X HSMV - Naples, October 2014

NUMERICAL INVESTIGATION OF WATER SLAMMING LOADS ON WAVE-

PIERCING CATAMARAN HULL MODEL

Ahmed A. Swidan, Giles A. Thomas, Dev Ranmuthugala, Irene Penesis, Walid Amin, Australian Maritime College,
University of Tasmania, Australia

SUMMARY

With the increasing demand for faster and lighter ferries the need for predicting motions and sea loads for efficient
structural design and safe operation has become necessary. Operation at speeds of around 40 knots in the open ocean,
where water impacts can result in structural damage and crew injuries emphasises the need for the development of
reliable tools to accurately predict slam loads.
This work investigated the behaviour of a quasi-2D section model of wave-piercing catamaran fitted with a centrebow
during the impact phase using finite-volume Computational Fluid Dynamics (CFD), overlapping grids, Reynolds-
Averaged Navier-Stokes (RANS) equations and Volume of Fluid (VOF) method to solve the water-entry problem. The
computed vertical acceleration and slamming pressures are compared to previously published drop test experimental
data and show good agreement. It was found that taking into account the compressibility of the trapped air is necessary
to accurately predict slamming pressures.

1. INTRODUCTION
To accurately predict hydrodynamic impact loads on
light-weight high-speed catamarans, it is necessary to
include slam events in estimating global loads in the
preliminary design stages. There is a lack of data
concerning unconventional hull forms in rough seas,
particularly data on motions and loads during full scale
Centrebow
trials and/or experimental tests, in order to conduct
Demihull
statistical studies for predicting the behaviour of such
vessels. In severe sea states many wave encounters result Figure 1: INCAT 112m high-speed catamaran fitted with
in rapid and significant local hydrodynamic loads called centrebow [4]
slamming.
These techniques can be divided into: full scale sea-trials
Over the past two decades, high-speed catamarans have
[5], scale model experimental tests [6], theoretical
been used for both commercial and military operations,
approaches [7], empirical formulae [8] and numerical
thus extending their service area from protected waters to
simulations [9].
the open ocean where frequent and large slams can result
Pioneering work on slamming was carried out by Von
in severe local and global loads.
Karman [10] and Wagner [11], which were followed by
A severe slam load acting on high-speed catamarans is
a significant number of publications, as summarised in
called wetdeck slamming. For a typical catamaran, this
Ship’s Structure Committee report (SSC-385A) [12].
occurs on the flat area of the fore body. For a catamaran
Kapsenberg [13] reviewed the problem of slamming
fitted with a centrebow, as shown in Fig. 1, slamming
from a practical perspective and highlighted the
occurs when the archway between the demihull and
challenge in predicting slamming loads, especially when
centrebow impacts the surface.
the angle between the hull and the free-surface is small
From a structural design perspective, primary importance
due to the effect of air inclusions and aerated water.
is given to the impact load, as the main source of
Almost all of these works focused on solving the slam
structure failure and buckling of ship frames tends to be
problem in either conventional hull forms, such as
hydrodynamic loads with very short time incidences [1].
wedges, or cylindrical structures as found in offshore
These slam events can excite the natural modes of the
systems. However relatively limited attempts have been
structure, a process called whipping, which can have a
made to solve the water-entry problems of 3D bodies or
significant influence on reducing the vessel’s fatigue life
unconventional hull forms such as catamarans.
as discussed by Thomas et al [2, 3].
Chuang et al [14] developed a method of predicting
Slamming is a complex non-linear problem. Many
slamming pressure in waves based on Wagner’s theory
researchers have developed various techniques to predict
[11] and relative velocity, known as the ‘k-factor
the behaviour of vessels during a slam event.
method’.

