Вы находитесь на странице: 1из 14

J Mar Sci Technol

DOI 10.1007/s00773-010-0105-y

ORIGINAL ARTICLE

Slam events of high-speed catamarans in irregular waves


Giles Thomas • Stefan Winkler • Michael Davis •

Damien Holloway • Shinsuke Matsubara •


Jason Lavroff • Ben French

Received: 15 February 2010 / Accepted: 12 August 2010


 JASNAOE 2010

Abstract Slam events experienced by high-speed Keywords Slamming  Wave loads  Catamaran 
catamarans in irregular waves were characterised through High-speed craft  Model experiments
experiments using a hydroelastic segmented model. The
model was designed to represent the dynamic behaviour
List of symbols
of the full-scale Incat 112 m vessel and to allow the
CBT Centrebow truncation
measurement of the slam load on the centrebow and wet
Fi0 Slam force on transverse beam (N)
deck. It was tested in irregular head seas at two speeds
FT0 Total slam force on centrebow (N)
relating to Froude numbers of 0.32 and 0.60. Nearly 300
F10 Slam force on forward transverse beam (N)
slams were identified in the test data and analysed with
F20 Slam force on aft transverse beam (N)
respect to kinematic parameters. Slams were found to pffiffiffiffiffiffi
have a large range of magnitudes, with the largest
Fr Froude number Fr ¼ v= gL
equivalent to 1785 tonnes full scale (approximately 70% g Acceleration due to gravity
of vessel displacement); however, the majority of events hCBT Immersion at CBT (m)
were of relatively low severity. Differences in slam hw Wave elevation (m)
characteristics were found for the two model speeds tes- l1 Distance between strain gauges mounted on elastic
ted; slams at the slower speed generally occurred further links and pin joint (m)
forward on the hull, prior to the wave crest and with a l2 Distance between strain gauges mounted on elastic
bow down pitch angle. Immersion of the centrebow to the links 1 and 2 (m)
two-dimensional filling height of the cross-section L Length at waterline
between the centrebow and demihulls is shown to be a M10 Bending moment at elastic link 1 on transverse
better indicator of slam occurrence than immersion to the beam (N m)
top of the archway. M20 Bending moment at elastic link 2 on transverse
beam (N m)
t Time (s)
v Velocity (m/s)
vRV Relative vertical bow velocity (m/s)
vv Vessel vertical velocity at CBT (m/s)
G. Thomas (&)  S. Winkler  S. Matsubara  B. French vw Vertical velocity of water surface (m/s)
Australian Maritime College, w2 Distance between forward and aft transverse beams
University of Tasmania, Locked Bag 1395, (m)
Launceston, TAS 7250, Australia
e-mail: giles@amc.edu.au
x1 Longitudinal position of total slam force (m)
Z Vertical displacement (m)
M. Davis  D. Holloway  J. Lavroff hCBA Angle of centrebow archway ()
School of Engineering, hp Pitch angle ()
University of Tasmania,
Hobart, TAS, Australia
hRI Relative impact angle ()

123
J Mar Sci Technol

hw Angle of wave surface () Although full-scale data are extremely valuable, their
f Damping ratio acquisition is often difficult. There is no control over the
environmental conditions (wave height, wave period, wave
spreading) and often little or no control over the operating
conditions (vessel speed, trim, displacement, heading, time
travelled in a specific direction). This makes it difficult to
1 Introduction draw wide-ranging conclusions concerning the influences
on slam behaviour. It is also not usually possible to
Large high-speed catamarans are currently utilised by both instrument the full-size vessel as fully as desired; for
commercial and military operators. Their ability to carry example, the fitting of fixtures such as pressure transducers
relatively large payloads at high speed ensures that they are which penetrate the hull is generally not feasible.
an effective platform for a range of operational require- Model testing offers the opportunity to investigate slam
ments, e.g. car, truck and passenger ferries and naval behaviour with greater control over the environmental and
transport and support ships. To maximise their transport operational conditions than in full-scale trials. In addition,
efficiency (by maximising payload and speed whilst min- the model may be more extensively instrumented and
imising power) it is crucial that their design is optimised more revealing techniques can be used than for a full-size
to minimise structural weight. For this to be achieved, vessel. In order to measure wave loads and to correctly
knowledge of the sea loads imparted on such vessels, for a model slam behaviour and the dynamic response to
wide range of operating conditions, is critical. slamming, a hydroelastic model is required [6]. A hydro-
The largest loads experienced by large high-speed elastic model may be continuously elastic, with a stiffness
catamarans are generally slamming loads. Slam events distribution similar to that of the full-scale vessel [7].
occur when the vessel motion causes an impact between However, such a technique does not allow for the model’s
the cross deck structure and the water surface. As well as stiffness to be altered once it has been constructed, and
causing high local pressures, slams can impart a large measuring catamaran slamming loads with such a con-
global load onto the vessel. In severe sea conditions an figuration would be challenging. In contrast, a hydroelastic
extreme slam event may result in damage to the vessel [1]. model may be split into a series of segments that are
The dynamic response of the structure, observed as a joined using a backbone beam which has the required
shudder on-board the vessel after a slam, is known as stiffness. Segmented models may utilise either a continu-
whipping, and can have significant implications with ous backbone beam through the segments and segment
respect to the fatigue life of the vessel’s structure [2]. gaps [8–10] or elastic links that join the rigid segments
As well as providing structural design information, [11–13]. The structural response and hence wave loads can
understanding the slamming behaviour of high-speed be measured using a force balance [14], load cells [12, 15]
catamarans is vital for setting appropriate operational or strain gauges [9].
guidelines. For example, operators need to recognise the For model tests that investigate slamming specifically,
implications of operating at specific speeds and heading pressure transducers are usually used to measure the
angles in a particular wave environment. slamming pressures on the underside of the hull [15, 16].
Conducting measurements on full-scale vessels is an For catamaran models, as well as slam pressures, the slam
important method for investigating slamming behaviour. load on the wet deck may be measured using force trans-
The installation of a series of strain gauges, a bow-mounted ducers [17].
wave sensor and motion sensors on several vessels during In the present work, a hydroelastic segmented model of
delivery voyages and commercial operations has allowed a wave-piercer catamaran has been designed and con-
conclusions to be drawn on a range of aspects, including structed to obtain experimental data on slamming behav-
slam occurrence rates with respect to sea conditions and iour. It was extensively instrumented with strain gauges,
vessel speed, slam magnitude through structural responses pressure transducers, linear voltage displacement trans-
and whipping behaviour [3–5]. ducers (LVDTs) and a series of wave probes. The model
A key finding has been the identification of the range was tested in irregular head seas for a variety of significant
of slam events experienced by high-speed catamarans. wave heights and modal periods.
Occasionally the vessel will experience an extreme event The data were analysed to identify slam events and to
which results in severe structural loadings. Understanding characterise them according to a range of parameters,
the mechanisms for the occurrence of such extreme slam including relative vertical velocity, slam location on the
events is important, since it may lead to design or opera- cross structure archway, vessel pitch angle and wave
tional changes to promote their avoidance. location.

