Wave induced extreme hull girder loads on containerships
J. Juncher Jensen (V, Corresponding Author), Preben
Terndrup Pedersen (FL), Bill Shi (V), Sue Wang
(M),Martin Petricic (V), Alaa E. Mansour (FL)
This paper provides simple but rational procedures forprediction of extreme wave- induced sectional hull girder forces with reasonable engineering accuracy. The procedures take intoaccount main ship hull characteristics such as: length, breadth, draught, block coefficient, bow flare coefficient, forward speed and hull flexibility. The vertical hull girder loads are evaluated for specific operational profiles Firstly a quadratic strip theoiy is presented which can give separate predictions for the hogging and sagging bending moments and shear forces and for hull girder loads. Then this procedure is used as a base to derive semi-analytical formulas such that approximate wave load calculations can be performed by a simple spreadsheet program. Due to the few input parameters this procedure can be used to estimate the wave-induced bending moments ai the conceptual design phase. Since the procedure is based on rational methods it can be applied for novel single hull ship types not presently covered by the rules of the classdìcation societies or to account for specific operational profiles.
KEY WORDS Slamming, Whipping,
Wave-induced hull girder response, Quadratic strip Theory, Timoshenko beam,Stochastic seaway, Extreme loads
INTRODUCTION
The objective of this paper is to present a simplified procedure for determining wave hull girder loads acting on containerships including transient loads such as slamming and green water effects. The high frequency transient loads are combined with lower frequency wave induced loads, and the entire simplified solution is presented in closed-form equations. Whipping accelerations due to forward or stern slamming are also included in the calculation procedure.
The theoretical basis for the present analysis are the procedures described in [1], -[5], supported by information regarding the
effect of weather routing,
[6],[7], springing, [8], [9] and, stochastic correlations between wave and whipping responses, [IO].High frequency loads such as slamming and green water play an important role in the safety and operability of containerships. There is no well-established procedure for determining the total
response of ships to these loads combined with the low
frequency loads such as vertical bending moment. Becauseof
the random nature of these loads, the sOlutions can become complex and not suitable for routine engineering evaluatiOns. At present the main loads on merchant ship hulls are based on rules developed by the Classification Societies. For normal ship types the Classification Societies have developed empiricaldesign rules for sectional forces based on theoretical and
experimental analyses and feed back from service experience. The Classification rules for initial scantling selection do not explicitly take into account such factors as draft, hull flexibility, operational procedures etc. Only as a possible second step will non-lineas seakeeping analyses be performed [Il].SMTC-072-2008 Jensen
Assessment of non-linear sectional hull girder loads due to
impulsive wave induced loads due to slamming on flared bows, forward bottom parts and possibly flat stern areas of ships is an
important task during the design phase ofnovel ships.
Thus, for advanced vessels more specific wave-induced hull girder forces have to be determined. This is normally done by specialists using non-linear strip theories and in recent years also numerical calculations based on comprehensive 3D time simulations coupled with finite element models.
Therefore, for novel ship designs there is a need to develop new design oriented prediction tools which can be used in the iñitial
design phase and serve as a gauge for detailed
numerical calculations using numerical methods based on non-linear strip theory or comprehensive 3D time simulations [12].With this purpose in mind the overall aim of this paper is firstly to present a quadratic strip theory which is suited for numerical prediction of the effect of non-linearities due to bow flare and
for the effect of hull
flexibility and non-linear loads on
springing. Secondly, to show how to use this analysis tool toderive closed form estimates for the wave frequency hull girder sectional bending moments for ships with fine forms such as
containerships and at the same time to determine the hull girder whipping bending moments due to slamming. The moments caused by whipping are combined with the spatially distributed wave frequency wave-induced bending moment M,,,(x,i) taking into account the phase angles and the effect of damping. The resulting long term Weibull and Gumbel distributions for the peak values taking into account the operational profile of the ship is derived.
The result is a procedure which is more ship specific and
therefore more accurate than the present class rules but just aseasy to use.
The only ship parameters entering the closed form expression
are
Deift University of Technôlogy
Ship Hydromechanics Laboratory
Library
Mekelweg'2, 2628 CD Deift
The Netherlands
Length L of the vessel (either the rule length or the length overall)
Moulded (max) breadth B
Actual draught d (any trim is neglected) Block coefficient C5 (minimum 0.6) Bow flare coefficient C1 (ABS definition)
Freeboard F (used
only in green water loadcalculations)
Local dead rise angle a, breadth B,and draught d1 at the location of slamming
Two- and three-node natural periods T,, T3 (used in the decay rate of whipping)
Relative damping ,,
, in the two- and three-node
modes (used in the decay rate of whipping)In addition the operational profile must be specified interms of speed and heading distributions, operational areas, total timeat sea and, possibly, effects of weather routing.
THEORETICAL BACKGROUND
The basis for the theory used herein for the estimation of the
wave frequency bending moments and shear forces
is a quadratic strip theory formulated in the frequency domain which originally has been presented in [1]. Therefore, we shall only summarize the governing equations and the assumptionsinvolved.
Measurements have shown that the springing and whipping deflections are primarily in the form of the lower hull girder vibration modes. That is, the heave and pitch rigid body modes and the two and three node vertical vibration modes. Since it is generally accepted that the lower modes of ship hull vibrations can be determined quite accurately by modeling the hull as a single non-uniform beam, we shall base the equations of motion
in the vertical plane on the following Timoshenko beam
formulatiOn:aa
EI i +
'iI +
ax[
(
at}]
a
aw /-IGA(i +-
c)J=
m,r2.-5t2--ÍuGAl+17fJ!_
m F(x,t).-q(x,t)
at2 (2.1) =where EI(x) and pGA(x) are the vertical bending and shear
rigidities, respectively; q(t, x) is the slope due to bending;
w(x,t)
isthe total
deflection; i is an internal damping coefficient; and X is a longitudinal coordinate in an x,y,z -coordinate system fixed with regard to the undisturbed ship so that the z-axis is in the vertical direction. Finally, m(x) is thehull mass per unit length, mr2(x) is the mass moment of
inertia aboût the horizontaly-axis, F(x,t) is the low frequencyexternal force per unit length which is nonlinear in w and
q(x,t) denotes the transient slamming load
per unit length along the hull girder.The boundary conditions to the equations (2.1). express that the bending moments and shear forces are zero at the ends of the
ship.
