Journal of Coastal Research SI 113 651-656 Charlotte, North Carolina 2024

CFD Modelling of a Tsunami Event

Curtis Graham† and Nagi Abdussamie‡*
† ‡
Centre for Maritime Engineering and Hydrodynamics College of Engineering and Technology
Australian Maritime College, University of Tasmania University of Doha for Science and
Launceston, Tasmania, Australia Technology, Doha, Qatar www.cerf-jcr.org

ABSTRACT

Graham, C. and Abdussamie, N., 2024. CFD modelling of a tsunami event. In: Phillips, M.R.; Al-Naemi, S.,
and Duarte, C.M. (eds.), Coastlines under Global Change: Proceedings from the International Coastal
Symposium (ICS) 2024 (Doha, Qatar). Journal of Coastal Research, Special Issue No. 113, pp. 651-656.
Charlotte (North Carolina), ISSN 0749-0208.
www.JCRonline.org This paper presents and discusses numerical results of a tsunami wave event and its interaction with a vertical,
surface-piercing cylinder by employing Computational Fluid Dynamics (CFD) code. Focusing on generating
tsunami-like wave events, common numerical tsunami approximation methods are investigated. A systematic
approach was adopted including the development of numerical wave tank and wave-structure interaction
analysis. The first step involves the development of a tsunami model using wave superposition, followed by
validation against observed data through numerical sensitivity studies. Subsequently, the model is employed to
simulate tsunami interactions with a fixed and a heaving cylinder. The comparison of the developed tsunami
model results with solitary and Stokes 5th order approximations reveals conservative but inadequate
representations of tsunami forces by both methods. This underscores the limitations of current approximations
in capturing force magnitude and direction, emphasizing the need for improved numerical modelling
techniques in tsunami analysis.

ADDITIONAL INDEX WORDS: Natural hazards, tsunami, numerical modelling, wave superposition.

INTRODUCTION Wave Tank (NWT) will simulate the problem dynamics,

With increasing concerns over coastal protection and climate providing insights into tsunami generation, propagation, and
change impacts, accurately modeling of tsunami events is structural interactions. The results aim to enhance numerical
crucial (Marras & Mandli, 2020; Nouri et al., 2010). The key analysis and industry practices, offering scalable models to
characteristics of tsunamis are their sheer force, velocity and justify design method selections and improve future design
unpredictability. Numerical analysis of tsunami propagation is criteria.
essential for designing coastal and offshore structures to The paper is structured as follows: First, the NWT
withstand these extreme metocean conditions throughout their development, including geometry, meshing, boundary
lifespan. conditions, and domains. Next, tsunami wave profile generation
Tsunami events are becoming increasingly frequent due to and propagation analysis through the NWT. Then, wave force
climate change, affecting regions like Japan and Canada. Several estimation on a fixed cylinder and analysis of a surface-piercing
notable tsunamis such as the 2011 Tohoku Tsunami event (Zhao cylinder with different wave types. Finally, response estimation
et al., 2019) and those occurred in the Douglas Channel, British of a heaving cylinder, detailing a one-degree-of-freedom (1DOF)
Columbia, caused by landslides in the Kitimat arm on October heave decay test and interactions with various waves.
17, 1974, and April 27, 1975 (Murty, 2003; Skvortsov &
Bornhold, 2007). These events produced wave heights of 6.1 m METHODS
to 8.2 m. Recent studies aimed to replicate such tsunamis to CFD Approach
estimate wave heights and runup (Byres & Szabo, 2007; Kirby In this work, the commercial Navier-Stokes, CFD code
et al., 2016; Murphy et al., 2016), though they did not consider STAR-CCM+, was used for simulating tsunami wave events and
tsunami-structure interactions. Accurate prediction of metocean their interaction with a vertical cylinder. Based on isothermal
loading conditions remains a significant challenge for designing and laminar flow assumptions, a system of partial differential
coastal and offshore infrastructure. equations governing the conservation of mass and momentum of
This study focuses on modeling a tsunami wave event and its a fluid was solved numerically using finite volume method
interaction with a truncated cylinder using a Computational (Versteeg & Malalasekera, 2016). In addition, the free surface
Fluid Dynamics (CFD) techniques-based code. A Numerical equation and its motion were solved and captured using the
Volume of Fluid (VOF) model. This is a two-step approach
____________________
starting with the development of the NWT followed with
DOI: 10.2112/JCR-SI113-129.1 received 23 June 2024; accepted in
revision 30 July 2024. simulating the interaction between the generated tsunami event
*Corresponding author: nagi.abdussamie@udst.edu.qa with a fixed and a heaving cylinder. Verification and validation
©
Coastal Education and Research Foundation, Inc. 2024 were conducted to assess CFD predictions. The main steps
conducted are summarized below:

XX
 Graham, and Abdussamie
_________________________________________________________________________________________________

 Tsunami Generation: A tsunami event was generated cylinder surface. These refinements aimed to develop a finer
using the NWT, wave superposition, and the VOF mesh near the cylinder, especially at the free-surface-cylinder
model to replicate the characteristics of a realistic interaction, to provide a more accurate simulation in critical
tsunami event. The results were validated against a regions. The domain is similar to that in Figure 1, but during
target wave event with 8.2 m wave height. free decay tests, the domain size was increased to minimize
 Wave Forces Prediction: A vertical cylinder was domain interference and wake reflection.
placed in a 3D domain based on the developed
NWT. The cylinder was fixed, and hydrodynamic Table 1. Tsunami generation NWT trimmer sizes.
forces were predicted using three methods including Meshing parameter Relative cell size Cell Size (m)
Stokes 5th order, solitary, and the simulated (%)
tsunami event. Δx 6.25 0.3125
 Heave Motion Analysis: The cylinder was activated Δy 12.5 0.625
to heave only using the Dynamic Fluid Body Δz 6.25 0.3125
Interaction (DFBI) model. A free decay test was
conducted to estimate the cylinder's heave natural
period, verified against theoretical values. Heave
motions were then obtained using Stokes 5th order,
solitary, and the simulated tsunami event.

NWT Development
The first component describes the development of the
NWT
numerical wave tank (NWT), which undergoes several iterations
and refinements, each justified and detailed. To assess the
NWT's geometry, meshing conditions, boundary conditions, and
domain interactions, a simple NWT was created, validated with
regular wave analysis.
The tsunami profile was generated in a long, narrow NWT, 1
m wide, with a 150 m wave propagation area and a 600 m
damping area to prevent reflected waves. The NWT was
designed with slip walls to minimize computational
requirements, maintaining consistent wave propagation. The
water depth was set at 54 m, reflecting the approximate depth at
the expected impact site. The cylinder had an 18.9 m diameter, Local surface mesh at the cylinder
18 m draft, 30 m height, a center of gravity 0.42 m above the
waterline, and a mass moment of inertia of 2.99 x 109 kg.m2 in
roll/pitch and 2.25 x 108 kg.m2 in yaw. See the schematic in
Figure 1.
Figure 1. NWT and mesh refinement details.
Mesh Generation and Optimization
For both the NWT and cylinder, the mesh was optimized
using mesh refinement domains, which do not interfere with the Tsunami Profile Generation
simulation but provide a reference for the mesh. Two volumetric To produce accurate results for a full-scale CFD tsunami
refinements were added to the main background mesh of the analysis, a realistic wave profile must be defined. Common
NWT domain. The first was a free surface volumetric industry practice uses approximate Stokes or solitary waves, but
refinement, reducing the target mesh size at the tank's free a realistic profile yields more accurate results. Following
surface. The second, the overset refinement, provided a experimental research by (Perez del Postigo Prieto et al., 2019)
transitional mesh between the background and overset mesh by and referencing typical tsunami behavior from (Zhao et al.,
defining a custom anisotropic trimmer size in each axis, 2019), a wave profile for a submarine landslide-induced tsunami
reducing the target and minimum mesh size to a small was determined. The data set, graphed as surface elevation vs.
percentage of the base size (O'Connor et al., 2023). A base cell time, was scaled to replicate an 8.2 m tsunami event, similar to
size of 5.0 m was used throughout the domain, except in the free the 1975 tsunami in the Douglas Channel, British Columbia,
surface zone, where a fine mesh was applied, as summarized in Canada. The original 40-minute time record from (Zhao et al.,
Table 1. 2019) was reduced to 140 s to minimize computational time,
For the mesh around the cylinder, three refinements were justified by the study's aim to simulate tsunami-like events in
introduced. The first refined the free surface zone. The second, shallow water and investigate their interaction with structures.
the overset mesh refinement, allowed the highly refined internal To input the data into the CFD code, the wave profile was
mesh to transition into the coarser background mesh. The third replicated using wave superposition. Five regular waves were
refinement decreased the target size of the cells around the superimposed to create the wave profile shown in Figure 2. In