1
X HSMV - Naples, October 2014

This study found that even if a vessel is moving at a wetdeck archway during a slam event due to its
certain velocity in a direction other than the normal significant influence on the localised slamming
component to the impact surface, the impact load on the pressures.
wetdeck is only affected by the relative velocity The work presented here carries out CFD simulations
component normal to the impact point on the surface. using the Finite Volume Method (FVM) and is validated
Kaplan [15] considered the effect of forward speed in against published experimental drop test data, thus
calculating the slamming pressure on twin-hulled vessels establishing a numerical model to predict slamming loads
by applying Von Karman’s technique [10]. However, the and corresponding motions of wave-piercing catamarans.
selected 2D section was in the longitudinal direction This provides a preliminary indication on the suitability
rather than a transverse section. Thus the effect of two of the suggested to accurately predict the behaviour of
demihulls was not considered, consequently sealing off catamarans during slamming.
the escape of air and water when contact was not
considered.
2. NUMERICAL SIMULATION
Faltinsen [16] reviewed the problem of wetdeck
slamming on an initially calm free-surface. It was noted The numerical simulations were performed using STAR-
that for bottom and/or wetdeck slamming experimental CCM+ 8.06.007 on a PC to solve the problem of a free-
pressures should not be Froude scaled due to the falling 3D wave-piercer model free-fall on to a clam
significant effect of the entrapped air cushion between water surface at an initial velocity of 0.8984m/s at the
the free-surface and structures with small dead rise instant the demihull keel touches the free-surface. The
angles. Faltinsen also stated that slamming is a 3D motion was restricted to a single degree of freedom
problem; so that testing the water-entry problem purely (DOF).
in 2D will increase the loads significantly. To validate the numerical model, the computed motions
Davis and Whelan [7] designed a drop test rig in order to and slamming pressures were compared against
study slamming loads and corresponding motions on a experimental results from Whelan [19]. Whelan
series of quasi 2D hull form models including an INCAT investigated the behaviour of nine quasi 2D models,
wave-piercer. It was found that the 2D drop test can including a free falling 1/40-scale wave-piercing model
overestimate the peak slamming pressures by around entering still water.
three times the largest experienced slam loads on an 86m The model measuring 0.54m wide x 0.29m long x 0.22m
wave piercing vessel with the same transverse section. deep and the tank geometries are shown in Figs.2 and 3.
Yang et al. [17] solved the problem of water-entry of
wedge-shaped hull form, wave piercer with a flat top 0.54m
arch model and cylinder using the Finite difference
method on a fixed Cartesian grid to solve the Navier-
Stokes equations. The Constrained Interpolation Profile
(CIP) method developed by Yabe et al. [18] was used to P3
P4 0.22m
capture the free-surface deformation. For the wave- P2
piercer model, the study compared the velocity ratios and P1
flow visualisation against the drop tests conducted by
Davis and Whelan [7]. Although it was found that free-
surface elevation was in good agreement with Figure 2: Positions of four pressure transducers on the
experiments, a bigger separation was seen to occur at the wave piercer model
demihull sides by using the CIP method. In addition, the
velocity ratios were over predicted, particularly for
higher initial velocities.
These limitations can be overcome by using
Computational Fluid Dynamics (CFD) methods [13].
Swidan et al. [9] used CFD techniques to solve the
problem of water-entry of a free-fall wedge. The
computed slamming pressures and corresponding
motions were in good agreement with data from the drop
tests conducted by Whelan [19]. It was found that a
slight change in the location of the pressure transducers
can affect the results significantly. Also, computing
pressures in rapidly changing pressure zones, such as
near the wedge apex, needs particular focus and a
Courant number of 0.1 was considered sufficient to
capture the slamming pressures.
For CFD simulations of catamarans it is necessary to
consider the effect of the aerated water at the top of the