FORCES BASED ON A QUADRATIC STRIP
THEORY
The quadratic strip theory is formulated in the frequency domain like the usual linear strip theory. This has the advantage that the correct frequency variation of the hydrodynamic added mass and damping terms can be applied and that the extreme value predictions can directly be performed using the well known methods for slightly non-Gaussian responses, e.g. monotonic transformations [13]. Thereby the computational effort becomes several magnitudes smaller thät than for nonlinear time domaiñ analyses. The drawback is that this perturbation procedure
canñot model very
severe non-linearities such as bottom slamming loads and green water loads. Such loads are here treated separately and included in the load term q(x,i).Calculation of the steady state hydrodynamic forces F(x,t) Us based on the time derivation of the momentum of the added mass of water surrounding the hull. In addition to forces due to a change in momentum of the added mass of water, we include a damping term and a restoring terni, both dependent
on the
relative motion. Thus the fOrce per unit length of the hull acting at position x is taken in the formF(x
DtL
DtJ
Dt-f
,ï
dzwhere the difference between the absolute displacement of the ship in the vertical direction, w(x,t), and the modified surface
elevation of the ocean, h(x,t), is denoted by ;(X,t)
, theadded mass per unit length by m, and the damping by N. The (3.1)
modification of the real wave elevation h
(x,
t) takes into
account the exponential decay of the
wave particle motion through the Smith correction factor K.The operator DI Dt is
the total derivative with respect to time t; that is
Di
ai
ax
where V is the forward speed of the ship. The breadth of the
ship is denoted by B(z,x) and the draft by T(x)
. TheFroude-Krylov fluid pressure is denoted by
p.
If we neglect the z dependence in
m, N and B, the force
expression corresponds to a linear strip theory. However, herewe shall,
as described in [I], evaluateF(x,t) by
a perturbational method, taking into account linear and quadraticterms in the relative displacement z and thereby
in the displacement of the hull w and the wave surface elevation h. In order to do this we shall start by evaluating the waterline breadth B, the added mass m, and the damping coefficient N around =O:az
B0(x) + ;(x,t)B1(x)
-am
m (,x) m (O,x) + z
h(x,t)=acosy.i
+a1a[(&
k1)cos(y,1
+,)-Jk
k
m0(x) +(x,t)m1(x)
(3.2)ql(x,t)
=
(t) a. (x)
w(x,t)=
u. (t) v,(x)
The wave surface elevation in the stochasticseaway h and the pressure
p
are also expressed as sums of linear terms and
quadratic terms so that we have, for instance
h(x,i)
=hw
+ h2
or
(3.5) where a1 are wave amplitudes and
y'1 = -
k, (x + Vt)
- a t + 6',
where k. denotes the wave number in the x-direction, and the
frequency a, is given by w
=
for deep water waves, where g is the acceleration of gravity.Similarly, the deflections of the ship hull are expressed as a sum of a linear part and a quadratic part
w =wW
+ w
These assumption lead to
F(x,,t)
=
F(»
+ F2
where F' are linear terms in the displacerient
w and the(I) (2)
wave surface elevation ., and are terms which are
quadratic in these quantities as well
isin the linear terms
squared.
DYNAMIC RESPONSE
The solution to Eq. (2.1) and the associated boundary conditions with the excitation given by Eq. (3.1) is approximated as a series
in the form
(4.1)
where u. (t) are coefficients to be determined
and where{v. (x),a1 (x)}
are eigenfiznctions to the homogeneous, self-adjoining part of Eq. (2.1). These eigenfunctions are assumed orthogonalized so thatnl
mr2 a a
+mv,v]dx=ÖM
(4.2)wherem m, + m11 and S, is Kronecker's delta. Furthermore,
the eigen-functions are assumed to be ordered so that {v0,a0} is the heave mode, {v1 , a1 } is the pitch mode, {v2 , a2 } is the two-node vibration mode associated with the natural frequency
2 and so on.
If we substitute (4.1) in the differential equation(3.1), and use the orthogonality relations (4.2),we obtain the following set of equations of motion:
SMTC-072-2008 Jensen
(3.3)
N(;,x)
N(O,x) +
N(x) + z(x,t)N1(x)
(3.4)
cos y'i
-vii)]
(3.6)
ii1
+cu
+ûu =
(4.3) where
j
= 0, 1, 2,From the
linear and quadratic transferfunctions of the
responses, the spectral densities and extreme values in statiònary stochastic seaways are easily calculated using for instance a Gram-Charlier or a Hermite series expansionto model the
slightly non-Gaussian response. Especially the [-lermite series expansion, suggested by Winterstein [13], seems to provide an effective procedure.In the following we shall present a simplified solution to this problem by breaking the response down into two parts.
The first part will be a simplified closed form
rigid body solution (j= O and 1) to the quadratic problem formulated above, neglecting transient loads, i.e. q(x,1) = O. This results in an analytical expression for the stochastic distribution ofwave-frequency sectional hull girder fôrces.
The second part will be a closed form approximation to the transient slamming induced hull girder sectional forces, i.e. including the transient slamming loads and green water loads through the load term q(x,t).
