Prediction of Slamming Loads on Catamaran Wetdeck using CFD

Ahmed Swidan1, Giles Thomas2, Dev Ranmuthugala1, Irene Penesis1, Walid Amin1, Tom Allen3, Mark Battley3.
1.
Australian Maritime College, University of Tasmania, Australia
2.
University College London, UK
3.
Centre for Advanced Composite Materials, University of Auckland, New Zealand

ABSTRACT
Wetdeck slamming is one of the principal hydrodynamic loads acting on catamarans. CFD techniques are shown to successfully
characterise wetdeck slamming loads, as validated through a series of controlled-speed drop tests on a three-dimensional catamaran
hullform model. Simulation of water entry at constant speed by applying a fixed grid method was found to be more computationally
efficient than applying an overset grid. However, the overset grid method for implementing the exact transient velocity profile resulted
in better prediction of slam force magnitude. In addition the splitting force concurrent with wetdeck slam event was quantified to be
21% of the vertical slamming force.

KEY WORDS
Catamaran; Slam; Drop-test; CFD; Water-entry.

INTRODUCTION
A major challenge in structural optimisation of high-speed
catamarans is the accurate prediction of sea loads imparted on
such vessels. The most important hydrodynamic loads
experienced by large high-speed catamarans are generally
referred to as wetdeck slamming loads. This sudden
hydrodynamic load, which occurs mainly on the cross deck
structure between demihulls (see Fig. 1) due to wave impacts,
can result in structural failure (Faltinsen (2012), Thomas et al.
(2002)). For high-speed catamarans the majority of wetdeck
slamming loads are located in the vicinity of the centrebow,
predominantly in the archway between the centrebow and the
demihull as shown in Fig. 1 (Thomas et al. (2011)). Accurate
prediction of wetdeck slamming loads in this particular vicinity
of high-speed catamarans is necessary for efficient structural
design as well as better seakeeping.
Past experience has shown that Computational Fluid Dynamics
(CFD) is one of most efficient numerical tools in predicting
quasi-two dimensional (2-D) slamming loads, (Swidan et al.
(2014), Lewis et al. (2010)). However to date CFD has not been
used to investigate three-dimensional effects in slamming which
are known to be significant (French and Thomas (2014)).
There are several available methods for assessing the magnitude
of slamming loads. Whilst performing measurements in fullscale trials can provide valuable data, it is generally unsuitable
for the validation of numerical results due to the high
uncertainty in isolating actual slam loads (Jacobi et al. (2012)).
An efficient method for obtaining a high-quality validation
dataset of slamming loads and pressures is drop test
experiments. However, currently the only available data for
catamaran drop tests are restricted to 2-D models impacting
water under gravitational force, such as the 2-D free-falling drop
test conducted by Davis and Whelan (2007).

Swidan

Fig. 1: Arched wetdeck with centrebow of 98 m wave-piercer

catamaran (Incat (2015)).
The present work investigates the use of CFD to model threedimensional (3-D) wetdeck slamming loads acting on a
catamaran hullform model during water-entry on an initially
calm free surface. The slamming pressures and loads were
predicted for a single constant speed using two CFD simulation
techniques: the overset grid method and the traditional fixed
grid method. The CFD predictions were validated through
results from drop test experiment using the Servo-hydraulic
Slam Testing System (SSTS) (Battley and Allen (2012));
applying the exact velocity profile measured during the drop test
experiments to the numerical simulations.

NUMERICAL METHOD
The numerical simulations were executed using the Reynolds
Averaged Navier Stokes (RANS) solver STAR-CCM+. The
code resolves the incompressible RANS equation in integral
form utilising the finite volume method. The motions of the
model were restricted in all degrees of freedom except in the
vertical direction. To validate the computed results against
experimental data, simulations were performed using the
velocity profile measured during the experiment.