123
J Mar Sci Technol

2 Model experiments (14 mm2, machined from solid aluminium) bolted into a
40 mm2 hollow-section aluminium backbone beam. The
2.1 Hydroelastic catamaran model elastic links enabled the measurement of the vertical
bending moment at the segment gaps using strain gauges.
A hydroelastic segmented model was designed and con- The links were also designed have the appropriate stiffness
structed to allow the measurement of the global wave and to give a first longitudinal bending mode for the model of
slam impact loads experienced by high-speed catamarans. 13.8 Hz; this being scaled to represent the main longitu-
A segmented model allows the simulation of the hydro- dinal bending mode of the full-scale vessel at 2.1 Hz [20].
elastic behaviour of the full-scale ship; the flexural natural Prior to the construction of the first 112 m Incat cata-
frequencies were scaled consistently with the full-scale maran, the first longitudinal mode of vibration required
vessel and were used to accurately model the whipping of the 2.5 m catamaran model was predicted using the
response to waves. Model flexibility was achieved by the longitudinal bending mode frequencies measured by con-
introduction of flexible links at the segmentation sections. ducting exciter tests on 86 and 96 m Incat catamarans [18].
The segments themselves were designed and constructed to The first longitudinal modal frequency of the 112 m cata-
be of high stiffness so that flexibility was only introduced maran at a loaded ship displacement of 2500 tonnes was
in the model at the elastic links. Such a model is recom- predicted to be 2.1 Hz and was used as a basis for estab-
mended when global wave-induced loads are to be lishing the first longitudinal modal frequency at model
measured [6]. scale of 13.79 Hz. Note that subsequent to the experiments
A 112 m Incat wave-piercer catamaran was used as the described in this paper, the first longitudinal mode of the
basis for the hydroelastic model. The principal details of 112 m catamaran was subsequently confirmed through full-
the full-scale vessel are given in Table 1. Each demihull scale measurements to be 2.4 Hz [19]. It should be noted
had two transverse cuts, dividing it into three segments; see that whilst the overall dynamic structural behaviour of the
Fig. 1. These segments were connected by elastic links full-scale vessel was modelled, the local elasticity of the
shell plating was not.
Table 1 112 m Incat wave-piercer catamaran particulars A full description of the three degree of freedom theo-
Length overall 112.6 m
retical model developed to predict the longitudinal bending
frequency of the catamaran model as a function of the
Length waterline 105.6 m
effective link stiffness and the mass distribution of the
Beam demihulls 5.80 m
physical model is given in Lavroff et al. [20]. The theory
Beam overall 30.5 m
takes into consideration a single catamaran demihull con-
Draft 3.93 m
sisting of three separate rigid hull segments joined together
Speed 38 knots
by forward and aft torsion springs. During a slam event, the
Deadweight 1380 t maximum
forward and aft segments of the hull vibrate in a rotational
Main engines 49 MAN 28/33D (each 9000 kW)
direction opposite to each other and against the rotational
Propulsion 49 Wartsila LJX 1500 water jets
stiffness of the torsion springs in order to replicate the first
longitudinal mode of vibration of a catamaran demihull.
To allow the measurement of slam loads, the centrebow
and forward portion of the wet deck was constructed as a
separate segment from the two demihulls; see Fig. 1. The
cut between the demihulls and centrebow was designed to
enable the global wave loads to be measured on the
demihulls, whilst any slam load on the centrebow and wet
deck would act purely on the central segment, before being
translated through to the demihulls. This is illustrated in
Fig. 2, which shows a section through the demihull and
centrebow segments at the aft end of the centrebow. The
centrebow segment was supported on two transverse
19 mm2 hollow aluminium beams. Each beam was pin-
connected to the demihull backbones and incorporated two
small solid aluminium elastic links (see Fig. 3). The
Fig. 1 3D diagram of the segmented catamaran model consisting of
bending moments in the four elastic links located in the
six separate hull components: 1, centrebow; 2, 3, forward demihulls, transverse beams could be used to calculate the location
4, 5, aft demihulls; 6, midship demihulls and magnitude of the vertical force on the centrebow and

123
J Mar Sci Technol

Table 2 Hydroelastic segmented catamaran model particulars


Length overall 2.5 m
Displacement 27.4 kg
Trim Level
Radius of gyration 0.64 m
LCG 0.954 m from transom
VCG 0.167 m above keel
Trim tabs 7
Forward segment length 0.35L
Fig. 2 Transverse section through demihulls and centrebow, showing
Mid-segment length 0.25L
separate segments and the capacity of the centrebow to accept all the
vertical slam force Aft segment length 0.40L
Gaps between segments Sealed with latex and waterproof tape