2
X HSMV - Naples, October 2014
Figure 3: Wave piercer model and tank dimensions
Table 1: Experimental instrumentation [19]:
Characteristic Pressure transducer Accelerometer
Range +/- 0-3,447 kPa +/- 30g
Sensitivity 4.1 kPa/mV 66 +/- 4 mV/g
Frequency
0 to 1500 Hz. 0 to 1500 Hz.
response
Face Diameter 3.8 mm Nil
Linearity 0.25% Nil
Four probe points were used in the simulation to capture
the pressure distribution around the archway of the hull,
with the positions coinciding with the locations of the
pressure transducers on the physical model (see Fig.2).
The characteristics of the pressure transducers and
accelerometer used by Whelan [19] are given in Table 1.
This simulation cannot be simplified to a 2D simulation
model, i.e. without a gap between the model and the tank
wall unlike simple wedges or cylinders due to the need
for the fluid trapped between the demi-hulls to escape as
the slam progresses (see Fig.3).
However, due to symmetry, the simulated model was
reduced to a quarter of the full domain, with the use of
symmetry boundaries, as discussed in section 3.2.
3. COMPUTATIONAL SET-UP
3.1 Grid Generation
The Finite Volume Method (FVM) was used to define
the discrete flow field due to its inherent
conservativeness. In this work, the domain was
subdivided into 600,000 hexahedral cells for one quarter
of the geometry, with finer cells around the centrebow
and in the confined gap between the hull and the tank
walls with a minimum of 10 cells including 5 prism
layers were used for inflation where rapid flow variations
can occur as shown in Figs.4 and 5.
In order to permit body motion under the effect of
gravitational forces without deforming the cells and
hence affecting the cell quality and/or resolution, the
overlapping grid method was used.
Figure 4: Symmetry, side and perspective, views of
initial general arrangement of the numerical model and
the tank side.
Figure 5: Perspective view of cross-section in grids of
quarter wave piercer numerical model.
The general procedure, as originally developed by
Starius [20] for using the overlapping method is to
generate two main grids, one around the model which
moves with it, called overset grid, and another stationery
grid at the far field, called the background grid, which is
adapted to the free surface and the outside boundaries,
such as the bottom and sides of the tank. These two grids
interpolate information within the overlapping zone
where donor and acceptor cells intersect (for more details
Swidan et al. [9]).
The overlapping method can be considered a locally-
structured and globally un-structured method. This is due
to the ability to use unstructured grids with higher-order
discretisation to resolve rapid flow variations by
assembling other independent structured blocks with
different cell sizes with data passed between them [21,
22]. This can offer the advantage of higher flexibility and
faster computational time compared with other methods,
such as deformed or re-meshed methods.
To minimise the error in interpolation, a cell size of
12.5mm was set for both the background and overset
mesh within the overlapping zone.
3
X HSMV - Naples, October 2014
The final background region had around 200,000 cells,
while the overset region included approximately
400,000cells.
3.2 INITIAL AND BOUNDARY CONDITIONS
Simulation of the free-falling quasi-2D catamaran hull
model was performed under realistic initial and boundary
conditions. Only one quarter of the geometry was
considered, imposing symmetry conditions in the
geometrical symmetry planes of the catamaran hull
model. There were two main dimensionless parameters
of importance when simulating the model, the mass
number m*=mm/ρTL2 and a dimensionless drop height
(H/L), where mm is the model mass, ρ is the water
density, T is the length of the model (0.29m), L is the
beam of the model (0.54m), and H is the drop height
measured from the top of the archway of the wetdeck to
the free-surface at the instant when the model starts
falling.
The wave piercing hull model, weighting 24.4kg, was
allowed to free-fall under the influence of gravity.
Figure 6: 3D wave-piercer catamaran model
computational domain
The top boundary of the simulation domain, which
represents air, was treated as a pressure outlet. The two
planes that dissected the catamaran model were treated as
symmetry planes. The starboard and bottom planes of the
domain were treated as slip walls. The tank wall adjacent
to the catamaran model and the hull surfaces were treated
as non-slip walls (see Fig.6).
3.3 GOVERNING EQUATIONS
The problem of the free-falling catamaran model
impacting on to the water surface is numerically
represented by the RANS equations in which the effect
of turbulence is solved using SST k-ω eddy-viscosity
model. The equations to be solved for a viscous 3D flow
are governed by continuity, three momentum
components and two equations for the Shear Stress
Transport (SST) K-ω turbulence model. The latter was
selected due to the insensitivity to free-stream conditions
of the K-ɛ model in the far-field, while retaining the
advantages of the K- ω model near the walls [23].
A volume fraction equation, Volume of Fluid (VOF)
method, accounts for the free surface deformation of air
and water fluid mixture. Water/ air are considered to be
two fractional components of a single effective fluid
characterised according to the proportion of each fluid
[24].
For this case STAR-CCM+ uses the 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 Finite Volume Method (FVM), the governing
equations on its integral form are approximated over
each control volume by the midpoint rule. This means
that before interpolating all integrals the discrete form of
the equations are solved at the centroid of faces and
volumes, converting the volume integrals to surface
integrals. The objective is to obtain a set of linear
algebraic equations, with the total number of unknowns
in each equation system corresponding to the number of
cells in the grid [25].
To complete the mathematical model, the fluids flow
when using the overset grid (moving grid) as the grid
movement is not prescribed by initial conditions, rather it
follows the motion of the moving body. This requires,
for this case, a coupled solution between the previous
mentioned equations and one equation for the simulated,
1DOF linear motion. In addition, the space-conservation
law equation is solved to conservatively express the
transport motion since the control volumes of the overset
mesh moves and changes its location as the body moves
[26].
3.4 NUMERICAL ACCURACY
This section presents the investigation into the effect of
venting clearance (see Fig.4) and air elasticity on the
computed results.
Assuming the model to be infinitely rigid, pressure
transducers were flush-mounted to the curved structure
surface neglecting the compressibility of water. The
effect of frictional forces between moving parts in the
drop test rig were assumed to be negligible. Symmetry
was assumed across the centerline during experimental
drop tests based on symmetric flow visualisation. The air
4
X HSMV - Naples, October 2014