Prediction of Slamming Loads on catamaran wetdeck using CFD

Symmetry
plane
Test model

The overset grid method was originally developed by Starius

(1977). The numerical domain consists of two main grids, one
around the moving body that follows the time history of the
body motion referred to as the overset grid, and another
stationary grid representing the total domain, called the
background grid. These two grids interpolate information within
the overlapping zone where donor and acceptor cells intersect.
Linear interpolation was utilised among each moving acceptor
cell centroid and the adjacent four donor cells centroids for this
3-D case (for more details Swidan et al. (2013)).
The main particulars of the test model shown in Fig. 2-a are:
length (L) 500 mm, beam (B) 638 mm, height (H) 327.6 mm and
total mass 14.8 kg, while the tank geometry is shown in Fig.2-b.
Five pressure sensing locations, namely; P1 to P5, were used to
capture the pressure distribution along the archway of the
wetdeck both in the simulations and in the experiments, the
latter through pressure sensors installed on the physical model
(see Fig. 2-a).
The lowest point located on the demi-hull keel of the model was
set as the origin of the earth-fixed global coordinate system. The
boundary conditions used were velocity inlet at the bottom of
the tank, pressure outlet at the top of the tank, non-slip wall at
the hull surface of the model and tank wall, while the plane that
bisects the catamaran model vertically was treated as a
symmetry plane, as shown in Fig. 2-b. No damping boundaries
were employed to save computational time, since the slamming
event occurs over a very short time there will be little or no
influence of reflected free-surface waves.

Fig. 2: Overview of generic catamaran hullform model, (a)

Geometry with five pressure sensors locations, (b)
Computational domain and (c) Generated mesh.
The water-entry process can be simulated using various CFD
techniques. In the present work a comparison was made
between two numerical methods:

allowing the body to move relative to the initially

stationary fluid using overset grid method within
STAR-CCM+; and

applying an inlet flow at a given velocity with the body

remaining stationary using the traditional fixed grid
method.
The computed results using both methods were compared
against the drop-test experimental data.

Swidan

A slow growth rate was selected for mesh generation to ensure

smooth transition between neighboring cells. The grid was
generated with seven density levels in order to capture pressure
distribution and related peaks around the sensors and refine the
mesh around the hull shape. The finite volume meshing method
was implemented for all simulations to solve the differential
equations for a viscous 3-D multiphase flow represented by
RANS equations in which the effect of turbulence is solved
using two equation Shear Stress Transport (SST) k- eddy
viscosity model.
To resolve the near wall turbulence quantities all y+ wall
treatment was selected that attempts to combine the high y+
wall treatment for coarse meshes and low y+ treatment for fine
meshes, where y+ is a non-dimensional wall distance. This
turbulence model was selected based on previous experience by
the authors in solving water-entry problems (Swidan et al.
(2014)) and since it has the advantages of the k- model near
the boundary layer and retaining the insensitivity of k- in the
far field (Menter (1994)).
The Volume of Fluid (VOF) formulation, accounts for the free
surface between the air and water fluid mixture and the possible
complex splash-up during the water-entry phase. In this method,
the assumed incompressible water and air interface (freesurface) is formulated on the basis of a single-fluid with
discontinuous physical properties based on a phase fractional

Prediction of Slamming Loads on catamaran wetdeck using CFD

function according to the proportion of each fluid. Therefore,

the same set of basic governing equations describing
momentum, mass, and energy transport in a single phase flow is
solved. The density of each cell that goes to the momentum
equation has a single time dependent value that is based on the
fractional density of the mixed fluids per each cell.
Surface tension is not important in the water-entry problem but
it can lead to more accuracy in capturing the water splash-up. In
order to avoid smearing of water-air interactions careful
attention was be paid to discretising the grid in the vicinity of
free-surface and water jet formation. To preserve the sharpness
of the interface between the water and air, the High Resolution
Interface Capturing (HRIC) scheme is used for the discretisation
of the non-linear convective term in the momentum equation of
the volume fraction (Wacawczyk and Koronowicz (2008)).
This scheme is activated by default in the code for Courant
numbers1 less than 0.5.
The applied physical model in all simulations uses the Eulerian
multiphase segregated iterative method to solve the conservation
equations for mass, momentum and energy (RANS equations)
for each phase. This model solves the flow equations for the
velocity components and pressure in an un-coupled manner.
First, the linearized components of the momentum equations are
the prevailing pressure and mass fluxes through the control
volume faces (inner-iterations), followed by a Semi-Implicit
Method for Pressure-Linked Equations (SIMPLE) to resolve the
pressure-velocity coupling, while the linkage between the
momentum and continuity equations is achieved through
predictor and corrector stages.
For the temporal integration an implicit Euler scheme was
applied for maximum stability (Dragomir et al. (1998)). For the
case of overset mesh the iteration loop is extended to update the
position of the overset region. The governing motion equation in
the vertical direction of the rigid body is then solved to compute
the new body position and adapt the corresponding overset grid
and update the boundary conditions as well.
Grid Independence Study
To study the sensitivity of computed results against the used
sizes of cells to discretise the domain and the 3-D geometry of
the catamaran hull model, grid-density independence studies

and time independence studies were conducted, as shown in Fig.