SIMPLIFIED
LINEAR AND NON-LINEAR
WAVE-FREQUENCy RESPONSES
Neglecting hull flexibility, i.e. taking intoaccount only heave and pitch modes (j =O and I) in Eq. (4.3) and neglecting the
bottom slamming pressure q(x,i) and including linear
and quadratic terms in the equations (2.1) and (3.1) - (3.7) given above the wave induced bending momentsMgdbodyqrjc and shear forces Qrigidbody,quadratic have been calculated for a numberof different ship types. Based on these comprehensive numerical calculations an estimation procedure of the long term wave-induced bending moments based on closed form expressions have been suggested in [2]. These expressions are based on analytically determined response operator functions for
simplified ship hull geometries. That is, the same type of
expressions as sought in the presentpaper.In [2] the influence of the ship principal characteristics on the linear part of the vertical wave-induced bending moment has been derived through an analytical analysis of a homogeneously loaded box shaped vessel. Closed form expressions for the
standard deviation of the linear part of the wave-induced
bending moment amidships for ships sailing in long crestedirregular waves modeled by the Pierson - Moskowitz (ISSC) wave spectrum were derived. The result is expressed as
SM =
¡IT
/g
I4ff2I(JßhI/3
It, 2L
Tg)
L)
where the significant wave height H and the zero-up-crossing
period represent the wave environment and
F(Fn)= I +3Fn2
is a speed dependence factor, valid for F,, <0.3,
F,(Cb)
=[(i
9) +0.6a(2_t9)];
9=2.5(1Cb); Cb=max(0.6,Cb)
is a block coefficient dependent factor and,
I1(i)=3OOO.exp(I4.1.i°
\ j; 6.9+O.I3fJz
I2(Y)=e(l+0.15y)
y=42..L;
V;
13(r)= (o.95_5.4r+1.14T2)
Finally,
fi
is the heading with fi = 1 80° for headsea.This analytical prediction of the linear
response has been compared with numerous results of direct strip theory calculations in [2] and good agreement is found. A 10percent reduction is usually assumed to account for short-crested waves,[19].
Similar closed form expressions are derived for the relative motion in [4],[5].
The frequency response functions cDv,, D9 for heave (w) and
pitch (, for
the vertical wave-induced motionsof a
homogeneously loaded box-shaped vessel (LxBxT) are
derived analytically by the linear strip theory. By neglecting thecoupling terms between heave and pitch and assuming
a constant sectional added mass equal to the displaced water, the frequencyresponse functions for heave and pitch become,[4]:cIç =IFI
cIo IGI (5.2)
with
2
(A2
2'
=[(1_2a2)
+kBa2J
J 2 .kL
F = KfsIfl(---)
24
[. kL
kL
kL
G = Kf
i sin(---) ---5---cos(--5----)
(k0L)2L[
2 2 2 where ke=IosßI
is the effective wave number and
f
=J(1_kd)2 (A)2
The Smith correction factorK 5 approximated by
K = exp(k0d)
(5.3)
(5.6)
The sectional hydrodynamic damping is
modeled by the
dimensionless ratio A between the incomiñg and diffracted waves through the approximation, [141A = 2sin(--kBã2)exp(_kdä2)
where a is related to the encounter frequency
W, the wave
number k, the wave frequency »= the ship speed V and the heading angleß
throughw = cokV cosßmlãjco
(5.8)
For block coefficient C8 less than one, the breadth B is replaced by BC8.
The frequency response functions for heave and pitch
accelerations are simply obtained by multiplication with
w2.
These frequency response functions for motions andaccelerations have been validated with model test results for a tanker, a container ship and two high-speed mono-hull vessels and reasonable good agreement is generally found, [4].
The phase angles between the motion components are found
from the equations of motion for heave and pitch.
Thecoefficients in the equations are the same, but the amplitudes of the forcing functions are different, and the two motions are9Ø0
degree out of phase. Due to the identical coefficients the phase
angles are the same as well. The phase angle of
the forcing function, i.e. the excitation force in heave or the excitation moment in pitch, relative to the incoming wave is1kd
f
A2Sifl(s1,)
kBä3f
COS(jor.e)where the coefficients follows from the f-function in Equation
(5.5).
The phase angle for the heave and the pitch motions relative to the forcing function is found from the inertia, the damping, and the restoring coefficient in the equation ofmotion:
(54)
cos(e,) = (1 2kdã2)i7
A2sln(ew)=_k_2
(5.5) where the coefficients follows from the ij function in Equation (5.3). It is seen that this phase angle is between0° and 180°. This ends the discussion of closed form expressions for the linear motions and loads.
The nonlinear part represented by the quadratic terms in Eq. (3.1) give separate predictions for hogging and sagging bending moments. That is, the contribution from momentum slamming caused by immersion and re-emergence of the linearly flared forward and aft part of the vessel is included in the quadratic
(5.7) strip theory described above and therefore also inthe resulting nonlinear sectional forces.
Non-linear numerical calculations in [I] show that the non-linear hogging bending moment peaks become slightly lower
than the linear prediction. On the other hand,
the sagging bending moment peaks become much largerand increase with the bow flare.In [21 an empirical procedure is developed
whereby first the skewness of the response is approximated by an analytical expression in the wave spectrum parameters H0,T, the flare coefficient C1 and the Froude number F;,:K3 =
O.26HC1Í1 expí_101
min(
-.5jcosßl)
(5.11) The flare coefficient C1
is here defined as the difference
between the upper deck area and the water plane area in the forward 20% of the ship hull normalized the freeboard in thisarea and with the ship length. Together with the analytical
approximation for the linear standard deviation of the bending moments sM this skewness is used to derive the probability distribution for the peaks of the wave induced sagging bending moments by application of Hermite series, [13]. Hence, the probability P that an individual peak M exceeds a given level mbecomes
SMTC-072-2008 Jensen
5
(5.9)
The main assumptions behind the present whipping analysis are: The rigid body ship motions are not influenced by the slamming impacts.
The dynamic transients are such that they are without significant effects from previous waves
The mathematical model for the flexible
hullis based on
Es.(2; 1) and we shall superimpose the effect of the bottom slammiñg load q(x,t) on the previously described nonlinear wave-induced loads calculated for the rigid vessel.Assuming the flexible displacements and velocities to be zero at the time of impact, t = O,the transient solütion to Eq. (4.3) for
j=2,3,... isfoundas
u. (t)
=_L je__T)
siny (tr)
q(x,r) v (x) dxdr
(6.1)
where
/3
ç2
=Jc? -fi12
(6.2) The vertical bending moment distribution afterthe occurrence of the slamming event is then calculated asM(x,t)
MÇjgjdbody quadratic(x,t)
+
± u,
[m,r2 (Ç)a
(Ç)
-
(xÇ)mv (c)] dÇ
(6.3) provided the slamming pressure is
acting forward of the
considered section at x. In Eqs. (6.1) and (6.3)a (x), y. (x)
are the same eigenfunctions as used in Eq.(4.-3).In the following we shall seek a simplified analytical solution to the transient bending moment expressed by Eq. (6.3).