at the top of the arch, between the structure and disturbed
free-surface, was assumed to compress adiabatically, i.e.
with no change in heat energy.
The error bound associated with the venting clearance
space between the model’s end flat face and the tank
wall during experiment, as stated by Whelan [19], was
more than 15% in percentage terms and 1mm in absolute
terms. This study was carried out using three different
grid sizes from 500,000 to 3000,000 cells, including a
range from 10 to 45 cells between the model’s flat face
and tank wall with a Courant number of 0.1.
It was found that simulating the same venting clearance
between 5 to 6mm space as in the experiment resulted in
an error of around 78%, as shown in Fig. 7. This error
can be attributed to boundary condition limitations due to
pressures and vertical accelerations result in large errors.
At 7mm clearance, the error bound associated with wall
clearances was around 15% in percentage terms and
1mm in absolute terms when compared against
experimental data. In addition, the boundary condition
limitations were minor at this venting clearance.
Moreover, two simulations were carried out at 8mm
venting clearance and stable results were achieved. Thus,
a 7mm venting clearance was used in all presented
simulations.
Experimental
15
Cell Size= 0.5mm
13
Cell Size= 0.25mm

Acceleration (gm/s^2)
the rotation of flow around the sharp ended edge of the 11
Cell Size= 0.1mm
wetdeck and in a very tiny space of the venting
9
clearance. Thus a detailed study was carried out by
increasing the venting clearance spacing from 6mm to 7

8mm, in increments of 1mm, as shown in Fig. 8. 5

In this study, the finest cell size at the venting clearance
3
where equal 0.02 ∆x, where ∆x is the venting clearance
based on grid independence study, see Fig 9. 1

All the presented results in Fig. 9 considered the air -1

compressibility effect. 0 0.02 0.04 0.06 0.08 0.1
Time (Second)
Fig. 8 shows that with a venting clearance of 6mm for
simulations using air as ideal gas or as a constant density Figure 7: Sensitivity to the grid resolution at 5mm
fluid, the predicted peak values of the slamming venting clearance

Figure 8: Venting clearance error study and corresponding time record uncertainty of vertical acceleration and slamming
pressures distribution

To investigate the sensitivity of the computed results to of the model was numerically solved using various cell
the cell sizes and the time step, the vertical acceleration sizes at a Courant number of 0.1, as shown in Fig. 9. In