Journal of Coastal Research, Special Issue No. 113, 2024

XX
 CFD Modelling of a Tsunami Event
_________________________________________________________________________________________________

this figure, the free surface elevation in meters is plotted against

time, showing the 8.2 m tsunami and the five superimposed
waves in dashed lines. The wave profile is disregarded beyond
140 s, where it becomes unstable and irrelevant. Realistically,
the free surface would oscillate about the original area without
exceeding the maximum tsunami wave height.

Figure 3. Numerical tsunami wave elevation at incident, 50 m

and 150 m wave probes. WP = Wave Probe

Figure 2. Tsunami wave profile generated using wave Wave Forces on a Fixed Cylinder
superposition. In industry practice, tsunami modeling often approximates
tsunamis as simplified wave types like Stokes 5th order or
solitary waves. These methods, however, may not accurately
CFD simulations were performed using a 2nd order implicit represent the force dynamics of a tsunami, which differs
unsteady solver with a time step of 0.01 s and five iterations per significantly due to its unique wave profile. To assess the
time step. The time step and mesh sizes were refined to keep the discrepancies of industry methods with a tsunami model, the
Courant number below 1 and as low as possible. The Courant results obtained through the numerical tsunami model will be
number, representing the percentage of a cell that fluid passes compared with both a Stokes 5th regular wave of equal height
through in a given time step, was maintained at a maximum with an associated period equal to the maximum period of the
value of approximately 0.79, based on the minimum cell size tsunami (140 s) and a solitary wave, modelled using a cnoidal
and maximum fluid velocity (O'Connor et al., 2023). wave profile with an equal wave height to the tsunami and a
In the 150 m propagation area, three numerical wave probes period that represents the major peak in the tsunami profile as
(WP) were placed at 0, 50, and 150 m from the tsunami source. recommended by (Zhao et al., 2019), see Figure 5.
These probes measured the free surface elevation at various
distances to ensure the wave maintained the desired profile as it
approached the interaction area. The NWT domain arrangement
with the free surface mesh and boundary conditions is shown in
Figure 1.

RESULTS
Tsunami Profile Results
Surface elevation measurements were conducted using three
wave probes over a 150 m distance. Figure 3 shows the surface
elevation against time at the incident, 50 m, and 150 m wave
probes. The wave profiles closely align with the desired profile
from Figure 2, showing a minimal error of 0.27% in peak height
compared to the targeted points. Maximum peak deviations over
150 m were observed at just 0.01%.
The speed of the generated tsunami was determined to be Figure 4. CFD predictions of tsunami wave against the recorded
approximately 22.8-24.8 m/s (82-89 km/h), which corresponds event.
well with the theoretical estimation of 23 m/s (83 km/h) for
shallow water tsunamis. Figure 4 overlays the surface elevations
for the 140 s simulation, demonstrating the close match between Figure 6 illustrates the wave forces acting on the fixed
the CFD and input tsunami profiles. cylinder, including horizontal force (top), vertical force (middle),
Overall, the tsunami wave profile developed in the NWT is and resultant force (bottom). Table 2 summarizes the force
validated, as the surface elevation closely aligns with the desired results from different wave profiles: tsunami, solitary wave, and
values. Using this tsunami generation method via wave Stokes 5th order wave. The peak wave heights align closely
superposition, the tsunami can be numerically tested against across profiles, with negligible errors (within 1%), ensuring
coastal and offshore structures. comparable maximum force results (Figure 5). However,

Journal of Coastal Research, Special Issue No. 113, 2024

XX
 Graham, and Abdussamie
_________________________________________________________________________________________________

significant variations were observed in maximum horizontal

forces (Fx). The solitary wave induced a 31% increase, while the
Stokes wave showed a 97% decrease compared to the tsunami
profile. This disparity is attributed to differences in wave
steepness and water level variations around the cylinder. Despite
these variations, resultant forces remained similar, with
increases of 4.79% and 1.85% for the solitary and Stokes waves,
respectively.