Fig. 3 Photograph of forward centrebow transverse beam (looking


forward) showing the elastic link and pin joint connection to the port
demihull

wet deck. The centrebow segment was manufactured to


enable the fitting of pressure transducers on the starboard
side of the centrebow and wet deck. Fitting plugs were Fig. 4 Hydroelastic segmented model of the Incat 112 m catamaran
embedded into the hull at 84 locations, extending from
frame 55 to frame 82 (full-scale frame spacing was target displacement of 27.4 kg. Therefore, minimising
1200 mm, frame 0 at the transom). weight in the model’s design and construction was crucial,
All of the gaps between segments were sealed using a especially to provide sufficient ballast mass so as to
combination of waterproof tape and latex rubber. Particular achieve the target trim and radius of gyration values. The
attention was given to the gap between the centrebow and particulars of the model are given in Table 2 and the model
demihulls, ensuring realistic water flow up the side of the is shown in Fig. 4.
hull towards the archway.
Two transverse beams were connected to the midship 2.2 Test facility
segments of each demihull, forming a rigid midship seg-
ment which comprised the midship demihull segments and The experiments were conducted in the Australian Mari-
the central and aft transverse wet decks. A longitudinally time College towing tank, which is 100 m long, 3.55 m
orientated aluminium section was attached to these main wide and has a water depth of 1.5 m. A powered carriage
transverse beams at the model centreline and extended aft was used to tow the model, which has a maximum speed
from the rear crossbeam. The model was towed from a of 4.6 m/s. The waves were generated by a computer-
hinge fitting attached to this longitudinal section, with a controlled single-flap paddle wavemaker that was posi-
slider located further aft constraining the model in yaw. tioned at one end of the tank. The computer software
The model was constructed from sandwich carbon fibre controlling the wavemaker paddle provides the ability to
and Divinycell foam core to produce hull segments with produce a wide variety of waveforms, including regular
large inherent stiffnesses in order to keep the model weight waves and irregular wave spectra.
to a minimum. The overall model size was restricted to a A two-post system was used to attach the model to the
length of 2.5 m due to the size of the test facility, giving a carriage. The posts moved freely in the vertical direction in

123
J Mar Sci Technol

linear bearings on the carriage frame. The forward post was wave profile prior to the test run. An acoustic wave probe
connected to the model using a hinge allowing rotation (UltraLab ULS-40D) recorded the wave profile during the
around the lateral axis only and restricting all relative run from a position on the carriage offset transversely
translational modes. The aft post was connected via a ball (1200 mm from model centreline) but in the same longi-
bearing whose base was allowed to slide longitudinally. tudinal position as the bow of the model. In order to
The model was therefore free to heave and pitch whilst accurately identify the slope of the wave at the bow of the
being constrained in roll, yaw, sway and surge. model, three resistance wave probes were set up side by
side with a small longitudinal offset (20 mm). This com-
2.3 Instrumentation bined instrument was attached to the carriage in line with
the forward transverse beam of the centrebow at a distance
A set of instruments was utilised to measure the global 1490 mm from the model centreline.
wave loads, slam loads, slam pressures, model heave and
pitch, and encountered wave elevation. Strain gauges 2.4 Slam load calibration
installed in a half-bridge configuration on the top and
bottom surfaces of each of the demihull elastic links Figure 6a shows a schematic diagram of the slam force
enabled measurement of the global wave-induced strains acting on the forward transverse beam. The beam was
and structural bending moments. Gauges on the four cen- pinned at both ends A and B with two elastic links. Strain
trebow transverse beam elastic links allowed the slam gauges were mounted on each of the elastic links, and were
forces on the centrebow and forward wet deck to be separated by a distance l2. The port and starboard strain
evaluated from the bending moments in these links, the pin gauge measurement positions were each located at a dis-
fixings to the demihulls being connections of zero moment. tance l1 from the pinned joints A and B, respectively.
The slamming pressures were measured using piezore- Strains were measured at each of the elastic links to eval-
sistive pressure transducers (Endevco 8510C-50), which uate the bending moments M10 and M20 . The slam force, Fi0 ,
are rugged, miniature, high-sensitivity instruments. The was evaluated from the magnitude and position of the
piezoresistive-type pressure transducers measure the pres- measured bending moments:
sure difference on the sides of a silicon diaphragm, and M10 þ M20
therefore a vent tube is introduced so that the gauge pres- Fi0 ¼ : ð1Þ
l1
sure can be read directly. Seven pressure transducers were
utilised, located in a line along the top of the archway Figure 6b shows a schematic diagram of the forward
between the centrebow and the demihull; see Fig. 5. The and aft transverse beams; the beams were separated by a
vertical motion of the model was measured at each of the distance w2. The total slam force acting on the centrebow,
two tow posts using a LVDT. This provided data to cal- FT0 , was the summation of the forces acting on the forward
culate model heave and pitch motions. transverse beam, F10 , and aft transverse beam, F20 :
Both static and moving wave probes were used to FT0 ¼ F10 þ F20 : ð2Þ
measure the wave elevation. A static resistance wave
probe, located 7 m from the wave maker, recorded the The longitudinal position of the total slam force, x1, was
evaluated using:

Fig. 6 a Centrebow forward transverse beam configuration for


evaluating slam loads from measured bending moments. b Centrebow
forward and aft transverse beam configuration for evaluating slam
Fig. 5 Schematic of model instrumentation (dimensions in mm) loads from measured bending moments

123
J Mar Sci Technol

F10 4.0
x1 ¼ w2 : ð3Þ Measured Spectrum
F10 þ F20 3.5 JONSWAP Spectrum
3.0

S 0 (m2/(rad/s))
To calibrate the centrebow configuration the model was
turned upside down and the sheer line of each demihull 2.5
segment was mounted flush with a flat horizontal granite 2.0
bench top. This restrained the demihull segments but
1.5
allowed the centrebow to move freely in the vertical
1.0
direction. Varying calibration loads were applied to the
centre line of the hydroelastic model at several positions on 0.5
the centrebow wet deck and the centrebow keel to simulate 0.0
a slam load. 0 0.5 1 1.5 2 2.5 3