5
X HSMV - Naples, October 2014

this figure, the cell sizes at the venting clearance ranged It was observed that including the elasticity of air could
from 0.1∆x to 0.05∆x. In addition, air compressibility enhance the computed vertical acceleration from
was included in this study. It was observed that there quantitative as well as temporal perspectives.
were only slight changes in the numerical solution results
despite the change in cell size. Therefore, 0.075 ∆x was 170
Experimenal
considered sufficient to perform the numerical simulation 150

Vertical Acceleartion (m/s^2)

CFD (compressible air)
to minimise computational time. 130
CFD (incompressible air)
110
170
Experimental 90
Vertical Acceleration (m/s^2)

150
0.1 ∆x_Compressible 70
130 0.075 ∆x_Compressible
50
110 0.05 ∆x_Compressible
30
90
10
70
-10
50 0 0.02 0.04 0.06 0.08 0.1
Time (Sec.)
30
Figure 10: Vertical acceleration of INCAT model as a
10
function of time.
-10
0 0.02 0.04 0.06 0.08 0.1
Time (Sec.)
Figure 9: Sensitivity mesh study vs. drop test data

4. RESULTS AND DISCUSSION

The numerical results were compared with the physical
drop tests results obtained by Whelan [19]. Firstly the
free surface elevations for the physical model on the
starboard side as recorded by a high-speed camera at
various time intervals were compared with computed
results on the port side as shown in Fig. 11.
The visual comparison indicated that the computed free-
surface elevations and the air cushion beneath the top
arch were in good agreement with those observed in the
experimental drop test.
Figure 11 shows that between times 110ms and 120ms
air bubbles were trapped and mixed with the water at the
top archway void, given that slamming pressures
occurred between the times of 110.3 and 110.6ms, as
presented in Figs. 15, 16, 17, and 18. Therefore, the
effect of the compressibility of air on the computed
results was investigated due to the existence of air
inclusion during the slam event.
The vertical acceleration of the tested model was
computed for incompressible and compressible air, i.e.
assuming that air has a constant density and an ideal gas
respectively. Thus, air density was computed in the
momentum equation and corrected in the continuity
equation until the solutions converged.
Figure 10 presents the vertical acceleration for both
simulations which were practically identical until the
computed slam began, i.e. 1.4%, for compressible air and
2.7%, for incompressible air, before the experimental
peak acceleration rapidly increased at 0.112s.
Although the trend of the numerical model’s acceleration
was similar to the physical model, the computed results
over estimated the vertical acceleration in the range of
4.4% for compressible air and 5.6% for air having a
constant density (see Fig. 10).