3 and Table 1.
The total number of generated cells varies from around 0.7
million, 1.2 million and 2.3 million cells for coarse, medium and
fine fixed grids, while for overset grids from 0.8 million, 1.4
million to 2.5 million cells for coarse, medium and fine overset
grids respectively. The slight increase in the applied total cell
densities of overset grids were due to the extended refinement of
the overlapping region, where the moving overset mesh
interpolate with the fixed grid.
Fig. 3 illustrates the computed vertical force against time for
three levels of mesh refinement where 0 time corresponds to the
instant where the demihull keel touches the initially calm water
surface. Fig. 3-a presents the results of using the fixed grid
method while Fig. 3-b shows the computed results employing
the overset grid method, given that the same computational grid
refinements around the hull have been utilised for both the
stationary grid and the overset grid methods.
Table 1 provides a summary of the independence grid studies
and associated uncertainties against the measured data. The first
column presents the grid size, the second the applied Courant
number and the rest of table compares between the applied
numerical methods; namely, overset method and fixed grid
method. The comparison conducted is regarding three main
parameters, as following; t% refers to the percent of deviation
between numerical and experimental results regarding the time
duration as the total slam force peak, while Fmax% is the
difference between the computed slam force peaks and the
experimental data in percentage. The simulation time has also
been stated in Table 1. The simulations were performed using
12 nodes of the high-performance computing cluster (consisting
of 648 cores and 132 nodes), at the Australian Maritime
College.
It was decided to use the medium sized meshes for the presented
results on the basis of the slight change in timing of slamming
peak force with respect to the applied coarse stationary grid, as
shown in Fig. 3-a and as presented in Table 1. The chosen timestep varies, in order to satisfy a Courant number of a range from
0.1 to 0.5 in the same domain.

Table 1: Summary of numerical independence study with associated uncertainty against experimental data
Courant
Overset Grid
Fixed Grid
No1.
t %
Fmax %
Simulation t % Fmax %
Simulation
time (hr.)
time (hr.)
Coarse
0.1
9.9% 0.6
11
6%
5
8.5
Grid
0.5
9.9% 0.6
10
6%
5
8
Medium 0.1
10%
0.5
17
5%
4.8
14
Grid
0.5
10 % 0.5
14.5
5%
4.8
12
Fine
0.1
10%
0.5
30
4.9% 4.8
26
Grid
0.5
10%
0.5
28
5%
4.8
23.5
1

Courant number = velocity (m/s) x time step (s)/cell size (m)

Swidan

Prediction of Slamming Loads on catamaran wetdeck using CFD

Table 2: Instrumentation Details

No. of
Gauge
Manufacturer
Channels
Precision
Load cell
3
Transducers
Pressure
PCB
5
Transducers
Piezotronics
Position
1
Vishay
sensor

Model
LPC
5t
113B26
REC
139L

Maximum
Range
5000 kg
68950
kNm-2
3m

The test model was sized to ensure that there would be a gap
between the model and the tank wall of double the models
overall beam. This was to minimise boundary condition effects
and the possibility of wave reflections, as well as to facilitate
replication of the experimental set-up in numerical simulations.

Fig. 3: Computed vertical slamming forces by applying three

different grid sizes for, (a) fixed grid and (b) for the overset
grid.

EXPERIMENTAL SETUP
To provide high-quality experimental data suitable for
validation purposes a series of drop-test experiments was
conducted using the Servo-hydraulic Slam Testing System
(SSTS), at Industrial Research Limited, Auckland, New Zealand
(Battley and Allen (2012)).
Fig. 4 illustrates the main mechanical components of the SSTS.
The hydraulic system can achieve a range of controlled waterentry velocities up to 10 m/s, with the required hydraulic power
for each target velocity controlled by a servo-proportional
control valve. This advanced servo-valve controls the hydraulic
power based on an electronic feedback signal from the rig
position transducer. Details of the used instrumentation on the
test rig are given in Table 2.