To determine the influence of the bow inertia on the Wave Frequency (rad/s)
predicted slam force from strains measured during slam Fig. 7 Comparison of measured and idealised wave spectra
impact, accelerations were measured using an Endevco
accelerometer mounted on the centreline of the centrebow
forward transverse beam during towing tank tests per- speeds were tested: 20 knots and 38 knots at full scale (1.53
formed in regular waves. The peak accelerations mea- and 2.89 m/s model scale). A comparison of the measured
sured on the model were then used to evaluate the inertia wave spectrum in the test region of the towing tank is
of the centrebow to predict the accuracy of the (quasi- compared with the required idealised wave spectrum in
static) slam load calculation. Based on the accelerometer Fig. 7; excellent correlation between the two spectra can be
measurements, the calculated bow inertia force repre- seen.
sented up to 17% of the total peak sagging quasi-static
slam load if the peak quasi-static force and peak accel- 2.6 Data acquisition
eration occurred at the same instant. Thus, the slam forces
on the hull could be up to 17% larger than those To acquire the data, two acquisition systems were used.
measured. The first system was the permanent on-board towing tank
The bow inertia loads were initially considered to be carriage system (NI 6036E series, National Instruments,
insignificant, because the bow segment was designed to Austin, TX, USA). This acquired data from the carriage
have a considerably higher natural frequency (60 Hz) than speed, LVDTs, wave probes and pressure transducers. The
the first longitudinal mode whipping frequency (13.79 Hz), second DAQ system was a National Instruments Compact
so the moments recorded in the forward transverse beam RIO (cRIO) data acquisition system, which was used to
elastic hinges would give a direct indication of the slam record the strain gauge from the elastic links. To aid syn-
forces. However, tests in which bow acceleration was chronisation between the two sets of data, a common
measured (undertaken subsequent to the main series of signal—an LVDT—was recorded by each system simul-
tests described in this paper) did indicate that bow inertia taneously. For adequate resolution of the pressure trans-
loads during slamming were not completely negligible. ducer signals, the carriage DAQ was sampled at 5000 Hz,
Since bow acceleration records were not available for most with the cRIO system recording at 500 Hz.
of the towing tank tests, it was not possible to correct for
bow inertia loads in each case. 2.7 Test run method

2.5 Test conditions The full wave spectrum was required to have been
established in the test section of the tank prior to com-
The main aim of the work was to investigate the conditions mencing a run. Since the lower-frequency waves in the
for the occurrence of slam events and to characterise the spectrum travel at a faster speed than the high-frequency
slams with respect to magnitude and kinematic parameters. waves, it was important to wait for the high-frequency
The tests were conducted in a sea state corresponding to a waves to progress a sufficient distance down the tank
full-scale significant wave height of 3.75 m and a modal before starting each test run. The required waiting time
period of 8 s. The wave conditions were simulated using a was calculated based on the group velocity of the highest
JONSWAP sea spectrum. These conditions and a limited frequency waves in the spectrum. For each wave condition
fetch idealised spectrum were chosen since they are tested, the equivalent of 20 min of full-scale data were
representative of waters typically experienced by high- recorded, which was achieved by combining a series of
speed catamarans during commercial ferry operations. Two test runs.

123
J Mar Sci Technol

400.0

300.0

200.0
micro strain

100.0

0.0
5 7 9 11 13 15
-100.0

-200.0

-300.0

-400.0
Time (s)

Fig. 8 Hydroelastic catamaran model raw strain gauge data, star-


board forward demihull Fig. 10 Vertical load on the centrebow, showing global wave loading
and slam loads
140

120 be excellent, with similar behaviour being exhibited in


100 both. Scaled finite-element modal analysis for the Incat
112 m gave a value of 2.1 Hz for the two-node mode,
Stress (MPa)

80
which was matched by the model. The zero-speed damping
60 ratio of 0.019 was close in value to that obtained for 86 and
40 96 m catamarans (0.035 and 0.018). The damping ratio f is
20
defined as:
0 lnðx1 =xmþ1 Þ
160 165 170 175 180 185 f¼ ð4Þ
2pm
-20
Time (s ec ) for f  1, where x1 and xm?1 represent the amplitudes of
Fig. 9 Incat Hull 050 raw strain gauge data [19] the signal at times t1 and tm?1 separated by a number of
cycles m. This result was somewhat fortuitous, since the
damping cannot be easily controlled, being largely asso-
3 Results and discussion ciated with the structure and outfit.

3.1 Dynamic behaviour 3.2 Slam identification

A key aim of model development was to be able to real- The slam load data, derived from the strain gauge mea-
istically simulate the slamming and whipping behaviour of surements in the centrebow (see Sect. 2.4), was an ideal
high-speed catamarans. It is critical that the model tests can signal to use to identify slam events in the time records.
be regarded as a good simulation of full-scale slamming. The centrebow only became loaded when it entered the
Raw strain gauge data from the starboard forward water and slam events could be clearly discerned from
demihull elastic link is shown in Fig. 8. The underlying global loads due to their distinct nature. This is illustrated
global wave loads can be clearly seen, as well as a series of in Fig. 10, where the load on the centrebow for a typical
slam events resulting in significantly larger strain values test run is shown. The event at 17.75 s is a slam event, as
(e.g. slam events at 7.5, 8.2, 12.3 and 13.6 s). Following exhibited by the rapid increase in the load due to the impact
these slam events, the transmission of energy into the with the water. In contrast, at 18.5 and 19 s there are two
structure to cause whipping of the model is visible, with the global wave loadings due to the immersion of the centre-
primary two-node longitudinal mode being excited. The bow in the water, but no slam event. A slam was defined,
modal frequency, shape and damping were ascertained using a similar approach as that taken with full-scale
using the strain gauge data from the demihull elastic links measurements by Thomas et al. [5], as having occurred if
[20] during impact tests. These data can be qualitatively the maximum load exceeded 10 N and the maximum rate
compared with that obtained from full-scale measurements of change of load prior to the peak was at least 5000 N/s.
on high-speed catamarans. An example of such data is This value was chosen since it gave a rate of change that
shown in Fig. 9 for Incat Hull 050 [18]. Qualitative clearly differentiated the slam events from the global wave
agreement between these two sets of data is considered to loads. This is illustrated in Fig. 11, where the same data