6
X HSMV - Naples, October 2014
Figure 11: Computed free surface elevations as a
function of time on left of the figure vs. experimental
water-entry of catamaran model on right of the figure, as
recorded by high-speed photography by Whelan [19].
-1.4
-1.2
Velocity (m/s)
-1
-0.8
Experimental
CFD (compressible air)
CFD (incompressible air)
-0.6
0 0.02 0.04 0.06 0.08 0.1
Time (Sec.)
Figure 12: Computed velocity profiles for the INCAT
model vs. experimental data.
The velocity profiles are presented in Fig. 12. For
incompressible air, the velocity tended to decelerate
faster than for the compressible air, with both being
lower than the measured data from the vertical
accelerometer during the drop test. However, the velocity
profiles for both simulations were in good agreement
with the experimental data.
Figure 13 presents the locations of the four pressure
probes during slam event after a time of 80ms.
Peak pressures can be a good indicator of the relative
pressure experienced in different impact situations,
measuring or computing the true peak pressures is
complicated due to uncertainty in pressure
measurements, such as the sampling rate, which
according to Det Norske Veritas should not be less than
20kHz [27].
The represented pressure data was averaged over the
finite sensing area. It was necessary to ensure that the
sensing area was flush mounted to the curved section of
the model. This could not be achieved except by
flattening the iso-surface longitudinally at the pressure
transducer positions.
Figure 14 presents the time history of the measured and
the computed results pressure at probe P1, firstly by
assuming air as compressible and then as a constant
density gas.
Jet forming
P3 P4
P2
P1
Figure 13: Simulation on impact after 80ms involving
locations of the four pressure probes and jet forming
around probe P2
25
P1 (Experiemental)
P1 (Compressible)
20
P1 (incompressible)
15
Pressure (kPa)
10
0
0 0.02 0.04 0.06 0.08 0.1
Time (Sec.)
-5
Figure 14: The effect of air compressibility on pressure
prediction at probe P1.
This figure presents two peak pressures; the first
occurred at 64ms after the centrebow apex impacted the
free-surface, and the second occurred after the archway
nearly filled with water and a small amount of entrapped
air.
It is clear that including air compressibility improved the
predicted pressure from both the time and magnitude of
maximum pressure perspectives.
The computed result including air compressibility under
estimated the measured peak value at probe P1 by10%
and it occurred 2.5% before the measured data. It is clear
that the time duration of the numerical compressible slam
event is nearly the same as in the measured data.
On the other hand, it was observed that neglecting air
compressibility will lead to a sharper slam pressure,
faster slam event, and under estimation of the magnitude
of the maximum pressure by around 15% against drop
test data.
Figure 15 shows that good agreement was achieved
between the computed peak values and the experimental
data for probe P2. However, a pressure drop into the
negative region occurred in the experimental data for no
apparent reason around 80ms. This could be due to the
jet forming at the free-surface occurring at the same time
7
X HSMV - Naples, October 2014

75
as the pressure transducer touched the surface, as shown P3 (Experimental)
in Fig.13, or more likely, the pressure sensing area not 65 P3 (Compressible)
being flush mounted to the curved surface, thus P3 (incompressible)
55
separation may have occurred between the sensing

Pressure (kPa)
surface and the water. Between the occurrence of the 45

slam event and 80ms, both the numerical solutions and

35
experimental data showed the same trends.
Hence, neglecting the air compressibility resulted in an 25

increased numerical uncertainty from 1.3%, for

15
compressible air to 13%. Also, only a slight effect was
observed with regard to the timing of the slam event for 5

both simulations.
-5 0 0.02 0.04 0.06 0.08 0.1

Time (Sec.)
25
P2 (Experimental) Figure 16: Pressure at probe P3 with respect to time
110
20
P2 (compressible) P4 (Experimental)

15 P2 (incompressible) 90 P4( Compressible)

Pressure (kPa)

P4 (Incompressible)
10
70

5
Pressure (kPa)
50

0
0 0.02 0.04 0.06 0.08 0.1
30
-5

10
-10 Time (Sec.)

0 0.02 0.04 0.06 0.08 0.1

Figure 15: The effect of air compressibility on pressure
prediction at probe P2
By considering the compressibility of air, the computed
peak pressure value at probe P3 correlated better with the
experimental data. The comparison showed that the
computed pressure time history was similar to that
predicted by the drop test. However, both computed
results under estimated the slamming pressure by around
13% for compressible air to 22% when air
compressibility was neglected (see Fig.16).
The pressure at probe P4 is shown in Fig. 17, where the
time history of both the computed results and
experimental data show that the first slam event
happened at the same time period. This represents the
maximum slamming pressure when compared against the
other three measured pressures.
After this slam event, a rapid increase in the vertical
acceleration was experienced, see Fig. 10. This was
followed by an increase in pressure in probes P3, P2 and,
in finally P1, as shown in Figs. 14, 15, and 16. The
magnitude of peak pressure was seen to be directly
proportional to the pressure where the slam event begun.
Figure 17 presents the trend line for the computed
pressures at probe P4 and is in good agreement with the
experimental trend line. The peak value at probe P4
decreases as expected, when the air compressibility is
neglected.
-10
Time (Sec.)
Figure 17: Time history of computed slamming pressure
at probe P4 vs. experimental data
From analysis of the results, it was apparent that
although CFD overestimated the vertical acceleration of
the numerical model, or in other words the slamming
loads per unit length by around 4%, all the pressures
were slightly under predicted.
In addition, it was found that slamming pressure was
significantly affected by the aerated water and air
compressibility beneath the top arch of the wetdeck,
whilst only a slight effect was observed in the computed
vertical acceleration. Thus, slamming pressures should
not be used as the only indication for the magnitude of
total slamming loads on catamaran models due to the
effect of aerated water and air bubbles in computing the
slamming pressures at certain positions. Hence,
slamming loads can be accurately indicated by
measuring the total forces using load cells despite the
extrapolation uncertainties.
5. CONCLUSIONS
The primary aim of the presented work was to evaluate
3D CFD methodology for predicting the magnitude of
slamming loads, corresponding motions and flow
visualisation for a free-falling wave-piercing catamaran
model in still water.
The CFD methodology included the jet formation of
water around the hull section during water-entry and the
compressibility of the air mixed in water at the top
archway. In general, considering air elasticity as a