Fig. 4: Schematic of test installation, at the Industrial Research

Limited Auckland, New Zealand

Swidan

RESULTS AND DISCUSSION

Table 3 presents a summary of the simulation conditions as well
as the principal results. The measured velocity profile during the
drop test (at a target velocity of 4 m/s) was implemented for the
CFD simulations. Fig. 5 presents the time histories of the
measured immersion (on the left axis) and velocity (on the right
axis) of the physical model, where the 0 timing corresponds to
the demihull keel touching the free-surface.
In general the model enters the water at an approximate velocity
of 3.85 m/s. The slight velocity reduction at around 6 ms
corresponds to an immersion of 23.5 mm where the centrebow
keel touches the water free-surface. At 36 ms, when the velocity
again reduces, the model is immersed by around 140 mm, which
corresponds to the archway of the wetdeck commencing
immersion.
Thus, in order to accurately compare the computed results
against the measured data it was deemed necessary to
implement the same measured velocity profile to the numerical
simulations. Table 3 provides a summary of the main measured
data against computed results, where Fmax, Pmax and t refers to
slam force peak, slam pressure peak and the penetration time
respectively. The penetration time is calculated from the total
force signal/results and defined as the time of duration of the
total force peak with reference to penetration.
Although implementing the same profile for both numerical
methods and applying same physics setup, it is interesting to
observe the differences in both the trends and magnitude of the
computed hydrodynamic force, as shown in Fig. 6. Applying the
overset grid method overestimated the measured slam force
peak by less than 0.5% and shows good overall correlation with
the measured data. However, from a timing perspective the
computed vertical force peak occurs prior to the measured slam
force by approximately 4 ms for both the first and second peaks.
This could possibly be due to a slight lag in response of the
output signals of load cells comparing to the used Piezotronics
pressure transducers signals, where the CFD results correlates
well with the measured pressure signals from the timing
perspective.

Prediction of Slamming Loads on catamaran wetdeck using CFD

In contrast, applying the fixed grid technique with a tabulated

inlet velocity profile equivalent to the measured velocity,
underestimated the total force by 4.8% when compared against
the experimental data, as presented in Fig. 6. In addition, the
fixed grid technique could not detect the second peak of the
measured force that occurred at around 57 ms, this could be due
to the higher sensitivity of the moving body subjected to direct
acceleration or deceleration, as in the case of the overset grid,

Fig. 5: Experimental immersion and velocity data, for the target

velocity of 4 m/s

rather than simulating a fixed body with the transient inlet flow
applied, however this needs further numerical investigation.
Table 3: Results Summary
Test

Approach

Experiment

Drop test
Fixed grid
Overset grid

CFD

Fmax
kN
8.1
7.7
8.1

Pmax
kPa
228
236
243

t
ms
40
38
36

Fig. 6: Comparison of total slam force data against the CFD

results using both the overset grid and the fixed grid methods

Fig. 7: Pressure distributions along arched way of catamaran wetdeck time histories. Subplots (a-e) present the measured pressure time
histories against computed results using overset and fixed grid methods from P1 to P5 respectively.

Swidan

Prediction of Slamming Loads on catamaran wetdeck using CFD

Fig. 7 illustrates the time histories of measured and computed

pressure data at transducers P1 to P5 respectively, using both
overset grid and fixed grid methods. Fig. 7 presents a clear trend
of a reduction in the peak pressures with increasing immersion
can be seen as the slam location moves forward towards the bow
section.
In general good agreement was achieved between the computed
values and the experimental data for all transducers; however, it
was observed that allowing the body to move enhanced the
prediction of both the total slamming force and local slamming
pressures for the transient velocity profile. The computed results
using both simulation techniques were practically identical until
the slam occurs.
From the peak pressure distributions, it is clear that the slam
severity increases towards the aft part of the wetdeck. Thus
designers need to consider the large local loads that may act on
this particular vicinity.
From the CFD results it is possible to compute the transverse
component of force acting on the demihulls during a slam event,
as illustrated in Fig. 8. Special attention should be paid to this
transverse load, as its main consequence is the splitting force on
catamarans as well as the resultant dynamic split moment.
Although this load represents only 21% of the vertical slamming
force, which is still a significant load that can affect the integrity
of the hulls.

Fig. 8: Total transverse force acting on demihulls during modelwater impact.