123
J Mar Sci Technol

The relative vertical bow velocity, vRV, was found by


calculating the difference between the vertical velocity of
the vessel at the CBT, vv, and the vertical velocity of the
water surface, vw:
vRV ¼ vw  vv ; ð7Þ
where
dhw
vw ¼ ð8Þ
dt
dz
vv ¼ : ð9Þ
dt
Note that in Eqs. 5, 6 and 7, the influence of the steady
Fig. 11 Time trace of the rate of change of vertical load, clearly and unsteady wave pattern due to the vessel motion is
showing four slam events at t = 17.75, 19.6, 20.1 and 20.5 s ignored, as it is assumed to be insignificant in relation to
the incident waves.
A sample of the results of the kinematic analysis is
shown in Fig. 10 have been converted to rate of change of
shown in Fig. 12 for a typical test run with heave at LCG,
vertical force on the centrebow; the slam events can be
pitch angle, wave elevation, relative bow immersion at
discerned even more clearly.
CBT, relative vertical velocity at CBT, slam force and
pressure at the top of the archway.
3.3 Kinematic analysis
3.4 Slam characterisation
Kinematic analysis was conducted for each of the identified
slam events to allow characterisation of the slamming
A total of 284 slams were identified in the data records and
process. A set of measurable kinematic parameters was
analysed. The majority of the slams were relatively low in
developed to help identify key relationships; these param-
magnitude, as illustrated in Fig. 13, which shows the per-
eters included the relative bow immersion and impact angle,
centage of occurrence of slams of varying magnitude for
the relative vertical bow velocity and the peak slam force.
both the speeds tested. This is a similar distribution to that
A reference point for many of these parameters is the aft
found from full-scale vessel measurements, for example on
truncation point of the centrebow, which is the aft end of
Incat Hull 042 [5]. It is clear that few extreme slam events
the centrebow where it terminates, aft of which the cross
occur in relation to the smaller events, with a total of only 3
deck structure is solely a flat wet deck. It is abbreviated as
slams having a slam load of over 150 N. The most severe
CBT and (as shown in Fig. 5) is 1902 mm forward of the
slam recorded was a load of 187 N, which relates (using
transom of the model.
Froude scaling) to a full-scale load of 1785 tonnes. Of note
The relative impact angle, hRI, was defined as the angle
is the difference in behaviour with vessel speed: more
between the centrebow archway and the water surface at
larger slam events appear to occur at the higher speed.
the CBT. It was calculated as the difference between the
The slam occurrence rates were determined by con-
wave angle hw and the sum of the pitch angle hp and
verting the number of slams found into the number of
the centrebow archway angle hCBA; see Eq. 5 below. The
slams per hour. The slam events per hour increased from
centrebow archway angle, which was the angle to the
318, for an operational speed of 20 knots, to 484, for 38
horizontal of the fore and aft run of the top of the archway
knots. It should be noted that the model was operating
at the CBT, was 7.7.
without a ride control system, so full-scale values of
hRI ¼ hw  hp  hCBA : ð5Þ slamming rates would be expected to be significantly
reduced.
The slope of the wave, hw, was found from the
A standard approach to explaining slam behaviour is
difference in wave elevation as measured by two of the
through the relative vertical velocity between the vessel
moving resistance wave probes.
and the water surface prior to impact. The relative vertical
The immersion of the model at the CBT, hCBT, was
bow velocity was determined for each slam event and is
calculated by determining the difference between the ver-
plotted in Fig. 14 in relation to the slam load. The data
tical displacement at the CBT (combining heave and pitch
show that a relative vertical velocity value of only 0.25 m/s
motion), z, and the wave elevation at this point, hw.
(1.7 m/s full scale) is required for a slam to occur. There
hCBT ¼ hw  z: ð6Þ does not appear to be a significant difference is the

123
J Mar Sci Technol

Fig. 12 Kinematic analysis


overview of a typical test run,
showing heave at LCG, pitch
angle, wave elevation, relative
bow immersion at CBT, relative
vertical velocity at CBT, slam
force and pressure at the top of
the archway

200
Percentage of occurrence

60 180 Fn = 0.32
50
160 Fn = 0.60
40
30 140
Slam load (N)

20 120
10
100
0 Fn = 0.60
80
Fn = 0.32
60
40
20
Slam load (N)
0
Fig. 13 Distribution of slam loads for various model speeds 0 0.5 1 1.5 2
Maximum relative vertical velocity prior to slam (m/s)

relationship between velocity and slam magnitude for the Fig. 14 Slam load as a function of the maximum relative vertical
two speeds tested. velocity prior to a slam occurring
Generally, it would be expected that there would be a
correlation between the relative vertical velocity and the for a particular relative vertical velocity, there is a large
magnitude of the slam. Whilst Fig. 14 shows a general range of possible slam magnitudes; for example, with a
trend that an increase in relative vertical velocity leads to relative vertical velocity of 1.1 m/s, the slam load could
an increase in slam load, the relationship is weak. In fact, realistically be of any value between 20 and 150 N. This

123
J Mar Sci Technol

200 200
Fn = 0.32 Fn = 0.32
180 180
F = 0.60
Fn
160 CBT 160 Fn = 0.60

140 140

Slam load (N)


Slam load (N)

120 120
100 100
80 80
60 60
40 40
20 20
0 0
1650 1700 1750 1800 1850 1900 1950 2000 -5 -4 -3 -2 -1 0 1 2
Slam location from transom (mm) Pitch angle (deg)

Fig. 15 Slam load as a function of slam location measured from the Fig. 17 Slam load as a function of pitch angle (positive is bow up)
transom
the majority of the pressure peaks are located in the
3
vicinity of the centrebow truncation and further aft, and
2.5 CBT there is very little slam pressure in the forward region of
the centrebow. In future tests it would be beneficial to
instrument the model with more pressure transducers, thus
Peak Pressure (psi)