Two types of slamming loads are considered. The first isbottom slamming which occurs when the relative displacementat any section exceeds the draught so that the local section emerges and then reenters the water. The magnitude of the bottom pressure impulse is here estimated from the slamming load per unit length. Its value can be derived from calculations perfòrmed in [151 for the instantaneous force q on the wedge:
q(t)
3'
p±3t
(6.4)4tan a
where the wedge angle is denoted a and with
p, i, t being
the density of water, the vertical relative velocity at the
slamming location, and the time measured from the wedge apex hitting the water, respectively. For bottom slamming the time at maximum immersion including account for the water up rise onJ=2
P[Mm]=expí!u2(m))
(5.12) where u is given by1+ .Ji + 4(z
+
2 1(3-
m i - -i with)=--, m
=-
andr=(V1+/c3 /18)
6sMr
In thesubsequent analysis the hogging bending moment is taken
as the
linear result. Whipping can make this assumptionsomewhat non-conservative, but further evaluations of full-scale data are needed to clarify this topic.
WHIPPING ANALYSIS
Assessment of non-linear sectional hull girder loads due to impulsive wave induced loads caused by slamming on flared bows, forward bottom parts and possibly flat stern areas of ships is an important task during the design phase of novel ships. In the following we shall analyze the combined effects of such slamming induced forces and the wave-induced rigid bodyloads derived above. Traditionally two types of slamming forces have been considered. That is a) bow flare impact which is modeled by the momentum slamming effect due to rapid immersion of the flared bow and b) the bottom slamming effect caused by sudden impact of the relatively flat stem or stern bottom. Part of the non-linear effect on the hull girder sectional forces due to immersion of the flared bow is included in the quadratic strip theory and is the main cause for the skewness given by Eq. (5.11). Since the rise time for this momentum slamming is of the order of the wave period i.e. typically with a rise time of I s
or more the dynamic effects of this momentum slamming
caused by the linearly flared bow section
is very limited. Therefore, in the following transient response analysis we shall only consider bottom slamming and slamming caused by large changes of curvature of the bow above the still water plane with rise times which are relatively small compared to the lowest hull girder natural period. That is, we shall consider situations where a slamming impact force is delivered on to the stern or stem of the ship and produces a shutter throughout the entire hull. This results in whipping stresses and a sudden deceleration, which are superimposed to the steady state wave-induced sectional hull stresses and accelerations. Whipping caused by emergence of the bow or stern which acts like a negative bow flare is not considered in here. But this may have an effect on the whipping response of full form ships.The calculation of slamming effects requires consideration
of
hull flexibility. The maximum slammiñg loads do not typically occur when the wave induced loads are the largest, and suchphasing needs to be considered in the combination of the
calculation of the combined load effects.a section with dead rise angle a and local breadth B, is
approximately
2B tana
(6.5)
3
2z
taking into account that the water up-rise is about half the
immersion of the section according to the results shown in [15].
The second type of slamming load is caused by a rapidly
decreasing angle of the bow sections with the horizon, both in the transverse and in the longitUdinal direction. That is the angle the wave hits the shell plating under. It is noted that the quasi static calculation procedure presented above includes the 2D
effect of a linear increase of the waterline breadth with
immersion and that the momentum bow flare slamming considered here is confined to the effect of the large waterline increase most container vessels have in the fore ship above the
still water level in order to increase deck area and to protect cargo on the deck from water impact. A mathematical
representation of this momentum slamming loadcan be derived
considering the bow forward of FP as a simple half cone (which often will model the bow fairly well since it is surface which can
de developed from flat plates). Then, the momentum impact of that body can be determined by a von Karman approach where
the added mass of a cone is taken to be equal to added mass of a
disc having the same radius as the cone's penetration of the still water plane, i.e.
r
=z cotß
where z is the relative immersionand ß is angle between a horizontal plane and thecone surface. The added mass of a disc in an infinite fluid is 8/3 ? p Thus for a half cone neglecting the Wagner correction, we find
M=pr
According to [16] the slamming load terms proportional to the forward velocity Vplay a less significant role for the whipping loads. Therefore, neglectingthe effect of forward speed Von the slamming load for a given vertical velocity ± we find the impact of the bow flare slamming load as
'impac: =
f F(t)dt
=
f D(Mi)
=
p(z z)cot3ßi
=p(,3
ij3)±
The problem now remains to choose the vertical position of the section which generates the bow flare slamming impact. This position will depend on impact velocities, natural frequencies of
the hull girder, the flare angle ß andso on. Numerical studies
are needed to determine appropriate values.
In the following it is assumed that in order to develop a
slamming impact the vertical velocity shall also be larger than the classical threshold value by Ochi. In addition the duration of
SMTC-072-2008 Jensen
or
the slam t, must be small compared to the relevant whipping
period in order to cause dynamic amplification. Hence, for
bottom slamming we have
=
max
(O.o93JZ
--
tana)
(6.7)Denoting the effective longitudinal extent of the slamming area
ßL and the longitudinal centre of pressuréx0 then the weighted
impulses of the bottom slam, see Eq.(6.4) is:
I,
=
v,(x0)ßL
q(t)dt
=
--4-ßL. p. v(x0)B,2±
(6.8)Now, using eigenfunctions (a,,v,) normalized such that the
vertical deflections V,
(L)
equals one, then the uncoupled onedegree of freedom equations (4.3) can be expressedas
ü,
+1,2u,
=P(t)/M,
(6.9)where
=17/
Ç,;
M,
=r[m,r2a,2 +m,v,]
dx
and
¡(t)
denotes the generalized slamming load, i.eJ(t)
=
v,(x0)ßLq(t)
Let us introduce the damping corrected frequency
û
=cJi _2
, then the free vibration solution can bewritten
u, (t)
=
u, (0)e" cos(,t) +
z,(0)+ ,û,u,(0)
e" sin(O.t)
(6.10)û,
Let us also introduce a characteristic time 7. as the smallest
positive value of t satisfying the condition z, (t) = O for
(6.6) u,(0) = O and ú,(0) O. Then by integration Eq. (6.10) we
get the following expression for the time T to maximum values of the coordinate fUnctionsu,.