8
X HSMV - Naples, October 2014
function of pressure is better in predicting slamming
pressures and corresponding motions of wave piercing
catamarans than air has a constant density due to the
existence of air inclusion during the slam event.
The presented CFD methodology was used to predict the
magnitude and peak values of slamming pressures at four
positions along the top archway. Computed pressures far
from the initial impact where slam occurs resulted in
under predicting slamming pressures. Special attention
must be paid to ensure that the pressure transducers are
fitted flush to the hull, particularly in the case of curved
hull surfaces such as in the catamaran model used in this
study.
The CFD correctly predicted the velocity profile and the
maximum vertical acceleration of the tested model.
Thus, this CFD methodology presented can be used to
predict the maximum loads acting per unit length of the
model during water-entry. Although the numerical
vertical acceleration is marginally overestimated, the
slamming pressures were under estimated. It should be
noted that the slam pressures did not represent the
slamming loads due to the effect of aerated water in
computing the pressure distribution at certain locations.
Overlapping grids were successfully used to simulate the
motion of the numerical INCAT model under
gravitational forces. The computed flow visualisation
illustrated the ability of CFD to capture the trapped air
bubbles mixed with water at the top of the arch in the
hull cross section during the slam process. Wetdeck
slamming of wave-piercing catamarans is thus affected
by compressibility of the air inclusion at the highest
point beneath the wetdeck during the slam event.
REFERENCES
[1] F. J. Huera-Huarte, D. Jeon, and M. Gharib, "Experimental
Investigation of Water Slamming Loads on Panels," Ocean
Engineering, vol. 38, pp. 1347-1355, 2011.
[2] G. Thomas, M. Davis, D. Holloway, N. Watson, and T. Roberts,
"Slamming response of a large high-speed wave-piercer catamaran,"
Marine Technology, vol. 40, pp. 126-140, 2003.
[3] G. Thomas, M. R. Davis, D. S. Holloway, and T. Roberts, "The
effect of slamming and whipping on the fatigue life of a high-speed
catamaran," Australian Journal of Mechanical Engineering, vol. 3, p.
165, 2006.
[4] Maritime Denmark. (March 2013). KatExpress 2 on its way to
Denmark. Available:
http://www.maritimedenmark.dk/?utm_source=uknyhedsbrev22-03-
2013&utm_medium=email&utm_content=uknyhedsbrev2013&utm_ca
mpaign=uknyhedsbrev&Id=16757
[5] M.Davis, J.Whelan, and G.Thomas, "Computational Modeling of
Wet Deck Slam Loads with Reference to Sea Trials," presented at the
Ninth International Conference on Fast Sea Transportation, Shanghai,
China, 2007.
[6] W. Amin, M. Davis, G. Thomas, and D. Holloway, "Transient Wave
Loads on Large High Speed Catermarans," in 8th International
Conference on High Performance Marine Vehicles, 2012, p. 246.
[7] M. R. Davis and J. R. Whelan, "Computation of wet deck bow slam
loads for catamaran arched cross sections," Ocean Engineering, vol. 34,
pp. 2265-2276, 2007.