CONCLUSIONS
This project investigated wetdeck slamming loads for a generic
wave-piercer catamaran hullform model during water-entry on
an initially calm free-surface.
Numerical predictions using two CFD modeling techniques
were performed for the wetdeck slamming problem and
successfully validated against new experimental drop test of a 3D model.
For the problem of water-entry at constant speed, although both
techniques provided close predictions in comparison with the
experimental work, the overset grid technique employing a
moving body relative to stationary fluid presented significantly
better prediction of the slam force magnitude. However, the
fixed grid method with a stationary body and moving fluid is
more computationally efficient.
The transverse split force has also been quantified during a slam
event and found to be approximately 21% of the vertical load.

Swidan

ACKNOWLEDGEMENTS
The authors acknowledge the ongoing support of Revolution
Design and INCAT. The assistance of Industrial Research
Limited in providing access to the experimental facilities is
gratefully appreciated. The Authors also recognise the work of
Liam Honeychurch at the Australian Maritime College for
constructing the experimental model.

REFERENCES
Battley, M. & Allen, T. 2012. Servo-hydraulic system for controlled
velocity water impact of marine sandwich panels.
Experimental mechanics, 52, 95-106.
Davis, M. R. & Whelan, J. R. 2007. Computation of wet deck bow
slam loads for catamaran arched cross sections. Ocean
Engineering, 34, 2265-2276.
Dragomir, S. S., Pietro, C. & Anthony, S. 1998. Some Remarks on the
Midpoint Rule in Numerical Integration. RGMIA research
report collection, 1.
FALTINSEN, O. M. 2012. International Ship and Offshore Structures
Congress (ISSC) 2012 in Rostock/Germany. Ships and
Offshore Structures, 7, 309-310.
French, B. & Thomas, G. A. 2014. Slam characteristics of a high-speed
wave piercing catamaran in irregular waves. Royal Institution
of Naval Architects. Transactions. Part A. International
Journal of Maritime Engineering, 156, A-25-A-36.
Incat. 2015. HSV 2 Swift (98 Metre Wave Piercing Catamaran)
[Online].
http://www.incat.com.au/domino/incat/incatweb.nsf/0/732EE
FD364269304CA2573A10011D332/$File/0610312EH.jpg.
Jacobi, G., Thomas, G. A., Davis, M. R., Holloway, D. S., Davidson,
G. & Roberts, T. 2012. Full-scale motions of a large highspeed catamaran: The influence of wave environment, speed
and ride control system. International Journal of Maritime
Engineering, 154, A143-A155.
Lewis, S. G., Hudson, D. A., Turnock, S. R. & Taunton, D. J. 2010.
Impact of a free-falling wedge with water: synchronized
visualization, pressure and acceleration measurements. Fluid
Dynamics Research, 42, 035509.
Menter, F. R. 1994. Two-equation eddy-viscosity turbulence modeling
for engineering applications. AIAA Journal, 32(8).
STARIUS, G. 1977. Composite mesh difference methods for elliptic
and boundary value problems. Numer. Math, 28, 243258.
Swidan, A., Amin, W., Ranmuthugala, D., Thomas, G. & Penesis, I.
2013. Numerical Prediction of Symmetric Water Impact
Loads on Wedge Shaped Hull Form Using CFD. World
Journal of Mechanics, 3, 1-8.
Swidan, A. A., Thomas, G. A., Ranmuthugala, D., Penesis, I. & Amin,
W. 2014. Numerical investigation of water slamming loads
on wave-piercing catamaran hull model. In: RINA, ed. 10th
Symposium on High Speed Marine Vehicles (HSMV 2014),
October 2014 Naples, Italy.
Thomas, G., Davis, M., Holloway, D. & Roberts, T. Extreme
asymmetric slam loads on large high speed catamarans. In:
RINA, ed. High Speed Marine Vehicles, 2002 Naples, Italy.
9.
Thomas, G., Winkler, S., Davis, M., Holloway, D., Matsubara, S.,
Lavroff, J. & French, B. 2011. Slam events of high-speed
catamarans in irregular waves. Marine Science Technology,
16, 14.
Wacawczyk, T. & Koronowicz, T. 2008. Comparison of CICSAM and
HRIC high-resolution schemes for interface capturing.
Journal of Theoretical and Applied Mechanics, 46, 325-345.

Prediction of Slamming Loads on catamaran wetdeck using CFD