û,(7.)
=
O= 4,Q,
z(0)
sin(,7.)
tan(o,.T )
' =Jj
Let us now assume that a slamming load P(t)> O is acting
during a time t.. where Ot,
Let us also assume that no whipping motion is present at t=
O, that isui(0)
=
ù1(0) =OThen we can integrate Eq. (6.10) withrespect to time t and get
ú1(t5)=
.!!_
2û1u1 (ti) Ç2
' u1(t)dt
with
Ii
=
j(t)dt=.ßLpv,(xo)B,2±
If t,
«7
then we find the following approximation to the velocity at the end of the application of the slamming pressureI. u. (t5) =
After the time of impact the motiòn will continue
as a free
motion described by Eq. (6.10) . Since 15« T we can use the approximation that u(0)=
O and that ii, (0)I, I
M1. The maximum modal displacement is then found for t= asUI =
e°'
.sin(O 1T )' Ç21A4,
where Eq.
(6.11)has been used
together with standardtrigonometric relations.
Note, that T, denotes the rire time from the time of impact to the maximum value of the response. For superposition with the wave frequency induced loads also the phase angle for the time of maximum impact pressure needs to be considered. Here these
phase lags are taken relâtive to peaks in the wave-induced
sagging bending moments. Positive means that the whipping bending moments occur before these peak values. With this definition the phase lag for maximum bottom slamming pressure is taken as 2 7r ¡3 and the phase lag for the momentum bow flareslamming is taken as it /6. These phase lags are based
on numerical calculations using a non-linear strip theo!)'program.We shall now use this result to derive an approximation to the spatial distribution of bending moments along the hull girder for the case where the numerical hull girder bending frequencies fl and mode shapes are not known. For the lowest natural vibration mode shapes the shear deformations can be neglected. Then the
two-node vibration mode for a free-free beam with constant (6.11) mass and stiffness properties is
(6.12)
y (x)
2(sinh k + sin /cx)(cosh h/cL - cos kL)
2(cosh kLsinkLsjnh kLcoskL)
(cosh /cx + coskx)(sinhkL sin kL)
2(coshkLsinkLsinhkLcoskL)
where ¡cL =4.73004074 and the mode shapes are normalized
such that
v2(0) = v2(L) O. This solution satisfies theorthogonality conditions (10) with heave and pitch motions for the two first modes.
Here a more realistic non-homogeneous mass distribution m(x) is needed for calculation of the spatial distribution of the hull
girder sectional forces. For a symmetric
mass distribution,increasing linearly from zerb to 0.3L forward and aft
andconstant on the remaining 40 per cent middle section, the
resulting mode deflection becomes as shown in Figure 1.05
-OE5
Figure 1. Mode shape for the two-node vibration mode fora uniform (thin line) and a non-uniform beam (thick line)
This two node vibration mode for the non-uniform beam can be approximated by y2 (x) L )3
1.23
-
4.58(..J
Iv2(x+)= v2(--x)
O-O.5
Lusing a least-square fit. Similar results for the 3-node modeare
shown in Figure 2.
(6.13)
Figure 2. Mode shape for the three-node vibration mode for a uniform (thin uñe) and a non-uniform beam (thick line)
The maximum values for the accelerations in the individual vibration modes at the forward perpendicular given by
.(x,T) =
"
" (\ D
C
24
(6.15) Assuming that the wave frequency wave induced acceleration and the whipping induced accelerations are normal distributed with Rayleigh distributed peaks
we can express the most
probable largest value, MPL, of each of the three acceleration components in a short term sea state with duration T asMPL
= s0 (x)2 in
-MPL..
whip,inode ace=
ßL
24 ¡iM
¡=2,3
M1T
°5s±vI(xo)I(x)I21n(77h.
(6.16) Here we have taken into account the relative occurrence of whipping, and s. is the standard deviation of the linear wave induced acceleration and ; the associated standard deviation of the relative velocity at the centric of the slamming pressure. Still neglecting inertia effects associated shear deformation the approximate whipping bending momentcan from (6.3) be
determined asMwhipping (x,t)=
-X[
(x
-
Ç) m,v (c)]
dÇ
and the maximum values of whipping bending moments can be expressed as
SMTC-072-2008 Jensen
0 0.1 0.2 0.3 0.4 0.5 0.5 0.1 0.8 0.9
Figure 3. Normalized bending moment variationS2(x)/LM2 for a uniform (thin line) and a non-uniform beam (thick line) in
two-node mode 0.5 0.8 H 'rIII:l¼ 07 08 0.8 Mwhjpping (x, T) =
ßLpv(x0)
(6.17) where S,.(x) =fm(,(u)(u_x)du
The corresponding result for the shear force is
N 2
Q(x,T)=
fJ3Lpvj(xo)Bj±(xo)Le21TG(x)
G1(x) = f m(u)v,(u)du
(6.18) Based on the approximation to the two node vibration mode depicted in Figure 1, the variation of S2(x)/LM2 and G2(x)/M2
are shown in Figure 3 and Figure 4, respectively.
It is noted that the variation S2 (x)/L M0 for thenon-uniform beam
approximation of the hull rather well resembles the numerjòaj results for the spatial distribution of the wave induced bending moments obtained from the second order strip theory.
Figure 4. Normalized shear fòrce variationG2(x)/M2 for a uniform (thin line) and a non-uniform beam (thick line) in two-node mode
As the whipping bending moment depends linearly on the relative vertical velocity the standard deviation
Smwhipping
(x,7)
= N .2S.(x)
, (6.19)ßL2Bips2(xo)v(xo)ç2
e'''
1=2
24
LM,
can be added directly to the standard deviation of the wave-induced bending moment. It is noted that the correlation is taken
care of by the exponential decay factor in the
whippingresponse. The standard deviation s (x0) of the relativemotion
at the location x0 for a, ship sailing in ocean waves modeled by JONSWAP or Pierson-Moskowjtz spectra is determined as described in Section 5 of the paper by Eqs. (5.2)-(5.10).