Figure 5. CFD predictions of tsunami wave against the

recorded event.

For the vertical force, the peak Fz values are relatively

comparable with an increase of 6.93% and 0.94% from the
solitary and Stokes profiles respectively. The obtained results
suggest that the solitary and Stokes profiles do provide an
approximate peak Fz and peak resultant force which may be
sufficient, scenario depending. However, it must also be noted
that the maximum resultant force value should not be used as the
main comparative factor, as it is only a magnitude and
disregards the direction in which the resultant force acts.
Since the wave periods and under curve areas (representing
quantity of water mass) vary significantly, the wave profiles and
the force time history were integrated with respect to time to
determine the average wave height and resultant force imparted
on the cylinder during the event. The results of which can be
seen in Table 2. When compared with the tsunami, the solitary Figure 6. Wave forces acting on the cylinder: Horizontal force
and Stokes profiles show an increase of 13% and 114% (top); vertical force (middle) and resultant force (bottom).
respectively. Due to the wave profile itself, this trend does not
continue regarding average resultant force. The average resultant
force applied to the cylinder during the solitary and Stokes Table 2: Force results on the fixed cylinder due to tsunami
waves is increased by 126% and 190% respectively, compared modelled by different methods.
to the 2387 kN average applied during the tsunami. These Wave Hpeak Max Max Max HAvg Res. F
results, unlike the peak forces suggest that the solitary and Profile (m) Fz Fx Res. F (m) Avg (kN)
Stokes approximations produce results that are generally (kN) (kN) (kN)
incomparable as the disparity in the average results is significant.
Tsunami 8.241 10521 2231 10754 2.442 2387

Response of a Heaving Cylinder Solitary 8.281 11122 2922 11499 2.762 5389
The natural period of the cylinder in heave was predicted by Stokes 8.203 10855 69 10855 5.215 6926
CFD as 9.915 s versus the theoretical value of 9.89 s (<1% 5th
error). The response of a heaving cylinder to tsunami and other
solitary waves. Table 3 details the peak wave height (H*) and
wave types was analyzed using similar methodology and NWT
cylinder heave (Z*) for each profile, alongside average values
setup as the fixed cylinder. Figure 7 illustrates the heave
over the given periods. Interestingly, the response amplitude
response of the cylinder to Stokes 5th order, tsunami, and
operator in heave (Z*/H*) was found to be approximately 1.0 for
the simulated tsunami-like wave case. Wave buoys located near

Journal of Coastal Research, Special Issue No. 113, 2024

XX
 CFD Modelling of a Tsunami Event
_________________________________________________________________________________________________

and offshore can be used as a remote sensing system for height and speed, underscores the accuracy and validation of
environment and extreme weather events provided that their the CFD model. However, limitations such as idealized
structural integrity is maintained during such events. Results bathymetry and computational constraints may affect
indicate that the cylinder's heave varies significantly depending localized impact predictions.
on the wave steepness. The maximum heave during the solitary These results hold significant implications for coastal
wave was 19.0% higher than during the tsunami, while the management and disaster preparedness. The validated CFD
Stokes wave produced a 2.25% increase in maximum heave. By model provides a robust tool for predicting tsunami impacts,
integrating the under-curve areas, the wave height average and aiding in the development of resilient coastal infrastructure
the cylinder heave average can be determined. The wave height and effective mitigation strategies. Future research should
averages are greatly similar to the fixed cylinder analysis. The focus on improving model resolution and incorporating
average heave magnitudes show correlating numbers based on additional physical processes to enhance prediction accuracy
the average wave height increase, with the solitary and Stokes and reliability.
waves producing average heave increases of 40.7% and 131% CONCLUSIONS
over the 2.145 m average during the tsunami testing. This study employed a two-step CFD process to generate a
tsunami-like wave event with 8.2 m in height. The first step
involved developing a tsunami model using wave superposition,
validated against targeted profile. The second step simulated
tsunami interactions with a vertical, surface-piercing cylinder.
Key findings include:

 The tsunami wave profile closely matched the

characteristics of the targeted wave profile, with a
minimal error of 0.27% in peak height and travel
speeds aligning well with theoretical predictions.
 Solitary and Stokes 5th wave approximations were
found to inadequately represent the forces induced
by tsunamis, highlighting the need for improved
modeling techniques.
 A realistic tsunami profile based on recorded data
was developed using wave superposition, providing
a reliable method for future numerical analysis in
various CFD software packages.