[8] X. U. Guo-dong and D. Wen-yang, "Review of Prediction
Techniques on Hydrodynamic Impact of Ships " Marine Science, vol. 8,
2009.
[9] A. Swidan, W. Amin, D. Ranmuthugala, G. Thomas, and I. Penesis,
"Numerical Prediction of Symmetric Water Impact Loads on Wedge
Shaped Hull Form Using CFD," World Journal of Mechanics, vol. 3,
pp. 1-8, 2013.
[10] Von Karman, "The Impact on Seaplane Floats during Landing,"
National Advisory Committee for Aeroneautics (NACA), Technical
Note, October 1929.
[11] H.Wagner, "Über Stross-und Gleitvorgange an der Oberflasche "
Von Flüssigkeiten, vol. 12, No. 4, May 1932.
[12] J. Daidlooa and V. Mishkevich, "Hydrodynamic Impact on
Displacement Ship Hulls," Ship Structure Committee (SSC-385), 1995.
[13] G. Kapsenberg, "Slamming of ships: where are we now?,"
Philosophical Transactions of the Royal Society A: Mathematical,
Physical and Engineering Sciences, vol. 369, pp. 2892-2919, 2011.
[14] A. Korobkin, "Elastic response of catamaran wetdeck to liquid
impact," Ocean Engineering, vol. 25, pp. 687-714, 1998.
[15] P. Kaplan, "Analysis and Prediction of Flat Bottom Slamming
Impact of Advanced Marine Vehicles in Waves," ISP, vol. 34, March
1987.
[16] O. M. Faltinsen, M. Landrini, and M. Greco, "Slamming in
Marine Applications," Journal of Engineering Mathematics, vol. 48,
pp. 187-217, 2004.
[17] Q. Yang and W. Qiu, "Numerical simulation of water impact for
2D and 3D bodies," Ocean Engineering, vol. 43, pp. 82-89, 2012.
[18] Y. Takashi, y. Feng-Xiao, and U. Takayuki, "The Constrained
Interpolation Profile Method for Multiphase Analysis," Journal of
Computational Physics, vol. 169, pp. 556–593, 2001.
[19] J. R. Whelan., "Wetdeck slamming of high speed catamarans
with a centrebow," PhD, National centre of Maritime Engineering and
Hydrodynamics, Univesrsity of Tasmanai, 2004.
[20] G. Starius, "Composite mesh difference methods for elliptic and
boundary value problems," Numer. Math, vol. 28, pp. 243–258, 1977.
[21] I. Demirdzic, Lilek, Z., and Peric, M., "A collocated finite
volume method for predicting flows at all speeds," nt. J. for Numerical
Methods in Fluids, p. 16, 1993.
[22] H. HADˇZI´C, "Development and Application of a Finite
Volume Method for the Computation of Flows Around Moving Bodies
on Unstructured, Overlapping Grids," PhD, Technischen Universit¨at
Hamburg, 2005.
[23] F. R. Menter, "Two-equation eddy-viscosity turbulence modeling
for engineering applications," AIAA Journal, vol. 32(8), 1994.
[24] C. W. Hirt and B. D. Nichols, "Volume of fluid (VOF) method
for the dynamics of free boundaries," Journal of Computational
Physics, vol. 39, pp. 201-225, 1981.
[25] S.Dragomir, P.Cerone, and A.Sofo, "Some Remarks on the
Midpoint Rule in Numerical Integration," RGMIA research report
collection, vol. 1, 1998.
[26] H.Mørch, S.Enger, M.Peri´c, and E.Schreck, "Simulation of
Lifeboat Launching Under Storm Conditions," presented at the 6th
International Conference on CFD in Oil and Gas, Metallurgical and
Process Industries, Trondheim, Norway, 2008.
[27] D. N. VERITAS, ENVIRONMENTAL CONDITIONS AND
ENVIRONMENTAL LOADS. Høvik, Norway, 2007.