The relative occurrence of whipping due to bottom slamming is taken as
I
i(v
2Tlwh,i
=exp{--I
2 s2
where the last term follows from the requirement that the bottom must be out of the water prior to a slam. Fiñally, the probability P(MW+Wh > m) that
the combined wave and whipping
bending moment exceeds a value m in a stationary stochastic condition is written(6.20)
P(MW+Wh> m)
=(1
7Wh,2)P(M > m)(6.21)
+(wh2 l7Wh,3)P(MWh2 > m)+7Içh3P(MWh3 > m) Here P(MW > m) is calculated using the standarddeviationSM
and skewnessic3as calculated by Eqs. (5.1) and (5.11), whereas
P(MWh > m) applies the same skewness, but
a standard deviation equal to + SbS .The subscript i = 2,3 sign ifs' the relevant 2 and 3 node properties (natural frequency, damping factor, mode shape and bending moment variation).P(M,VhI > m) is calculated with the 2-node (i = 2) or the 2 and 3-node modes (i =3) included in Eq (6.19).
For container vessels green water loading on the forecastle deck
is of major concern since green water can cause various
damages to deck equipment and damage or even throw
contaiñers over board. Besides, the vertical component of this loading can add to the dynamic vertical hull girder loading. To evaluate this longitudinal load effect simplified expressions fôr the effect of green water loads on the wave-induced hull girder bending moments are determined principally as in [3] and added to the combined response assuming bottom slamming and greenwater loads are mutual exclusive. The only modification is that a threshold vertical velocity
(O.O93jZ)
is included together with a dynamic amplification factor of0.5 and the phase lag istaken to be 5
ir
/3. Green water loads are fürthermore onlyconsidered on sections in the forward twenty per cent of the
vessel.
LONG TERM PREDICTIONS
The long term analysis deals with the probability distributions of the extreme peaks taken over a period of typically 20 to 25 years. The basic assumption in the analysis is that the process
can be modeled as a sequence of stationary
processes with independent peaks. Witha given operational profile and a
relevant wave scatter diagram the long term distribution of the non-linear wave-induced wave frequency loads andaccelerations and the whipping loads can be calculated. The operational profile specifies the relative occurrence of
short-term conditions with constant forward speed, heading,
significant wave height and up-crossing period. Wave scatter
diagrams for the North Atlantic, the North Pacific, or the
Asia/Europe trades are generally used. Theselong term scatter diagrams can be modified in order to account for the effect of weather routing. Significant effects of weather routing on the long term wave scatter diagram have been found in [6], [7] and [18], reducing the predicted extreme values significantly. Having determined numerically the long-term individual peak distribution of a response R by averaging as discussed above, itis
often useful to approximate the result to an analytical
distribution. Here the Weibull distribution
P[Rr]=.LxpI_(!
y(0)
(7.1)(
a)
with the two free parameters a, b generally provides a good fit. v(r) is the up-crossing rate. From this expression the extreme value distribution follows from the Poisson up-crossing model or, asymptotically, the Gümbel distribution:
P[max(R)
<rio
<t <T]
=exp {v(r)T}
=
exP{_Nexp[_(L)J}
exp{exp(_c(r r,))};
rmp,
=
a/i7;
c
=
(1n N)1Here T and N are the total time and total
number of peaks, respectively. rmp, isdenoted the most probable maximum
response and has a return period of lITVALIDATIONS
A comprehensive survey of the literature was conducted on
full-b)
scale measurements of global hull girder loads and stresses on container ships. The objective of the
search was to identif'
results that are suitable for comparison with results predicted by the procedure developed in this paper.The search revealed that there exist some full-scale data, but few were reported in such details that they could be used to conduct an accurate comparison with the analysis results developed in
this
project. The main reason in most cases
is missinginformation necessary to conduct the comparison, such as some particulars and geometric characteristics of the containership. In
other instances there was
no or little
information on the environmental and operating conditions (speed and heading) under which the full-scale measurements were collected. Two studies were identified where only few assumptions had tobe made in order to complete the input required for the
procedure.The first study considers the thorough measurements described and analyzed by Okada et aI [17]. Three years of measurements have been recorded on board a container ship on route between Europe and Japan. The recorded data includes strain and wave height measurements. It is
shown that the wave height
distribution resembles those found from standard scatter diagrams for this route However, wave heights larger than 10m have not been measured. In thepresent study, the AsialEurope scatter diagram is therefore applied together with the weather routing modification detarmined in [6], [7]. Similar or even greater differences between measured wave data and standard scatter diagrams are observed in [18]. The operational profileis taken in accordance with the fairly detailed information given in [171, basically assuming service speed (24.5 kt) when the significant wave height is less than 6 m and a reduced speed of 6 kt for higher sea states. The heading is taken as slightlynon-uniform with larger weights
on head and near head sea
conditions.The main dimensions of the container ship
are length L Lb=
283.8 m, Breadth B = 42.8 m, draft d= 14 m and depth D = 24.4 m. The block coefficient Cb and the bow flare coefficient (ABS definition) C1 are not known, but takento be 0.6 and 0.2,
respectively, as these are typical values for
làrge container vessels. The section modulus W at the position of the strain gauges is W =47.183 m3 (from private communication with Dr. Okada). Hereby, all the required input data is defined for the present procedure as regards the rigid body response. However, reference [17] also contains a shOrt discussion of the importance of hydro-elastic effects (whipping and springing). They find whipping single stress amplitudes as high as 30 MPa and by use of a cut-off frequency below the two-node frequency (at 0.68Hz), they derived numerically from the
measured records individual peak stress distributions for the deck stresses with and without whipping (and springing) included. The results are presented in their Fig. 33, showing a marked increase in the extreme stresses (at 10 lével) when whipping is accounted for. These results can, to some extent, be used to validate the present procedure and it has been found that using the followingdata: For the slamming impact calculations the local bottom breadthB, 20 m, local draft d, =10 m, local dead rise
angle a =5
degrees and relative longitudinal extent fi = 0.07 yield results iñ fair agreement with the measured values, Figure 5. lt is noted that the. comparison deals with stress amplitudes defined as the average of the sagging and hogging stress amplitudes. The hogging peak following a sagging peak is taken as 80 per cent ofthe hogging peak at the same probability level in order to
account for the correlation between sagging and hogging peaks. This percentage ischosen from the
linear autocorrelation function of the bending moment. Furthermore, the calculated stresses in long-crested waves are reduced by IO per cent to account for real ocean waves being short crested, [19].With this in mind the example shows that the present procedure yields predictions in good agreement with available measurementsboth for the rigid body and the whipping bending
moment response. Especially it isnoted that the rapid increase in
whipping stresses at low probability levels (10to 1O) is
captured by the present procedure. No theoretical calculations are included in [17].Long term distnbutn of longmidìzml tending stress withc1d-offfrcqcy 03 and 1.0Hz.