2
allowing a more complete mapping of the pressures in the
1.5 archway to be achieved.
Another clear difference in slam behaviour for the
1
varying model speed is seen in Fig. 17. Here the pitch
0.5 angle of the vessel is plotted against slam load. For the
slower ship speed of 20 knots, the majority of the slams
0 occurred with a negative pitch angle, i.e. bow down,
1650 1700 1750 1800 1850 1900 1950 2000 2050 2100 2150
showing that the vessel is pitching down into the water as it
Location from transom (mm)
slams. There is a weak correlation between the slam load
Fig. 16 Slam pressure distribution along archway and the pitch angle, indicating that the larger slams at this
speed tend to occur with a larger pitch angle. In contrast, at
suggests that there are other important factors which are the higher vessel speed, the pitch angle at impact is gen-
influencing the magnitude of the slam load; possibilities erally close to zero or slightly positive.
include the immersion of the vessel and the relative angle These results suggest that the mechanics of the slam-
between the water surface and the hull. ming events are somewhat different for the two speeds. The
The majority of the slam forces are centred in the forward slam location and the bow-down pitch angle sug-
vicinity of the centrebow truncation. This is shown in gest that the slow-speed slam is more dependent on the
Fig. 15, where the slam location for each slam is plotted pitching motion to drive the bow into the water. The
against its slam load magnitude; the majority of slams combination of a small pitch angle and slam location fur-
occur within 100 mm of the CBT. This is a valuable result, ther aft suggests that a slam event at higher speed is more
as it provides designers with information on the zone in the akin to the vessel driving straight horizontally into a wave.
vessel most susceptible to high local impact loadings. Tests It is interesting to note that the resultant vertical velocities
in regular waves by Lavroff [21] showed a similar longi- for both types of slams are similar (Fig. 14).
tudinal location for the slam events. It is interesting to note It was proposed that the relative angle between the water
that the slams for the 38 knot condition tend to occur surface and the underside of the archway may have an
further aft than those for a full-scale speed of 20 knots. influence on the magnitude of the slam event. For example,
Mapping of the slam impact was achieved by using when a flat plate is dropped horizontally onto flat water, the
results from the pressure transducers located along the impact is significantly larger than if the plate contacts at an
starboard archway between the centrebow and the demi- angle to the water [22]. This was investigated by plotting
hull. The maximum pressure measured by each transducer the slam magnitude versus the relative impact angle at the
during a series of slam events is shown in Fig. 16. The CBT for each slam event; see Fig. 18.
vertical line indicates the truncation point of the centrebow This plot shows that there was in fact no clear rela-
and the start of the flat wet deck. This plot again shows that tionship between the relative impact angle and the

123
J Mar Sci Technol

200 100
180 Fn = 0.32 Fn = 0.32
160 Fn = 0.60 80

Immersion at CBT (mm)


140
Fn = 0.60
Slam load (N)

120 60
100
Centrebow
80 40
truncation
60
40 20 2D filling
height
20
0 0
-50 -40 -30 -20 -10 0 10 20 Max arch
height
Relative impact angle (deg)
-20
1780 1830 1880 1930 1980
Fig. 18 Slam load as a function of relative impact angle
Slam location from transom (mm)

200 Fig. 20 Immersion at the CBT for varying slam locations


180 Fn = 0.32

160 Fn = 0.60
140 max arch height
Slam load (N)

120
100
80
60
40
20
0
-20 0 20 40 60 80 100
Immersion at CBT (mm)

Fig. 19 Slam load as a function of immersion at the centrebow Fig. 21 Definition of 2D filling height, hf, based on area A1 equalling
truncation area A2

magnitude of the slam. This suggests that there are other interest is the immersion required for a slam at the cen-
factors which might have a greater influence on the slam trebow truncation, 44.4 mm. This relates to the two-
magnitude, such as venting of air from the archway and the dimensional filling height at the centrebow truncation. This
amount of mixing of air and water at impact. is defined in Fig. 21; the two-dimensional filling height, hf,
For a monohull vessel to slam, the vertical relative is reached when the bow is immersed so that the area of the
velocity needs to be greater than a threshold value and the rectangle (A2) with a width of the demihull centreline to the
keel of the vessel needs to enter the water [23]. In contrast, vessel centreline (b) and a height equal to the distance
a catamaran appears to need a relatively low threshold immersed is equal to the cross-sectional area of the arch-
value for a slam to occur (as shown in Fig. 14), and the top way A1. This physically represents the displacement of the
of the archway does not need to immerse to the level of the water by the demihull and centrebow into the archway
water surface. This is illustrated in Fig. 19, where the region as the bow immerses into the water.
immersion at the CBT is plotted for the slam events. The It is of interest that the slams at higher vessel speed
height of the archway at the CBT is marked as the vertical appear to require less immersion of the hull into the water.
line, showing that an immersion of 76 mm is required for This is consistent with the high-speed slams occurring
the archway to touch the water surface; however, it is clear further aft on the vessel with smaller angles of pitch.
that most slams are occurring with the vessel immersing The locations of slam events in the trace of wave ele-
less than this value. vation are shown in Fig. 22; the slams are marked as hol-
In Fig. 20, the immersion at the centrebow truncation is low circles. It appears that the majority of the slams
shown for each slam event plotted against its longitudinal occurred with the model impacting the wave just prior to
location. There is a clear trend here, with the immersion encountering the peak of the wave. This is shown more
increasing as slam location moves forward. Of particular clearly in Fig. 23, where each slam event is plotted based

123
J Mar Sci Technol

Fig. 22 Slam locations plotted


on a wave elevation time trace

200 200
Fn = 0.32 Fn = 0.32
180 180
Fn = 0.60
160 160 Fn = 0.60
Wave Peak
140 140
Slam load (N)

Slam load (N)

120 120
100 100
80 80
60 60
40 40
20 20
0 0
-50 -40 -30 -20 -10 0 10 20 30 40 50 0 10 20 30 40 50 60 70 80 90
Percentage of wave length prior to peak (%) Wave Height (mm)

