1Frequency of calculation time step of each mesh conditions for X2 (non-breaking wave, IHFOAM).

References Cuomo, G., Tirindelli, M., Allsop, W., 2007. Wave-in-deck loads on exposed jetties. Coast.
Eng. 54 (No. 9), 657–679.
Cuomo, G., Shimosako, K., Takahashi, S., 2009. Wave-in-deck loads on coastal bridges
ANSYS, 2009. ANSYS Fluent Theory Guide. ANSYS, Inc., Canonsburg, PA, 12th release.
and the role of air. Coast. Eng. 56 (No. 8), 793–809.
ANSYS, 2013. ANSYS Fluent Tutorial Guide. ANSYS Inc., Canonsburg, PA.
Do, T., van de Lindt, J.W., Cox, D., 2016. Performance-based design methodology for
ASCE/SEI (ASCE/Structural Engineering Institute), 2016. Minimum Design Loads for
inundated elevated coastal structures subjected to wave load. Eng. Struct. 117,
Buildings and Other Structures. ASCE/SEI 7-16, Reston, VA.
250–262.
Bea, R.G., Xu, T., Stear, J., Ramos, R., 1999. Wave forces on decks of offshore platforms.
Emanuel, K., 2005. Increasing destructiveness of tropical cyclones over the past 30 years.
J. Waterw. Port, Coast. Ocean Eng. 125 (3), 136–144.
Nature 436 (7051), 686.
Bradner, C., Schumacher, T., Cox, D., Higgins, C., 2011. Experimental Setup for a large-
FEMA, 2011. Coastal Construction Manual: Principles and Practices of Planning, Siting,
scale bridge superstructure model subjected to waves. J. Waterw. Port, Coast. Ocean
Designing, Constructing, and Maintaining Residential Buildings. FEMA s.l.
Eng. 137 (1), 3–11.
Fluent INC, 2005, August, 18. Simulation of Wave Generation in a Tank. Retrieved from.
Bullock, G.N., Obhrai, C., Peregrine, D.H., Bredmose, H., 2007. Violent breaking wave
http://willem.engen.nl/uni/fluent/documents/external/wave.pdf.
impacts. Part 1: results from large-scale regular wave tests on vertical and sloping
walls. Coast. Eng. 54 (2007), 602–617.

97
H. Park et al. Ocean Engineering 158 (2018) 78–98

Goda, Y., 1974. New Wave Pressure Formulae for Composite Breakwater. Copenhagen. Launder, B.E., Sharma, B.I., 1974. Application of the energy-dissipation model of
ASCE, pp. 1702–1720. turbulence to the calculation of flow near a spinning disc. Lett. heat mass Transf. 1
Goda, Y., 2010. Random Seas and Design of Maritime Structures, third ed. World (2), 131–137.
Scientific. S.1. Linton, D., Gupta, R., Cox, D., van de Lindt, J., Oshnack, M.E., Clauson, M., 2012.
Hayatdavoodi, M., Seiffert, B., Ertekin, R.C., 2014. Experiments and computations of Evaluation of tsunami loads on wood-frame walls at full scale. J. Struct. Eng. 139 (8),
solitary-wave forces on a coastal-bridge deck. Part II: deck with girders. Coast. Eng. 1318–1325.
88, 210–228. Maza, M., Lara, J.L., Losada, I.J., 2015. Tsunami wave interaction with mangrove forests:
Hayatdavoodi, M., Seiffert, B., Ertekin, R.C., 2015. Experiments and calculations of a 3-D numerical approach. Coast. Eng. 98, 33–54.
cnoidal wave loads on a flat plate in shallow-water. J. Ocean Eng. Mar. Energy 1 (1), Mokrani, C., Abadie, S., 2016. Conditions for peak pressure stability in VOF simulations of
77–99. dam break flow impact. J. Fluids Struct. 62, 86–103.
Higuera, P., Lara, J.L., Losada, I.J., 2013. Realistic wave generation and active wave Mori, N., Shimura, T., Yasuda, T., Mase, H., 2013. Multi-model climate projections of
absorption for Navier–Stokes models: application to OpenFOAM®. Coast. Eng. 71, ocean surface variables under different climate scenarios—future change of waves,
102–118. sea level and wind. Ocean. Eng. 71, 122–129.
Higuera, P., Lara, J.L., Losada, I.J., 2014. Three-dimensional interaction of waves and Park, H., Tomiczek, T., Cox, D., van de Lindt, J.W., Lomonaco, P., 2017. Experimental
porous coastal structures using OpenFOAM®. Part I: formulation and validation. modeling of horizontal and vertical wave forces on an elevated coastal structure.
Coast. Eng. 83, 243–258. Coast. Eng. 128, 58–74.
IHFOAM, 2017. Physics. Retrieved from. http://ihfoam.ihcantabria.com/model/physics/ Seiffert, B.R., Hayatdavoodi, M., Ertekin, R.C., 2015. Experiments and calculations of
. cnoidal wave loads on a coastal-bridge deck with girders. Eur. J. Mech. B-Fluids 52,
Iturrioz, A., Guanche, R., Lara, J.L., Vidal, C., Losada, I.J., 2015. Validation of 191–205.
OpenFOAM® for oscillating water column three-dimensional modeling. Ocean. Eng. Wiebe, D.M., Park, H., Cox, D.T., 2014. Application of the Goda pressure formulae for
107, 222–236. horizontal wave loads on elevated structures. KSCE J. Civ. Eng. https://doi.org/
Jin, J., Meng, B., 2011. Computation of wave loads on the superstructures of coastal 10.1007/s12205-014-0175-1.
highway bridges. Ocean. Eng. 38. Wilcox, D.C., 2008. Formulation of the k–x turbulence model revisited. J. Am. Inst.
Kaplan, P., 1992. Wave impact forces on offshore structures: Re-examination and new Aeronautics Astronautics 46 (11), 2823–2838.
interpretations. Houston, Texas. In: Offshore Technology Conference, pp. 79–86. Wu, Y.L., Stewart, G., Chen, Y., Gullman-Strand, J., Lv, X., Kumar, P., 2016. A CFD
Kaplan, P., Murray, J.J., Yu, W.C., 1995. Theorectical analysis of wave impact forces on application of new wave theory to wave-in-deck simulation. Int. J. Comput. Methods
platform deck structures. In: Proceedings of the 14th International Conference on 13, 2.
Offshore Mechanics and Arctic Engineering, pp. 189–198. Xiao, H., Huang, W., 2008. Numerical modeling of wave runup and forces on an idealized
Kirkg€ oz, M.S., 1995. Breaking wave impact on vertical and sloping coastal structures. beachfront house. Ocean. Eng. 35 (1), 106–116.
Ocean. Eng. 22 (1), 35–48.

98