Figure 5. Comparison between measured and predicted stresses in the deck of a container ship.
Full-scale measurements presented by Miyahara et al. in [181 have been collected over a period of two and a half years on board a post-Panamax containership on route between Japan and Europe. The main dimensions of the ship
are length L L=
281.4 m, Breadth B = 40.0 m, draft d l2.Sni and depth D = 24.2 m. The block coefficient Ch and the bow flare coefficient (ABS definition) C1 are not known from [18], but are taken tobe 0.6 and 0.2, respectively, as in the previous comparison. The section modulus W at the position of the strain gauges is not given in [18] and is, therefore, calculated using ABS Rules which give a value of W = 46.63 m3. For the calcülátion of bottom slamming loads,, the longitudinal extent of slamming is taken to be ß= O6L, the local breadth of slammingB, = 15 m, local draught d, = 12.5 m and the deadrise angle at slamming a = 10 deg. The operational profile of the vessel is derived from thedata given
in [18]. Frequency-speed histogram has beenmodified to fit the input format of the
spreadsheet program. Also, the time spent in harbor (with zero speed) was Climinatedulalive Pmbabiisy
i:
os i .uo7
i ro
i .1o5 i E404 i i no2iioi
LEq.ctSMTC-072-2008 Jensen
Il
12
E
from the analysis. Heading-wave height histogram has been normalized with respect to the time spent in each speed interval. An Euro-Asia scatter diagram is applied together with four different weather routing schemes, namely: I) weather routing derived from [18], 2) weather routing based on observations on containership in North Atlantic, 3)
weather routing with
stepwise decrease of the significantwave height and 4) no
weather routing. Scheme 3) was generated to furtherdemonstrate the effect of weather routing on the results. Scheme 4) was used as the default weather routing scheme in subsequent sensitivity analysis.
Figure 6 shows comparison
betweenmeasured and predicted stresses in the deckof a containership for all four weather routing schemes. Whipping contributions are not included as they were filtered out from the measurements presented in [18]. In general, the resultscompase
well with the full scale data, provided a reasonable weather routing scheme, 2) or 3), is applied.
a
n-E
î
Cumulative Probabi ity
Figure 6. Comparison between measured and predicted stresses in the deck due to hogging and sagging BM for various weather routing schemes
SENSITIVITY STUDIES
The sensitivity analysis was performed on the containership from [18] in order to assess the impact of various slamming parameters, scatter diagrams, operational
life and weather
routing schemes on the cumulative probability of the bending moment (in hogging and sagging) and on the most probable value of the bending moment. The values stated in section 8 were used as the default values. These are the results of this analysis:1) Impact of the extent of slamming
!' 10000
9000 8000'.
7000 6000 o . 5000 0.03*-- Extent of amming O.06L (&fault) s--- Extent of amming O.04L
a--Extent ofslamming=O.08L
Cumulative Probability
Figure 7. Cumulative probability of the stress in the deck due to hogging and sagging BM for different values of the extent of slamming
0.04 0.05 0.06 0.07 0.08 0.09 Extent of Slamming [% of L]
Figure 8. Variation of Most Probable Value (MPV) of BM with extent of slamming
2) Impact of the local breadth of slamming
Cumulative Probability
Figure 9. Cumulative probability of the stress in the deck due to
Scale Measurements as in North Atlantic R.
WRasrnref[lô]
a-- Full
....
4--W R.
No WL:
i jPO-= ilE Local Breadth = 15m (default)
e Local Breadth = IO m
aLocaJBreadth=2Om
-
hogging and sagging BM for different values of the local
breadth of slamming 12000 10000 8000 6000 4000 2000 o 8 10 12 14 16 18 20 22Local Breadth of Slamming [m]
Figure 10. Variation of MPV of BM with local
breadth of slamming3) Impact of the deadrise angle at slamming
Cumulative Probability
Figure 11. Cumulative probability of thestress in the deck due to hogging and sagging BM for different values of the deadrise angle at slamming 8000 7500 7000 6500 6000 0 Cumulative Probability
Figure 14. Cumulative probability of the stress in the deck due to hogging and sagging BM for different scatter diagrams
7000
6000 5000 4000 3000 2000 000- -
.
---
Sagglnge Hoing
-*-- Deadrise Angle at Slam. = IO deg (default)
Deadrise Angle at Slam. = 5 dog Deadrise Angle at Slam. = IS deg Deadrise Angle at Slam. = 20 deg
--
---*
o. i. 07
I. ;tLÌ_r!'''oI
I.,
Euro Asia (default)
Atlantic Pacific
*
-North+--VNorth
SMTC-072-2008 Jensen 13 5 10 15 20 25Deadrise Angle at Slamming [deg]
Figure 12. Variation of MPV of BM with deadrise angle at slamming
4) Impact of the operational life
O iO 15 20
25 30 35
Operational Life [years]
Figure 13. Variation of MPV of BM with operationallife
0
Eurn - As North Atlantic North Pacific
Operatknal LÎ [25 years]
Figure 15. Dependence of MPV of BM on scatter diagram
CONCLUSIONS
The paper presents simple but rational
procedures for the prediction of dynamic hull girder responses to wave frequency load and slamming load. The developed procedures have been applied to container carriers and the results are comparable with full scale measurements. The following findings are drawn from the analysis results of the applied vessels:Closed forni expressions for the wave bending moment as those presented in the paper can predict measured full-scale results with reasonable accuracy. The formulas take into account main ship hull characteristics such as: length, breadth, draught, block coefficient, bow flare coefficient, forward speed and hull
flexibility.