Fig. 23 Slam load as a function of the percentage of wavelength Fig. 24 Slam load as a function of the height of the wave back
prior to wave peak

will allow the incorporation of slam prediction, using an


on its slam load and the position of the impact in the wave.
empirical basis, into time–domain seakeeping codes such
The majority of the slams occurred within 30% of the
as the 2d?t code BEAMSEA developed by Holloway
wavelength (for the wave enveloping each slam event)
and Davis [24]. For example, such codes may incorporate
either side of the wave peak. For the higher model speed,
the data on immersion versus slam location to determine
the majority of the slams occur just after the wave crest.
if a slam event has occurred. Then the slam load versus
In order to analyse the effect of the wave characteristics
the relative vertical velocity may be used to estimate the
on slam magnitude, the wave causing each slam event was
magnitude of the slam load. This will result in such
examined in detail, with the height of the face of the wave
codes having additional outputs of slamming rates and
that the model slammed into and the height of the back of the
slam loadings as well as vessel motions and global wave
preceding wave being determined. The relationship between
loads.
slam magnitude and preceding wave back height was found
In addition, the mechanism for the occurrence of slam
to be stronger than that with the slam wave face height. The
events with extreme levels of loading is clearer. For
relationship between face height of the wave and slam load
example, in the test data, the three largest slam events with
is shown in Fig. 24; generally, an increase in face height has
loadings between 180 and 190 N (approximately 1700
a corresponding increase in slam magnitude.
tonnes full scale) all exhibited the following characteristics:
The results of this investigation have provided detailed
information on the kinematic parameters relating the • Relative vertical velocity above 1.3 m/s
slamming process of large high-speed catamarans. This • Slam location just aft of the CBT

123
J Mar Sci Technol

• Immersion to, or just below, the 2D filling height vice versa, a comparison with full-scale slamming data is
• Slam at the crest of the wave. illuminating. The use of strain gauge measurements of a
96 m high-speed catamaran and finite element analysis [1]
However, it is apparent that other factors must be
resulted in the estimation of an extreme slam event of 93%
influencing the slam behaviour that have not yet been
of the vessel’s displacement; this corresponds with the
identified, since some slams can have all of the above
largest slam measured during the model experiments of
characteristics but do not result in a large slam loading.
71% of the vessel’s displacement. The distribution of the
Further work is therefore required to investigate such fac-
occurrence of slams of varying magnitude was found to be
tors as the venting of, and the aeration within, the archway
similar to the distribution found from full-scale vessel
during a slam, in order to determine their influence on slam
measurements [5].
magnitude.

3.5 Scaling effects 4 Conclusions

The use of Froude’s law is appropriate for model tests Model tests were conducted using a hydroelastic seg-
involving water waves, since it ensures that, for the model mented model of a wave-piercer catamaran to obtain data
and full-scale vessel, the correct relationship is maintained on slamming behaviour. The whipping behaviour of the
between inertial and gravitational forces [25]. Since full-scale Incat 112 m vessel was replicated in the model,
slamming loads on ships are dominated by gravitational with the correct longitudinal modal frequency being
and inertial forces, Froude scaling is therefore suitable. achieved as well as an appropriate level of damping.
The Reynolds number governs the inertial and viscous A total of 284 slam events were identified in the data
forces, but it is not possible to achieve both Froude and records and analysed against a set of kinematic parameters.
Reynolds scaling simultaneously in model tests. It is pro- The largest slam event recorded was equivalent to a full-
posed that the differences between model and full-scale scale load of 1785 tonnes.
Reynolds numbers are not significant for slamming tests, A general trend that an increase in relative vertical
since the flow in the archway during slamming is naturally velocity leads to an increase in slam load was identified.
turbulent (note that turbulence stimulators were fitted to the However, there is a large range of possible slam magni-
bows of the model). tudes for a particular relative vertical velocity. The
The oscillatory flow of the slamming behaviour is majority of the slam forces are centred in the vicinity of the
governed by the Strouhal number. If the frequency of centrebow truncation.
oscillation is taken as the motion frequency of the ship and There were some clear differences in the slamming
water waves, then this number was satisfied in the model behaviour for the two speeds tested. The slower-speed
experiments. Surface tension forces, as governed by the slams occurred further forward on the hull, prior to the
Weber number, are normally ignored in ship model wave crest, with a bow-down pitch angle and deeper
experiments. For slam events, the surface tension may have immersion. In contrast, the slams at 38 knots were more
a visible effect on the appearance of the spray generated by akin to driving into a wave with the slams occurring further
the model when compared with full scale, with the size of aft, at close to level trim, with shallower immersion at the
the model-scale droplets of spray being too large compared centrebow truncation and just forward of the wave crest.
with full scale. However, since the slam loads are domi- Immersion of the centrebow to the two-dimensional
nated by inertial forces, it is proposed that similarity of sectional filling height was found to be a better indicator of
Weber numbers for model and full scale is not required, as slam occurrence than immersion to the top of the archway.
the spray is not expected to be in direct contact with the This demonstrates that the immersion of the demihulls and
hull. centrebow contribute to the filling of the archways and the
Faltinsen [26] proposes that if slamming is associated occurrence of a slam.
with the formation of gas pockets, which may be the sit- The occurrence of slam events with extreme levels of
uation with slamming in the archway of a catamaran, the loading appears to be dependent on having a relative ver-
Euler number must be the same in model and full scale. tical velocity above 1.3 m/s; a slam location just aft of the
However, this is hard to achieve experimentally, and Fro- CBT; immersion to, or just below, the 2D filling height;
ude scaling results in a model-scale Euler number that is and the slam being at the crest of the wave. However,
larger than the corresponding full-scale value. This will additional factors must be influencing the slam behaviour
lead to conservative predictions of slam loads. that have not been identified in this current work, since
Whilst it is not possible to investigate scaling effects by some slams may have all of the above characteristics but do
replicating the model tests in full-scale measurements, or not result in a large slam loading. Further investigation is