These findings enhance the accuracy and reliability of

numerical tsunami modeling, contributing to better coastal
Figure 7. Heave response of the cylinder to Stokes 5th order, protection strategies. The wave superposition method offers a
tsunami and solitary waves. robust alternative to traditional approximation methods,
promising more precise predictions of tsunami impacts.
Table 3: Results from Full-Scale Heaving Cylinder During ACKNOWLEDGMENTS
Tsunami, Stokes 5th and Solitary Waves. The authors are grateful to the University of Doha for Science
Wave H* (m) Z* (m) Havg (m) Zavg (m) and Technology (UDST) and the Centre for Maritime
Profile Engineering and Hydrodynamics of the Australian Maritime
Tsunami 8.230 8.166 2.444 2.145 College (AMC), University of Tasmania (UTAS) for facilitating
Solitary 8.417 9.504 2.791 3.018 this study.
Stokes 5th 8.203 7.986 5.215 4.959
LITERATURE CITED
DISCUSSION Byres, R., & Szabo, V. (2007). Recent Oil and Gas Port Projects
The CFD model effectively generated the 8.2 m tsunami in Northern British Columbia. In Ports 2007: 30 Years
event (a tsunami-like wave event occurred in April 27, 1975, of Sharing Ideas: 1977-2007 (pp. 1-10).
British Columbia) using wave superposition techniques. The Kirby, J. T., Shi, F., Nicolsky, D., & Misra, S. (2016). The 27
simulated wave profiles closely matched the desired profile, April 1975 Kitimat, British Columbia, submarine
with a minimal error of 0.27% in peak height. The estimated landslide tsunami: a comparison of modeling
speed of the tsunami (22.8-24.8 m/s) aligned well with approaches. Landslides, 13, 1421-1434.
theoretical calculations, (g/h)0.5 = 23 m/s, in which h = 54 m, Marras, S., & Mandli, K. T. (2020). Modeling and simulation of
confirming the model's reliability in dynamic simulations. tsunami impact: a short review of recent advances and
The close alignment between simulated and targeted future challenges. Geosciences, 11(1), 5.
tsunami profiles, along with minimal deviations in peak

Journal of Coastal Research, Special Issue No. 113, 2024

XX
 Graham, and Abdussamie
_________________________________________________________________________________________________

Murphy, E., Osborne, P., & Clohan, D. (2016). Water Level and Skvortsov, A., & Bornhold, B. (2007). Numerical simulation of
Tsunami Design Criteria for LNG Marine Terminals the landslide‐generated tsunami in Kitimat Arm,
in BC. In Ports 2016 (pp. 119-129). British Columbia, Canada, 27 April 1975. Journal of
Murty, T. (2003). A Review of Some Tsunamis in Canada. Geophysical Research: Earth Surface, 112(F2).
Submarine Landslides and Tsunamis, 175-183. Versteeg, H., & Malalasekera, W. (2016). An Introduction to
Nouri, Y., Nistor, I., Palermo, D., & Cornett, A. (2010). Computational Fluid Dynamics. 2007. Edinburgh,
Experimental investigation of tsunami impact on free England: Pearson Education Limited.
standing structures. Coastal Engineering Journal, Zhao, E., Sun, J., Jiang, H., & Mu, L. (2019). Numerical Study
52(1), 43-70. on the Hydrodynamic Characteristics and Responses
O'Connor, C., Mohajernasab, S., & Abdussamie, N. (2023). of Moored Floating Marine Cylinders Under Real-
Numerical investigation into water entry problems of a World Tsunami-Like Waves.
flat plate with air pockets. Journal of Ocean
Engineering and Science, 8(4), 386-400.
Perez del Postigo Prieto, N., Raby, A., Whittaker, C., &
Boulton, S. J. (2019). Parametric study of tsunamis
generated by earthquakes and landslides. Journal of
Marine Science and Engineering, 7(5), 154.

Journal of Coastal Research, Special Issue No. 113, 2024

XX