The effect of flare coefficient on slamming loads and other dynamic loads is important. In general, the larger the flare coefficient, the larger the bow slamming loads.
The impact of weather routing on life-time maximum loads can be important, but the effect ofoperational profile in terms of ship speed/heading combination is less important.
The impact of the choice of scatter diagram used in estimating the wave loads is important. In general, North Atlantic and North Pacific ship routes give approximately the
same magnitude
of
life-time wave loads, whereas the Asia/Europe route produces considerably lower values of waveloads.
It is important to include the influence of whipping stresses in the determination of the overall response.
The present paper deals with easy prediction of short and long term extreme loads in the vertical plane, i.e. the symmetric modes. Development of a similar procedure for estimation of the fatigue loading must in addition to the procedures described here also include research leading to a rational but similar simple procedure for calculation of hull girder springing caused by linear and non-linear wave excitation.
Sasng
Hogg,n8
Another subject for future research could be development of initial design oriented procedures for thecoupled horizontal bending and torsion response. For the case of containerships and open bulk carriers these anti-symmetric modes are important for fatigue evaluation of hatch corners on the deck structure, see for
instance [12].
DISCLAIMER
The views expressed in this paper are of the authors and do not necessary represent those of the American Bureau of Shipping.
REFERENCES
1. Jensen, J.J., Pedersen, P.T. "Wave-induced
Bending Moments
in Ships- a Quadratic Theoiy" Trans. RINA, Vol 121, 1979, pp
151-165.
Jensen, J.J., Mansour, A.E. "Estimation of Ship Long-term Wave-induced Bending Moment usingClosed-form Expressions". Trans. RINA, pp. 4 1-55, 2002.
Jensen, J. J., Mansour, A.E. "Estimation of the Effect of Green
Water and Bow Flare Slamming on the Wave-Induced Vertical Bending Moment Using Closed-Form Expressions", Proceedings
3rd
International Conference on Hydroelasticity in Marine Technology, pp. 155-161. Ed. R. Eatock Taylor, The Oxford University, Oxford, Sep. 2003.
Jensen, J. J., Mansour, A.E., Olsen, A.S. "Estimation of Ship Motions using Closed-Form Expressions", Ocean Engineering,
Vol. 31, pp 61-85, 2004
Mansour, A.E., Jensen, J.J., Olsen, A.S. "Fast Evaluation of
Container Securing Arrangements", Proc.PRADS'04,
Travemünde, pp 577-585, September 2004
Olsen, A.S., Schrøter, C., Jensen, JJ. "Encountered Wave Height Distributions for Ships in the North Atlantic Sea", Proc.
PRADS '04, Travemünde, pp 1043-1050, September 2004 Olsen, A.S., Schrøter, C., Jensen, J.J. "Wave Height
Distribution Observed by Ships in the North Atlantic", J. Ships and Offshore Structures, Vol. 1, No. 1, pp 1-12, 2005
Jensen, J. J., Dogliani, M. "Wave-induced ShipHull Vibration in Stochastic Seaways", Marine Structures, Vol. 9, pp 353-387, 1996.
Vidic-Perunovic, J., Jensen, J.J. "Non-linear Springing
Excitation Due to a Bi-directiona] Wave Field", Marine Structures, Vol. 18, No.4, pp 332-358, 2005
IO. Baarholm, G.S., Jensen, J.J. "Influence of Whipping on Long Term Vertical Bending Moment", J. Ship Research, Vol. 48, No.4, pp 26 1-272, December 2004.
ABS, Guidance for "SafeHull Dynamic Loading Approach" for Vessels 2006
Hirdaris, S.E.; Miao, S.H.; Price, W.G. and Temarel, P. "The influence of structural modeling on the dynamic behaviour
12000 10000 8000 6000 4000 2000
t
of a bulker in waves"
, Fourth International Conference onHydroelasticity in Marine Technology, Wuxi, China, pp 25- 33.
2006.
Winterstejn, S.R "Nonlinear Vibration Models for Extremes and Fatigue", AS, J. of Eng. Mech.Div., Vol 114, No. 10,
pp. 1772-1790, 1988.
Yamamoto, Y., Sugai, K., moue, H., Yoshida, K., Fugino, M., Ohtsubu, H. "Wave Loads and Response of Ships and Offshore Structures from the Viewpoint of Hydroelasticity', Proc. mt. Conf, on Advances iii Marine Structures, Admiralty Research Establishment, Dunfermline, Scotland, 1986
IS. Zhao, R., Faltinsen, O., "Water Enty of Two-dimensional
Bodies", J. Fluid Mechanics, Vol. 246, pp 593-6 12, 1993.
Ramos, J., Incecik, A., Soares, C. Guedes: Experimental study of slam-induced stresses in a containership Marine Structures, Vol.
13,no 1, Jan, 2000, pp 25-51
Okada, T., Takeda, Y., Maeda, T. "On board measurement of stresses and deflections of a Post-Panamax containership and its feedback to rational design", Marine Structures, Vol. 19, p
141-172, 2006
Miyahara, K., Miyake, R., Abe, N.,Kumano, A., Toyoda, M., Nakajima, Y. "Full-scale Measurements on Hull Response of a Large Container Ship in Service", Proc. OMAE2006, Paper no.
92233, June 4-9, Hamburg, Germany
Committee V.1.: Applied Design, Proc. International Ship and Offshore Structures Congress, Wuxi, China, September
1991.
SMTC-072-2008 Jensen