123
J Mar Sci Technol

thus required to assess the influence of such factors as the 8. Riska K, Kukkanen T (1994) Speed dependence of the natural
venting of, and the aeration within, the archway during a modes of an elastically scaled ship model. In: Proceedings of
Hydroelasticity in Marine Technology. Balkema, Rotterdam,
slam. pp 157–168
It is expected that although the model was not fitted with 9. McTaggart K, Datta I, Stirling A, Gibson S, Glen I (1997)
an operating ride control system, the slamming behaviour Motions and loads of a hydroelastic frigate model in severe seas.
of the full-scale vessel will be similar to that of the model, Trans SNAME 105:427–453
10. Dessi D, Mariani R, Coppotelli G (2007) Experimental investi-
except that the full-scale values of slamming rates would gation of the bending vibrations of a fast vessel. Aust J Mech Eng
be significantly reduced. Another possible influence on the 4(2):125–144
results is that the model is towed as opposed to being self- 11. Troesch AW (1984) Wave-induced hull vibrations: an experi-
propelled. Full-scale data have shown that vessels can be mental and theoretical study. J Ship Res 28(2):141–150
12. Hermundstad OA, Aarsnes JV, Moan T (1999) Linear hydro-
slowed after experiencing a series of large slam events. elastic analysis of high-speed catamarans and monohulls. J Ship
The results will allow the incorporation of an empirically Res 43(1):48–63
based slam prediction model into the 2D?t seakeeping code 13. Ge C, Faltinsen OM, Moan T (2005) Global hydroelastic
BEAMSEA originally developed by Holloway and Davis response of catamarans due to wetdeck slamming. J Ship Res
49(1):24–42
[24]. 14. Wahab R (1967) Amidships forces and moments on a CB = 0.80
‘‘Series 60’’ model in waves from various directions (NSMB
Acknowledgments This work has been supported by Incat Tas- Report No. 100 S). Netherlands Ship Research Centre, Delft
mania, Revolution Design Pty Ltd and an Australian Research 15. Colwell JL, Datta I, Rogers R (1995) Head seas slamming tests
Council Linkage Grant. The authors would like to acknowledge the on a fast surface ship hull form series. In: Proceedings of the
contribution of Mr Kim Chamberlin at the model testing stage of the RINA Conference on Seakeeping and Weather. Royal Institution
work. of Naval Architects, London, pp 1–18
16. Kapsenberg GK, Brizzolara S (1999) Hydroelastic effects of bow
flare slamming on a fast monohull. In: Proc 5th Int Conf on Fast
Sea Transportation (FAST 99), Seattle, WA, USA, 31 Aug–2
References Sept 1999, pp 699–708
17. Okland OD, Moan T, Aarsnes JV (1998) Structural response in
large twin hull vessels exposed to severe wet deck slamming. In:
1. Thomas GA, Davis MR, Holloway DS, Roberts TJ (2003)
Proc 7th Int Symp on Practical Design of Ships and Mobile Units
Transient dynamic slam response of large high speed catamarans.
(PRADS 98), The Hague, The Netherlands, 20–25 Sept 1998,
In: Proc 7th Int Conf on Fast Sea Transportation (FAST 03),
pp 69–78
Naples, Italy, 7–10 Oct 2003, 2(B1):1–8
18. Thomas GA, Davis MR, Holloway DS, Roberts TJ (2008) The
2. Thomas GA, Davis MR, Holloway DS, Roberts TJ (2005) The
vibratory damping of large high speed catamarans. Marine Struct
influence of slamming and whipping on the fatigue life of a high-
21:1–22
speed catamaran. In: Proc 8th Int Conf on Fast Sea Transportation
19. Lavroff J, Davis MR, Holloway DS, Thomas G (2010) The
(FAST 05), St Petersburg, Russia, 27–30 June 2005, Sect 4,
vibratory response of high-speed catamarans to slamming
pp 33–40
investigated by hydroelastic segmented model experiments. Int J
3. Roberts TJ, Watson NL, Davis MR (1997) Evaluation of sea
Marit Eng (RINA Trans Part A) (in press)
loads in high speed catamarans. In: Proc 4th Int Conf on Fast Sea
20. Lavroff J, Davis MR, Holloway DS, Thomas G (2007) The
Transportation (FAST 97), Sydney, Australia, 21–23 July 1997,
whipping vibratory response of a hydroelastic segmented cata-
pp 311–316
maran model. In: Proc 9th Int Conf on Fast Sea Transportation
4. Watson NL, Davis MR, Roberts TJ (1997) Shipborne measure-
(FAST 07), Shanghai, China, 23–27 Sept 2007, pp 600–607
ment of sea conditions and seakeeping response of high-speed
21. Lavroff J (2009) The slamming and whipping response of a
ferries. In: Proc 4th Int Conf on Fast Sea Transportation (FAST
hydroelastically segmented catamaran model (Ph.D. thesis).
97), Sydney, Australia, 21–23 July 1997, pp 713–718
University of Tasmania, August
5. Thomas GA, Davis MR, Holloway DS, Watson NL, Roberts TJ
22. Smith NJ, Stansby PK, Wright JR (1998) The slam force on a flat
(2003) Slamming response of a large high speed wave piercer
plate in free flight due to impact on a wave crest. J Fluids Struct
catamaran. Marine Technol 40(2):126–140
12:183–196
6. International Towing Tank Conference (2008) Testing and
23. Lloyd ARJM (1989) Seakeeping: ship behaviour in rough
extrapolation methods high speed marine vehicles: structural
weather. Ellis Horwood, Chichester
loads (ITTC Recomm Proced Guidelines 7.5-02-05-06). ITTC
24. Holloway DS, Davis MR (2006) Ship motion using a high Froude
Secretary, Rio de Janeiro
number time domain strip theory. J Ship Res 50(1):15–30
7. Watanabe I, Takemoto H, Miyamoto T (1988) Use of an elastic
25. Faltinsen OM (2006) Hydrodynamics of high-speed marine
model in studying wave loads on a ship in waves and its verifi-
vehicles. Cambridge University Press, Cambridge
cation by a full-scale measurement. In: Proceedings of the
26. Faltinsen OM (2007) Challenges in hydrodynamics of ships and
International Symposium on Scale Modeling. Japan Society of
ocean structures. Brodogradnja 58(3):268–277
Mechanical Engineers, Tokyo, pp 47–55

123

Вам также может понравиться